<%BANNER%>

Image Contrast and Quality Studies and System Dead-Time Studies for the University of Florida Backscatter X-Ray Imaging ...

Permanent Link: http://ufdc.ufl.edu/UFE0025146/00001

Material Information

Title: Image Contrast and Quality Studies and System Dead-Time Studies for the University of Florida Backscatter X-Ray Imaging System
Physical Description: 1 online resource (260 p.)
Language: english
Creator: Beharry, Kara
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: backscatter, contrast, paralyzable
Nuclear and Radiological Engineering -- Dissertations, Academic -- UF
Genre: Nuclear Engineering Sciences thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The University of Florida Backscatter X-ray (BSX) Imaging system has played an integral role for industries that employ non-destructive testing on equipment and target samples of varying materials. A BSX system provides the ability to image a target from the same side as the x-ray illumination. Due to its unique image acquisition capabilities, various simulations and experiments are continuously being conducted at the University of Florida in order to refine and improve the system and the system?s efficiency in order to fulfill the requirements that industry expects from it. In almost all detector systems there are limitations and the UF BSX imaging system it is no different. Experiments have been performed to analyze the image contrast of detectors in integrating mode, to examine image quality at high count rates, and to estimate the system dead time. Use of the integrating mode has shown that contrast improves for deep void defects in foam (Spray on Foam Insulation) without adversely affecting the contrast of shallow void defects. Analysis of the UF BSX imaging system has demonstrated that it follows a paralyzable model and because of this paralyzable behavior inversion of image contrast occurs at high count rates. Furthermore, based on the paralyzable model estimations of the system dead time (tau) was evaluated. The system dead time (tau) depended on material choice, detector type and position. For the Sodium-Iodide (NaI) detectors, values of tau were calculated that fell within the range of 5.27x10-07 s < tau NaI Series < 5.47x10-07 s. For the YSO detectors, two ranges are given corresponding to the tau values before the reset of the baseline restore constant and after the reset of the baseline restore constant respectively, 3.53x10-07 s < tau YSO Series < 3.79x10-07 s and 1.64x10-07 s < tau YSO Series < 1.95x10-07 s. Throughout the experiments it was observed that the maximum predicted count rate limit of 2.00x10+06 (cps) for the YSO detector was not being achieved. This behavior led to separate testing of the individual components that constitute the BSX system. Currently the reason for the YSO count rate limitation has not been definitively determined, but the problem is believed to be in the pre-amplifier. Additional testing and analysis needs to be performed.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Kara Beharry.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Dugan, Edward T.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0025146:00001

Permanent Link: http://ufdc.ufl.edu/UFE0025146/00001

Material Information

Title: Image Contrast and Quality Studies and System Dead-Time Studies for the University of Florida Backscatter X-Ray Imaging System
Physical Description: 1 online resource (260 p.)
Language: english
Creator: Beharry, Kara
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: backscatter, contrast, paralyzable
Nuclear and Radiological Engineering -- Dissertations, Academic -- UF
Genre: Nuclear Engineering Sciences thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The University of Florida Backscatter X-ray (BSX) Imaging system has played an integral role for industries that employ non-destructive testing on equipment and target samples of varying materials. A BSX system provides the ability to image a target from the same side as the x-ray illumination. Due to its unique image acquisition capabilities, various simulations and experiments are continuously being conducted at the University of Florida in order to refine and improve the system and the system?s efficiency in order to fulfill the requirements that industry expects from it. In almost all detector systems there are limitations and the UF BSX imaging system it is no different. Experiments have been performed to analyze the image contrast of detectors in integrating mode, to examine image quality at high count rates, and to estimate the system dead time. Use of the integrating mode has shown that contrast improves for deep void defects in foam (Spray on Foam Insulation) without adversely affecting the contrast of shallow void defects. Analysis of the UF BSX imaging system has demonstrated that it follows a paralyzable model and because of this paralyzable behavior inversion of image contrast occurs at high count rates. Furthermore, based on the paralyzable model estimations of the system dead time (tau) was evaluated. The system dead time (tau) depended on material choice, detector type and position. For the Sodium-Iodide (NaI) detectors, values of tau were calculated that fell within the range of 5.27x10-07 s < tau NaI Series < 5.47x10-07 s. For the YSO detectors, two ranges are given corresponding to the tau values before the reset of the baseline restore constant and after the reset of the baseline restore constant respectively, 3.53x10-07 s < tau YSO Series < 3.79x10-07 s and 1.64x10-07 s < tau YSO Series < 1.95x10-07 s. Throughout the experiments it was observed that the maximum predicted count rate limit of 2.00x10+06 (cps) for the YSO detector was not being achieved. This behavior led to separate testing of the individual components that constitute the BSX system. Currently the reason for the YSO count rate limitation has not been definitively determined, but the problem is believed to be in the pre-amplifier. Additional testing and analysis needs to be performed.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Kara Beharry.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Dugan, Edward T.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0025146:00001


This item has the following downloads:


Full Text

PAGE 1

1 IMAGE CONTRAST AND QUALITY STUDIES AND SYSTEM DEAD TIME STUDIES FOR THE UNIVERSITY OF FLORIDA BACKSCATTER XRAY IMAGING SYSTEM By KARA NELLA BEHARRY A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Kara Nella Beharry

PAGE 3

3 To my parents

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank God, for all the opportunities he has presented to me. I am truly grateful for all that I have and what He has provided. I would like to thank Dr. Edward Dugan and Dr. James Baciak, for all their invaluable guidance and insight into t his research. I also need to thank Dan Ekdahl for his time and efforts that have contributed to my research project. I would like to give a special thanks to my family who were a great source of support and motivation. I need to especially thank my brother Navin for his help, encouragement and patience. Thanks go out to everyone who has helped contributed to my growth, learning and success, in particular to my Orangeburg, SC friends who have waited patiently for me to succeed in my endeavors Thank s to the University of Florida, Department of Nuclear and Radiological Engineering for the financial funding and my research mate, Olivier Bougeant for aiding me along this academic path.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ...............................................................................................................4 LIST OF TABLES ...........................................................................................................................9 LIST OF FIGURES .......................................................................................................................11 ABSTRACT ...................................................................................................................................29 CHAPTER 1 INTRODUCTION ..................................................................................................................31 Background .............................................................................................................................31 Compton Backscatter Imaging ...............................................................................................31 Radiography by Selective Detection (RSD) ...........................................................................32 Overview of Previous Study ............................................................................................32 RSD Imaging ...................................................................................................................34 2 IMAGING SYSTEM OVERVIEW .......................................................................................37 Physics of Photon Interaction .................................................................................................37 In teraction Mechanisms ...................................................................................................37 Compton Effect ...............................................................................................................37 Photoelectric Effect .........................................................................................................38 Scattering Cross Section ..................................................................................................38 Angular Distribution of Scattered Photons ......................................................................39 UF BSX Imaging System .......................................................................................................39 RSD Scanning System .....................................................................................................39 Detectors ..........................................................................................................................40 3 COMPTON BACKSCATTER IMAGING RELATED STUDIES .......................................41 X R ay Backscatter ..................................................................................................................41 Detector Modes of Operation .................................................................................................42 Puls e Mode ......................................................................................................................43 Current Mode ...................................................................................................................43 4 IMAGE CONTRAST STUDY ...............................................................................................46 Experimental Settings .............................................................................................................46 X R ay Generator .............................................................................................................46 Data Sets ..........................................................................................................................47 Additional Information ....................................................................................................48 Target Material .......................................................................................................................49

PAGE 6

6 Poured Block Panel (SOFI) .............................................................................................49 Factors Affecting Count Rate ..........................................................................................50 Image Contrast ........................................................................................................................51 Contrast ............................................................................................................................51 Methodology ....................................................................................................................52 Results and Trends ..........................................................................................................54 5 UF BSX IMAGING SYSTEM DEAD TIME ........................................................................81 ............................................................................................................81 Introduction .....................................................................................................................81 Materials, Equipment, Settings and Measurements ........................................................82 Experimental set tings ...............................................................................................82 Baseline restoration ..................................................................................................83 Measurements ...........................................................................................................85 Data Analysis ..........................................................................................................................86 NaI Series Foam ...........................................................................................................87 SOFI and Nylon Differences ........................................................................................92 NaI Series Nylon ..........................................................................................................93 YSO Series Foam .........................................................................................................95 YSO Series Nylon ........................................................................................................97 6 DETERMINATION OF UF BSX SYSTEM DEAD TIME .................................................109 Interpretation of Results and Discussion ..............................................................................109 Models for Dead Time Behavior ...................................................................................109 Paralyzable Model .........................................................................................................110 Recorded Count Rate versus True Interaction Rate .............................................................111 Relationship between Observed Rate and True Rate ....................................................111 Plots of Observed Rate (m) versus True Rate (n) ..........................................................112 Calculation .............................................................................116 Using the paralyzable model equation ...................................................................117 ..............................................................................................117 Summary of Calculated UF BSX System Dead Time ..........................................................118 NaI Series ......................................................................................................................118 YSO Series ....................................................................................................................118 ...................................................................................................120 Degradat ion of Uncertainty based on Count Statistics ..................................................121 NaI design count rate limit .....................................................................................122 YSO design count rate limit ...................................................................................122 UF BSX system (detector and associated electronics) ...........................................122 Detectors .................................................................................................................123

PAGE 7

7 7 DEGRADATION OF IMAGE QUALITY ..........................................................................129 Image Quality .......................................................................................................................129 Purpose ..........................................................................................................................129 Setup ..............................................................................................................................129 Channel Contrast Analysis NaI ..................................................................................130 Channel Contrast Analysis YSO ................................................................................135 Unexpected Contrast Transition ...........................................................................................139 Hypothesis .....................................................................................................................139 Comparison ....................................................................................................................140 8 STATISTICAL ANALYSIS OF PILE UP EVENTS ..........................................................145 Count Statistics .....................................................................................................................145 Definition of Terms .......................................................................................................145 Paralyzable System ........................................................................................................145 Pile Up Free ...................................................................................................................146 ith Order Pile Up P(i) ...................................................................................................146 Probability Curves ................................................................................................................148 Plots of the Probabilities ................................................................................................148 NaI Series Probability Plots ........................................................................................148 YSO Series Probability Plots ......................................................................................151 Theoretical Probability Plots .........................................................................................152 Effect of Pile Up on the Fraction of True Events .................................................................154 Fraction of True Events that Escape PileUp ................................................................154 NaI Series ......................................................................................................................155 YSO Series ....................................................................................................................157 Theoretical Plot for the Fraction of True Event Rate ....................................................159 Rate of Observed Counts due to Pile Up ......................................................................159 NaI series ................................................................................................................161 YSO series ..............................................................................................................163 9 YSO DETECTOR COUNT LIMITATIONS .......................................................................166 YSO Detector ........................................................................................................................166 Individual Testing of Components .......................................................................................167 Approach .......................................................................................................................167 Fast Amplifier Component ............................................................................................167 Detector Component ......................................................................................................168 SCA Component ............................................................................................................170 Pre Amplifier Comp onent .............................................................................................173

PAGE 8

8 10 SUMMARY, CONCLUSIONS AND FUTURE WORK ....................................................177 Summary and Conclusions ...................................................................................................177 Contrast Study Pulse Mode versus Current Mode .....................................................177 ......................................................................................................177 YSO Detector Count Rate Limitations ..........................................................................179 Future Work ..........................................................................................................................180 Transition Changes i n Image Quality ............................................................................180 Count Rate Limitation ...................................................................................................180 APPENDIX A PLOTS OF THE OBSERVED TRENDS FOUND IN THE IMAGE CONTRAST STUDY .................................................................................................................................181 B UF BSX S YSTEM D EAD T IME C ALCULATIONS .........................................................237 C DEGRADATION OF IMAGE QUALITY ..........................................................................245 D ELECTRONICS SPECIFICATIONS ..................................................................................246 REFERENCE LIST .....................................................................................................................258 BIOGRAPHICAL SKE TCH .......................................................................................................260

PAGE 9

9 LIST OF TABLES Table page 41 Series used in the i mage c ontrast s tudy .............................................................................47 42 Series 1 poured block 2 inches f oam overlay ...................................................................48 43 Series 2 poured block 4 inches f oam overlay ...................................................................48 44 Series 3 poured block 6 inches f oam overlay ...................................................................48 45 Settings used for SOFI block scans ...................................................................................49 51 NaI detectors f oam & n ylon t argets ................................................................................83 52 YSO detectors f oam & n ylon t argets ..............................................................................83 53 YSO detectors c ontinuation of YSO s eries but with alterations .....................................83 54 Fast a mplifier baseline restore time constant ....................................................................85 61 Sample calculation using Series 2 NaI f oam to obtain true rate (n) .............................113 62 Calculation of system dead time ......................................................................................118 63 NaI s eries ........................................118 64 Summary of the UF BSX s ystem dead t YSO s eries ......................................119 81 Description of probability that a recorded count is caused by pile up. ...........................148 91 Decay time constants and detection surface area .............................................................166 92 Design count rate limit (cps) ............................................................................................166 93 Settings on pulse generator ..............................................................................................174 B 1 : Series 2 NaI f oam (2.0 mm beam aperture size) ....................244 B 2 : Series 1 NaI n ylon (1.5 mm beam aperture size) ...................244 B 3 Series 2 YSO n ylon (2.0 mm beam aperture size) .................244 B 4 : Series 3 YSO n ylon (2.5 mm beam aperture size, fast filter amplifier peak output voltage set at 7 V and the baseline restore constant reset to its fastest value of 110 ns) ....................................................................................................244

PAGE 10

10 B 5 System dead t : Series 4 YSO n ylon (4.0 mm beam aperture size, fast filter amplifier peak output voltage set at 7 V and the baseline restore constant reset to its fastest value of 110 ns) ....................................................................................................244 C 1 UF BSX s ystem d ead t ime for NaI detectors c hannel in n ylon ....................................245 C 2 UF BSX s ystem d ead t ime for YSO detectors c hannel in n ylon ..................................245

PAGE 11

11 LIST OF FIGURES Figure page 11 Lateral Migration Radiography (LMR). ............................................................................33 12 Compton Backscattering Imaging (CBI). ..........................................................................34 13 Compton backscatter i maging t echniques. ........................................................................34 14 RSD collimated and uncollimated detector with detec tion collimation plane. ................35 15 Rastering technique for RSD. ............................................................................................36 21 The Compton Effect ...........................................................................................................37 22 Photograph of the UF BSX imaging system BSX system mounted on the frame ............40 31 Energy weighted current mode contrast for a single pixel for varying spectral curves .....45 41 Photon scattering to absorption ratio in NaI and YSO as a function of energy. ...............46 42 Photograph of poured block foam panel ............................................................................49 43 Background regions surrounding a hole or defect. ............................................................52 44 Holes with constant diameter (D) lie in the x direction and holes with constant depth lie in the y direction. ..........................................................................................................53 45 Line along the diagonal of the block indicates the holes with increasing depth and increasing diameter. ...........................................................................................................54 46 Raw images from NaI and YSO in count mode (Series 12 inch thick foam overlay) .....55 47 Raw images from NaI and YSO in integrating mode (Series 1 2 inch thick foam overlay) ..............................................................................................................................55 48 Raw images from NaI and YSO in count mode (Series 2 4 inch thick foam overlay) ...56 49 Raw images from NaI and YSO in integrating mode (Series 2 4 inch thick foam overlay) ..............................................................................................................................56 410 Raw images from NaI and YSO in count mode (Series 3 6 inch thick foam overlay) ...57 411 Raw images from NaI and YSO in integrating mode (Series 3 6 inch thick foam overlay) ..............................................................................................................................57 412 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int for unprocessed images at constant diameter (D) of 0.75 inches ............................................59

PAGE 12

12 413 Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.75 inches ......................................60 414 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.75 inches ......................................60 415 Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.75 inches ......................................61 416 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant depth (d) of 0.5 inches. ...................................................61 417 Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches. .............................................62 418 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant depth (d) of 0.5 inches. .............................................62 419 Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches. .............................................63 420 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at increasing depth (d) and diameter (D). .........................................64 421 Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at increasing depth (d) and diameter (D). ....................................64 422 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at increasing depth (d) and diameter (D). ....................................65 423 Series 1: Comparing r elative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at increasing depth (d) and diameter (D). ....................................65 424 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.75 inches. ...........................................67 425 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.75 inches. .....................................67 426 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.75 inches. .....................................68 427 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.75 inches. .....................................68 428 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant depth (d) of 0.5 i nches. ...................................................69

PAGE 13

13 429 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches. .............................................69 430 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed image s at constant depth (d) of 0.5 inches. .............................................70 431 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches. .............................................70 432 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at increasing depth (d) and diameter (D). .........................................71 433 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at increasing depth (d) and diameter (D). ....................................71 434 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at increasing depth (d) and diameter (D). ....................................72 435 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at increasing depth (d) and diameter (D). ....................................72 436 Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.75 inches. ...........................................73 437 Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.75 inches. .....................................74 438 Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.75 i nches. .....................................74 439 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.75 inches. .....................................75 440 Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant depth (d) of 0.5 inches. ...................................................76 441 Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches. .............................................77 442 Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant depth (d) of 0.5 inches. .............................................77 443 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches. .............................................78 444 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at increasing depth (d) and diameter (D). ....................................79

PAGE 14

14 445 Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at increasing depth (d) and diameter (D). ....................................79 446 Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at increasing depth (d) and diameter (D). ....................................80 447 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at increasing depth (d) and diameter (D). ....................................80 51 UF BSX i maging e quipment s etup t he component specifications are all identified in Appendix B. .......................................................................................................................82 52 Equivalent circuit of a baseline restorer .............................................................................84 53 Image acquired for a uniform nylon block ........................................................................85 54 Series 1 NaI detector 1 foam (1.5 mm beam aperture size). ...........................................87 55 Series 1 NaI detector 2 foam (1.5 mm beam aperture). ..................................................88 56 Series 1 NaI detector 3 foam (1.5 mm beam aperture). ..................................................88 57 Series 1 NaI detector 4 foam (1.5 mm beam aperture). ..................................................89 58 Series 2 NaI detector 1 foam (2.0 mm beam aperture). ..................................................90 59 Series 2 NaI detector 2 foam (2.0 mm beam aperture). ..................................................90 510 Series 2 NaI detector 3 foam (2.0 mm beam aperture). ..................................................91 511 Series 2 NaI detector 4 foam (2.0 mm beam aperture). ..................................................91 512 Series 1 NaI n ylon (1.5 mm beam aperture). ..................................................................93 513 Series 1 NaI n ylon (1.5 mm beam aperture). ..................................................................93 514 Series 1 NaI n ylon (1.5 mm beam aperture). ..................................................................94 515 Series 1 NaI n ylon (1.5 mm beam aperture). ..................................................................94 516 Series 1 YSO f oam (2.5 mm beam aperture). .................................................................95 517 Series 1 YSO f oam (2.5 mm beam aperture). .................................................................96 518 Series 1 YSO f oam (2.5 mm beam aperture). .................................................................96 519 Ser ies 1 YSO f oam (2.5 mm beam aperture). .................................................................97

PAGE 15

15 520 Series 1 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 4 volts. ............................................................98 521 Series 1 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 4 volts. ............................................................98 522 Series 1 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 4 volts. ............................................................99 523 Series 1 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 4 volts. ............................................................99 524 Series 2 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts. ..........................................................100 525 Series 2 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts. ..........................................................100 526 Series 2 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts. ..........................................................101 527 Series 2 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts. ..........................................................101 528 Series 3 YSO n ylon (2.5 mm beam aperture) with the YSO fast filter amplifier peak o utput voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ..............................................................102 529 Series 3 YSO n ylon (2.5 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ..............................................................103 530 Series 3 YSO n ylon (2.5 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ..............................................................103 531 Series 3 YSO n ylon (2.5 mm beam ape rture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ..............................................................104 532 Series 4 YSO n ylon (4.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ..............................................................104 533 Series 4 YSO n ylon (4.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ..............................................................105

PAGE 16

16 534 Series 4 YSO n ylon (4.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ..............................................................105 535 Series 4 YSO n ylon (4.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ..............................................................106 536 Series 5 YSO n ylon (4.0 mm be am aperture and x ray voltage at 60 kVp) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ......................107 537 Series 5 YSO n ylon (4.0 mm beam aperture and x ray voltage at 60 kVp) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ......................107 538 Series 5 YSO n ylon (4.0 mm beam aperture and x ray voltage at 60 kVp) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). ......................108 61 Basic assumptions of the two models ..............................................................................109 62 Variation of the observed rate m as a function of the true rate n for two models of dead time losses ...............................................................................................................112 63 Measured count rate vs. True count r ate: Series 2 NaI 2 f oam (2.0 mm beam aperture). ..........................................................................................................................114 64 Measured c ount r ate vs. True c ount r ate: Series 1 NaI 2 n ylon (1.5 mm beam aperture). ..........................................................................................................................114 65 Measured c ount r ate vs. True c ount r ate: Series 2 YSO 5 n ylon (2.0 mm beam aperture). ..........................................................................................................................115 66 Measured c ount r ate vs. True c ount r ate: Series 4 YSO 8 n ylon (4.0 mm beam aperture and 55 kVp). ......................................................................................................115 67 Measured c ount r ate vs. True c ount r ate: Series 5 YSO 8 n ylon (4.0 mm beam aperture and 60 kVp). ......................................................................................................116 68 Comparing loss of uncertainty obtained by the detector to the combination of detector and associated electronics : Series 2 NaI f oam. ...............................................124 69 Comparing loss of counts obtained by the detector to the combination of detector and associated electronics : Series 2 NaI f oam. ....................................................................124 610 Comparing loss of uncertainty obtained by the detector to the combination of detector and associated electronics : Series 1 NaI n ylon. ..............................................125

PAGE 17

17 611 Comparing loss of counts obtained by the detector to the combination of detector and associated electronics : Series 1 NaI n ylon. ...................................................................125 612 Comparing loss of uncertainty obtained by the detector to the combination of detector and associated electronics : S eries 2 YSO n ylon. ............................................126 613 Comparing loss of counts obtained by the detector to the combination of detector and associated electronics : Series 2 YSO n ylon. .................................................................126 614 Comparing loss of uncertainty obtained by the detector to the combination of detector and associated electronics : Series 4 YSO n ylon. ............................................127 615 Comparing loss of counts obtained by the detector to the combination of detector and associated electronics : Series 4 YSO n ylon. .................................................................127 71 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 1 (without overlay). .................................................130 72 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed ima ges NaI detector 2 (without overlay). .................................................131 73 Trend observed for relative contrast as current (mA) increases for both processed an d unprocessed images NaI detector 3 (without overlay). .................................................131 74 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 4 (without overlay). .................................................132 75 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 1 (with overlay). ......................................................132 76 Trend observed for relative contra st as current (mA) increases for both processed and unprocessed images NaI detector 2 (with overlay). ......................................................133 77 Trend observed for r elative contrast as current (mA) increases for both processed and unprocessed images NaI detector 3 (with overlay). ......................................................133 78 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 4 (with overlay). ......................................................134 79 Image quality transition for NaI detector 3 (no overlay). ................................................135 710 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 5 (without overlay). ...............................................135 711 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 6 (without overlay). ...............................................136 712 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 8 (without overlay). ...............................................136

PAGE 18

18 713 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 5 (with overlay). ....................................................137 714 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 6 (with overlay). ....................................................137 715 Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 8 (with overlay). ....................................................138 716 Image quality transition for the YSO detector 8 (without overlay). ................................138 717 Representation of a paralyzable model. ...........................................................................139 718 Trend observed in NaI detector 1 that is attributed to the contrast changes at high count rates. .......................................................................................................................141 719 Trend observed in NaI detector 2 that is attributed to the cont rast changes at high count rates. .......................................................................................................................141 720 Trend observed in NaI detector 3 that is attributed to the contrast changes at high count rates. .......................................................................................................................142 721 Trend observed in NaI detector 4 that is attributed to the contrast changes at high count rates. .......................................................................................................................142 722 Trend observed in YSO detector 5 that is attributed to the contrast changes at high count rates. .......................................................................................................................143 723 Trend observed in YSO detector 6 that is attributed to the contrast changes at high count rates. .......................................................................................................................143 724 Trend observed in YSO detector 6 that is attributed to the contrast changes at high count rates. .......................................................................................................................144 81 Probability plot for Series 1 NaI 1 n ylon (1.5 mm beam a perture s ize) .......................149 82 Probability plot for Series 1 NaI 2 n ylon (1.5 mm beam a perture s ize) .......................149 83 Probability plot for Series 1 NaI 3 n ylon (1.5 mm beam aperture size) .......................150 84 Probability plot for Series 1 NaI 4 n ylon (1.5 mm beam aperture size). ......................150 85 Probability plot for Series 5 YSO 5 n ylon (4.0 m m beam aperture size). ....................151 86 Probability plot for Series 5 YSO 6 n ylon (4.0 mm beam a perture s ize). ....................152 87 Probability plot for Series 5 YSO 8 n ylon (4.0 mm beam a perture s ize). ....................152 88 Paralyzable system theoretical probability plot ............................................................153

PAGE 19

19 89 The fraction of re corded counts (upper curve) and true events (lower curve) that escape pile : Series 1 NaI 1 n ylon (1.5 mm beam aperture size). ...................................................................................................................155 810 The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile : Series 1 NaI 2 n ylon (1.5 mm beam aperture size). ...................................................................................................................156 811 The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile up a : Series 1 NaI 3 n ylon (1.5 mm beam aperture size). ...................................................................................................................156 812 The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile : Series 1 NaI 4 n ylon (1.5 mm beam aperture size). ...................................................................................................................157 813 The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile : Series 5 YSO 5 n ylon (4.0 mm beam aperture size). ...................................................................................................................157 814 The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile up : Series 5 YSO 6 n ylon (4.0 mm beam aperture size). ...................................................................................................................158 815 The fraction of recorded counts (uppe r curve) and true events (lower curve) that escape pile : Series 5 YSO 8 n ylon (4.0 mm beam aperture size). ...................................................................................................................158 816 Fraction of recorded counts and true events that escape pile up. ....................................159 817 Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu NaI 1 n ylon ( 1.5 mm beam aperture size ) ....................................................................161 818 Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu NaI 2 n ylon ( 1.5 mm beam aperture size ) ....................................................................162 819 Comparisons of the recorded true events free of pileup (rpf due to pile up (rpu NaI 3 n ylon ( 1.5 mm beam aperture size ) ....................................................................162 820 Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu NaI 4 n ylon ( 1.5 mm beam aperture size ) ....................................................................163

PAGE 20

20 821 Comparisons of the recorded true events free of pileup (rpf due to pile up (rpu YSO 5 n ylon ( 4.0 mm beam aperture size ) ..................................................................163 822 Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu YSO 6 n ylon ( 4.0 mm beam aperture size ) ..................................................................164 823 Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu YSO 8 n ylon ( 4.0 mm beam aperture size ) ..................................................................164 91 Recorded c ount r ate versus True i nteraction r ate: Series 1 Plast ic n ylon (4.0 mm beam aperture size). .........................................................................................................168 92 Recorded c ount r ate versus True i nteraction r ate: Series 1 Plastic n ylon (no aperture). ..........................................................................................................................169 93 Electronic components before the SCA and their corresponding pulse widths ...............171 94 Pictures A and B show the new SCA prototype while pictures C and D show the prototype mounted onto the motherboard ........................................................................172 95 Circuit of the SCA prototype provided by D. Ekdahl. .....................................................172 96 Equipment setup for testing the preamplifier. ..................................................................173 97 Frequency (MHz) versus Net c ount r ate (cps) ................................................................174 98 Simplified circuit of a charge amplifier configuration ....................................................175 A 1 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.5 inches. ...........................................181 A 2 Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.5 i nches. .....................................181 A 3 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant dia meter (D) of 0.5 inches. .....................................182 A 4 Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.5 inches. .....................................182 A 5 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI I nt. for unprocessed images at constant diameter (D) of 0.25 inches. .........................................183 A 6 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.25 inches. .........................................183

PAGE 21

21 A 7 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.25 inches. ...................................184 A 8 Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.25 inches. ...................................184 A 9 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant depth (d) of 0.25 inches. ...............................................185 A 10 Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth (d) of 0.25 inches. .........................................185 A 11 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant depth (d) of 0.25 inches. .........................................186 A 12 Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant depth (d) of 0.25 inches. .........................................186 A 13 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter ( D) of 0.75 inches. .............................................187 A 14 Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.75 inches. .......................................187 A 15 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for p rocessed images at constant diameter (D) of 0.75 inches. .......................................188 A 16 Series 1: Comparing relative contrast of images obtained by NaI I nt. and YSO Int. for processed images at constant diameter (D) of 0.75 inches. .......................................188 A 17 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.5 inches. ...............................................189 A 18 Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.5 inches. .........................................189 A 19 S eries 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.5 inches. .........................................190 A 20 Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.5 inches. .........................................190 A 21 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.25 inches. .............................................191 A 22 Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.25 inches. .......................................191

PAGE 22

22 A 23 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.25 inc hes. .......................................192 A 24 Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant di ameter (D) of 0.25 inches. .......................................192 A 25 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant depth (d) of 0.5 inches. .....................................................193 A 26 Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant depth (d) of 0.5 inches. ...............................................193 A 27 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.5 inches. ...............................................194 A 28 Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.5 inches. ...............................................194 A 29 Series 1: Comparing relative co ntrast of images obtained by NaI CR and NaI Int. for processed images at constant depth (d) of 0.25 inches. ...................................................195 A 30 Series 1: Compa ring relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant depth (d) of 0.25 inches. .............................................195 A 31 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.25 inches. .............................................196 A 32 Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.25 inches. .............................................196 A 33 Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at increasing depth (d) and diameter (D). ...........................................197 A 34 Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at increasing depth (d) and diameter (D). ......................................197 A 35 Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at increasing depth (d) and diameter (D). ......................................198 A 36 Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at increasing depth (d) and diameter (D). ......................................198 A 37 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed image s at constant diameter (D) of 0.5 inches. ...........................................199 A 38 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. f or unprocessed images at constant diameter (D) of 0.5 inches. .....................................199

PAGE 23

23 A 39 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.5 inches. .....................................200 A 40 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.5 inches. .....................................200 A 41 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.25 inches. .........................................201 A 42 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.25 inches. ...................................201 A 43 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.25 inches. ...................................202 A 44 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.25 inches. ...................................202 A 45 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at const ant depth (d) of 0.25 inches. ...............................................203 A 46 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth (d) of 0.25 inches. .........................................203 A 47 Series 2: Comparing relative contrast of images obtained by NaI C R and YSO CR for unprocessed images at constant depth (d) of 0.25 inches. .........................................204 A 48 Series 2: Comparing relative contrast of images o btained by NaI Int. and YSO Int. for unprocessed images at constant depth (d) of 0.25 inches. .........................................204 A 49 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at increasing depth (d) and diameter (D). .......................................205 A 50 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at increasing depth (d) and diameter (D). ..................................205 A 51 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at increasing depth (d) and diameter (D). ..................................206 A 52 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at increasing depth (d) and diameter (D). ..................................206 A 53 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.75 inches. .............................................207 A 54 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.75 inches. .......................................207

PAGE 24

24 A 55 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.75 inc hes. .......................................208 A 56 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.75 inches. .......................................208 A 57 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.5 inches. ...............................................209 A 58 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.5 inches. .........................................209 A 59 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.5 inches. .........................................210 A 60 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.5 inches. .........................................210 A 61 Series 2: Comparing r elative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.25 inches. .............................................211 A 62 S eries 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.25 inches. .......................................211 A 63 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.25 inches. .......................................212 A 64 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.25 inches. .......................................212 A 65 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant depth (d) of 0.5 inc hes. .....................................................213 A 66 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant dept h (d) of 0.5 inches. ...............................................213 A 67 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.5 inches. ...............................................214 A 68 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for p rocessed images at constant depth (d) of 0.5 inches. ...............................................214 A 69 Series 2: Comparing relative contrast of images obtained by NaI CR an d NaI Int. for processed images at constant depth (d) of 0.25 inches. ...................................................215 A 70 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant depth (d) of 0.25 inches. .............................................215

PAGE 25

25 A 71 Series 2: Comparing relative co ntrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.25 inches. .............................................216 A 72 Series 2: Compari ng relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.25 inches. .............................................216 A 73 Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at increasing depth (d) and increasing diameter (D). ..........................217 A 74 Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at increasing depth (d) and increasing diameter (D). ....................217 A 75 Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at increasing depth (d) and increasing diameter (D). ....................218 A 76 Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at increasing depth (d) and increasing diameter (D). ....................218 A 77 Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.5 inches. ...........................................219 A 78 Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at c onstant diameter (D) of 0.5 inches. .....................................219 A 79 Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.5 inches. .....................................220 A 80 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.5 inches. .....................................220 A 81 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.25 inches. ...................................221 A 82 Series 3: Comparing rel ative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.25 inches. ...................................221 A 83 S eries 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.25 inches. ...................................222 A 84 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.25 inches. ...................................222 A 85 Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant depth (d) of 0.25 inches. ...............................................223 A 86 Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth (d) of 0.25 inches. .........................................223

PAGE 26

26 A 87 Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant depth (d) of 0.25 inches. .........................................224 A 88 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant depth (d) of 0.25 inches. .........................................224 A 89 Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.75 inches. .............................................225 A 90 Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.75 inches. .......................................225 A 91 Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.75 inches. .......................................226 A 92 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.75 inches. .......................................226 A 93 Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.5 inches. ...............................................227 A 94 Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.5 inches. .........................................227 A 95 Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.5 inches. .........................................228 A 96 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.5 inches. .........................................228 A 97 Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.25 i nches. .............................................229 A 98 Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.25 inches. .......................................229 A 99 Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.25 inches. .......................................230 A 100 Series 3: Comparing relative contrast of images obtained by NaI Int. and Y SO Int. for processed images at constant diameter (D) of 0.25 inches. .......................................230 A 101 Series 3: Comparing relative contrast of images obta ined by NaI CR and NaI Int. for processed images at constant depth (d) of 0.5 inches. .....................................................231 A 102 Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant depth (d) of 0.5 inches. ...............................................231

PAGE 27

27 A 103 Series 3: Co mparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.5 inches. ...............................................232 A 104 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.5 inches. ...............................................232 A 105 Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant depth (d) of 0.25 inches. ...................................................233 A 106 Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant depth (d) of 0.25 inches. .............................................233 A 107 Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.25 inches. .............................................234 A 108 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.25 inches. .............................................234 A 109 Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at increasing depth (d) and increasing diameter (D). ..........................235 A 110 Series 3: Comparing relative contrast of images obtained by YSO CR a nd YSO Int. for processed images at increasing depth (d) and increasing diameter (D). ....................235 A 111 Series 3: Comparing relative contras t of images obtained by NaI CR and YSO CR for processed images at increasing depth (d) and increasing diameter (D). ....................236 A 112 Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at increasing depth (d) and increasing diameter (D). ....................236 B 1 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: NaI 1 f oam (2.0 mm beam aperture size). ..................................................................................................237 B 2 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: NaI 3 f oam (2.0 mm beam aperture size). ..................................................................................................237 B 3 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: NaI 4 f oam (2.0 mm beam aperture size). ..................................................................................................238 B 4 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 1: NaI 1 n ylon (1.5 mm beam aperture size). ..................................................................................................238 B 5 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 1: NaI 3 n ylon (1.5 mm beam aperture size). ..................................................................................................239 B 6 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 1: NaI 4 n ylon (1.5 mm beam aperture size). ..................................................................................................239

PAGE 28

28 B 7 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: YSO 6 n ylon (2.0 mm beam aperture size). ..................................................................................................240 B 8 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: YSO 7 n ylon (2.0 mm beam aperture size). ..................................................................................................240 B 9 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: YSO 8 n ylon (2.0 mm beam aperture size). ..................................................................................................241 B 10 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 4: YSO 5 n ylon (4.0 mm beam aperture size at 55 kVp). .................................................................................241 B 11 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 4: YSO 6 n ylon (4.0 mm beam aperture size at 55 kVp). .................................................................................242 B 12 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 4: YSO 7 n ylon (4.0 mm beam aperture size at 55 kVp). .................................................................................242 B 13 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 5: YSO 5 n ylon (4.0 mm beam aperture size at 60 kVp). .................................................................................243 B 14 Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 5: YSO 6 n ylon (4.0 mm beam aperture size at 60 kVp). .................................................................................243

PAGE 29

29 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science IMAGE CONTRAST AND QUALITY STUDIES AND SYSTEM DEAD TIME STUDIES FOR THE UNIVERSITY OF FLORIDA BACKSCATTER XRAY IMAGING SYSTEM By Kara Nella Beharry December 2009 Chair: Edward T. Dugan Major: Nuclear Engineering Sciences The University of Florida Backscatter X r ay (BSX) Imaging system has played an integral role for industries that employ nondestructive testing on equipment and target samples of varying materials. A BSX system provides the ability to image a target from the same side as the x ray illumination. Due to its unique image acquisition capabilities various simulations and experiments are continuously being conducted at the University of Florida in order to refine and improve the system and the systems efficiency in order to fulfill the requirements that industry expects from it. In almost all detector systems there are limitations and the UF BSX imaging system it is no different. Experiments have been performed to analyze the image contrast of detectors in integrating mode, to examine image quality at high count rates, and to estimate the system dead time. Use of the integrating mode has shown that contrast improves for deep void defects in foam (S pray on F oam I nsulation) without adversely affecting the contrast of shallow void defects. A nalysis of the UF BSX imaging system has demonstrated that it follow s a paralyzable model and because of this paralyzable behavior inversion of image contrast occurs at high count rates. Furthermore, based on the paralyzable model estimations of the system dead evaluated. The s ystem dead time ( ) depended on material choice, detector type and position. For the Sodium Iodide (NaI) detectors, were calculated that fell within the

PAGE 30

30 range of 5.27x1007 NaI Series<5. 47x1007 s For the YSO detectors, two ranges are given of the baseline restore constant respectively, 3.53x1007 YSO Series<3.79x1007 s and 1.64x1007 YSO Series< 1.95x1007 s Throughout the experiments it was observed that the maximum predicted count rate limit of 2.00x1 0+0 6 (cps) for the YSO detector was not being achieved This behavior led to separate testing of the individual components that constitute the BSX system. Currently the reason for the YSO count rate limitation has not been definitively determined, but the problem is believed to be in the pre amplifier. Additional testing and analysis needs to be performed.

PAGE 31

31 CHAPTER 1 INTRODUCTION Background The purpose of this research project is to present and explain current findings of the University of Florida (UF) Backscatter X r ay Imaging System (BSX ). The UF BSX system was created and is now currently being used by the University of Florida (UF) Nuclear and Radiological Department (NRE). Research at UF explores mainly backscatter Radiography by Selective Detection (RSD) and this technique is employed by industry today. The first study in this research project examined experimental observations from the BSX imaging system for the purpose of validating the use of the current (or integral) mode in detectors. This objective used a spray on foam insulation (SOFI) poured block as the target sample and the image contrast obtained from the detectors in count mode was compared to the image contrast obtained from a detector in current mode. The second study became more expansive than the first since initially the purpose entailed the determination of the system dead time for the BSX system From there this second study has evolved to encompass the physics and statistical analysis of the effects of pile up and the testing of individual components that comprise the BSX system to deduce the limiting factor that prevents the YSO detector from fully achieving its designed maximum count rate limit of 2 million per second Compton Backscatter Imaging Compton backscatter imaging (CBI) is a single sided imaging technique that has a variety of applications, including nondestructive testing and medical imaging. The CBI technique is unique; the source and the detectors are on the same side of the target. Such an arrangement facilitates a non intrusive examination from a single side thus allowing objects to be scanned where it is impossible to have a film or detector placed behind the target.

PAGE 32

32 Typical X ray scanning systems such as transmission radiography use forward scattered photons to obtain information about a sample. CBI instead uses information given by photons which are scattered back on the same side as the source. The energies and angles of photons that are backscattered depend on incident photon energy as well as the properties of the interacting medium. As a result, the difference in absorption and scattering cross section of the target materials creates differences in t he backscatter photon field intensity which allows for image reconstruction of the target and consequently provides insight and information about the target and its interior.1 Since the inception of CBI, a diverse set of imaging techniques have evolved using both collimated and uncollimated detectors, coded apertures, and hard x ray optics. 2Radiography by Selective Detection (RSD) Research at the University of Florida explores backscatter Radiography by Selective Detection (RSD), Lateral Migration Radiography (LMR), Shadow Aperture Backscatter Radiography (SABR) and Computed Image Backscatter Radiography (CIBR). In this research experimental and statistical analysis was based on backscatter RSD Overview of Previous Study Lateral Migration Radiography3 12 (Figure 1 1) was one of the first techniques developed at the University of Florida Nuclear Engineering Department. LMR shares some similarity to the RSD technique3 (Figure 1 2), but instead of counting primarily single collision backscattered photons the LMR technique primarily counts multiple collision backscattered photons that have laterally spread out from the illumination beam entry point.3 After some improvements and modifications to the LMR technique the Backscatter Radiography by Selective Detection (RSD) was developed. Adding adjustable collimators to the detector allows for the possibility of choosing backscattered photons to be counted. By preferentially selecting specific components of

PAGE 33

33 a scattered photon field, information relating to specific locations and properties of an imaged sample can be extracted.3 The utilization of collimators by RSD limits the scatter acquisition to a depth at or below the collimation plane. It does this by blocking the near scatter from the detectors, only allowing those photons from the collimation plane (CP) or below to enter. However, RSD differs from highly collimated techniques because the collimators are larger and dont limit the detector window to only a small voxel of interest.1 Hence, RSD lies between extreme collimated and uncollimated techniques (Figure 1 3) 1 2 Figure 11. Lateral Migration Radiography (LMR) 3

PAGE 34

34 Figure 1 2. Compton Backscattering Imaging (CBI) 3 Figure 1 3. Compton backscatter i maging t echniques: A) highly c ollimated B) not c ollimated RSD I maging 1 2 RSD uses a pencil beam source and it obtains data via a rastering technique .1 It is considered unique because it collimates to a plane instead of a particular point. Collimation to a plane allows it to use single and multiple scatter events, providing much better subsurface resolution than uncollimated techniques and being orders of magnitude faster than highly collimated techniques1 (Figure 1 4). A combination of first and multiple scatter events from

PAGE 35

35 various scan depths can be used to image a variety of defects, and scan for objects at different depths For the rastering techni que (Figure 15), the pencil beam passes over a voxel of interest and the detectors obtain data from that voxel. The beam moves in a continuous motion, rastering back and forth until the entire target area has been covered13 1 Since targets of interest will have different atomic and material properties, difference in absorption and scattering cross sections of the target are found and consequently this determines the backscatter radiation field intensity These differences are further highlighted in the data collected by the detectors allowing for changes and enhancement in contrast and thus the visibility of target features in an image. Figure 1 4. RSD collimated and uncollimated detector with detection collimation plane1 2

PAGE 36

36 Figure 1 5. Rastering technique for RSD. 1

PAGE 37

37 CHAPTER 2 IMAGING SYSTEM OVERV IEW Physics of Photon Interaction Interaction Mechanisms For an x ray based scanning system it is necessary to understand the mechanism behind photon interaction with matter. Generally there ar e five types of interactions with matter by x ray photons, but only three major types play an important role with regards to radiation measurements and the amount of energy released in a medium: photoelectric absorption, Compton scattering, and pair produc tion. In RSD the photon interactions are characterized by Compton scattering and photoelectric absorption, thus a brief description of these interaction mechanisms is presented in order to understand how an image is formed using the RSD scanning system. C ompton Effect The Compton effect, is simply the elastic scattering of a photon by an electron, in which both energy and momentum are conserved.16 The incident photon with energy E and wavelength 0 ), with 1 ). Figure 2 1. The Compton Effect

PAGE 38

38 The Compton scattering equation, which indicates the energy of the scattered photon, is given by 21()(1cos)oh hv h mc (2 1) w here moc2Photoelectric Effect the rest energy if the electron, is 0.511 MeV and all energies are measured in MeV. Equation 21 indicates that there is a one to The photoelectric effect is also an important interaction mechanism in the characterization of photon interactions in RSD imaging. At the lower photon energies the photoelectric effect is dominant. In the energy region in which backscatter is performed, t he Comptoneffect usually takes over at medium energies3Scattering Cross Section since the Comptoneffect dominance is very broad, extending from approximately 20 keV to 20 MeV for low Z media (e.g. carbon, air, aluminum, Spray on Foam Insulation), it is the competition between the photoelectric effect and Compton scatter and the control of one over the other in different materials at different energies that allows us to distinguish the different materials and features in the images. Cross e defined as the effective area for a collision. The total cross section measures the probability that an interaction of any type will occur when photons strike a target. Similarly, the total scattering cross section is the sum of the elastic and inelastic scattering cross section, which represents the probability of a scattering event. The probability of scattering is defined as the ratio of the blocked area in a beam to the beam area, see Equation 2 2 () nAdx ndx A dN ndx N (2 2)

PAGE 39

39 w here dx is the thickness of the target, n is the number of nuclei per unit volume, dN is the number of scattered particles, and N is the number of incoming particles. If the target is thicker and with boundary conditions x = 0 and N = No (1)ndx scatteredo oNNNNe then Equation 22 can be further developed into Equation 23 (2 3) Angular Distribution of Scattered Photons In the previous sections, the kinematics of the Compton scattering were explained. The angular distribution of scattered photons based o n the quantum mechanical theory of Compton scattering is given by the Klein Nishina16 2 24 ''2 2 24 2 9 2sin 2 1 8.9975510 4eo o odke vm dmcvvsr Nm k c formula, Equation 24 (2 4) The Klein Nishina relationship (de is the probabi lity of a photon passing normally through a layer of matter containing one electron per m2UF BSX Imaging System RSD Scanning System There are four major parts that comprise the RSD scanning system: x ray generator, detectors, the electronics and the image acquisition and processing. The system consists of a Yxlon MCG41 x ray generator mounted onto a scanning table with X Y scan motio n capabilities. There are four sodium iodide [NaI(Tl)] scintillation detectors and four Yttrium Orthosilicate (YSO Y2SiO5 ) (YSO) detectors. The sodium iodide detectors are positioned at the corners of an 18X18 centimeter square, centered on the x ray bea m. The YSO detectors orbit

PAGE 40

40 on an aluminum ring around each of the NaI detectors. The signal s recorded from the RSD scanning system are processed and displayed through a LABVIEW code.3 A schematic of the RSD system is shown in Figure 22. Figure 2 2. Phot ograp h of the UF BSX i maging s ystem BSX system mounted on the frame Detectors 3 For the current RSD system 8 detectors were used; 4 NaI and 4 YSO detectors. Each of the detectors is capable of generating its own image of the scanned object, but a cross correlated image can also be generated from any combination of detector images.2 Initially the BSX system only employed NaI detectors because of their ready availability, and later at the suggestion of Gintek, Inc., YSO detectors were examined and s ubsequently integrated into the RSD system. Each detector is capable of being collimated at different lengths or if suited no collimation can also be achieved. The use of an array of detectors with varying collimation produces images each with a unique vi ew that can be correlated to produce images with easily recognized object internal structure and /or defect details. 13

PAGE 41

41 CHAPTER 3 COMPTON BACKSCATTER IMAGING RELATED STUD IES The BSX imaging concept is extremely useful and there are many practical uses and reasons why such a system should continuously improve. Previous members of the Scatter X ray Imaging (SXI) research group have contributed to the present day RSD scanning system and it is because of this evolution of the scanning system that a series of study such as the evaluation of contrast obtained by detectors in integrating mode needs to be re explored. Many discussions both theoretical and experimental explored the use of detectors in count (or pulse) and current (or integrating) mode. Consequent ly it is important to compare the quality of images that are obtained by the count rate and current rate mode s This section will look at previous and related studies with the fo cus of findings that involve the use of a detector in either count or current mode. X R ay Backscatter Stephanie Brygoo M.S. thesis 200217 highlights the importance of nondestructive techniques. Brygoo 2002 reports that the detection of flaws in concrete foundations of a building or unveiling cracks in the wing of a plane without de stroying the target has always been a challenge for engineers and industry. In order to avoid damaging the samples being examined, scientists have developed various nondestructive techniques. Ultrasound techniques are the most famous, but there are other techniques like magnetic techniques or penetrating photon techniques.17 Initially it was the LMR technique that was first developed and was first applied at the University of Florida for the detection of buried land mines. Over some time, after many measur ements, research, and simulations it was reported in the publication Detection of Flaws Using Lateral Migration X ray Radiography by Dugan et al. 200318 that LMR is capable of

PAGE 42

42 detecting sub surface flaws and defects, relatively small composition changes and corrosion in materials and structures of industrial importance. The present day RSD x ray backscatter technique has been fine tuned and evolved into an efficient scanning machine that can examine a myriad of materials such as aluminum, plastics, honeycomb laminates, reinforced carbon composites, steel, and titanium alloys Experimental results obtained by Shedlock et al. 200619 have shown that this Non Destructive Examination (NDE) technique can even be used to detect boric acid deposition on a metal lic plate through steel foil reflective insulation commonly covering reactor pressure vessels. The scanning system is fully capable of detecting boric acid deposits with submillimeter resolution through such insulating materials. Because of the security related events and the increased intensity to combat terror and smuggling, there is an increasing need for airport security and border control. X ray imaging techniques based on CBI can serve as a tool to permit inspection and screening of a myriad of veh icles (cars, trucks, etc.) and seacontainers, luggage, and personnel. CBI is e specially useful when the inspector does not have the opportunity to access the target from the backside. One cause of concern is the exposure of humans to ionization radiation; x ray backscattering can be considered because adjustments to parameters, type of image and quality, effective penetration, and x ray tube voltage can be suitably altered so the target object receive a relatively low radiation dose. As a consequence, this adds significantly to the application spectrum for this imaging technique. 2 Detector Modes of Operation In 2006, Dugan et al.20 submitted a NASA final report with regards to scattered x ray imaging research. In this report two modes for detector operat ion were identified: pulse (counting) and current (integrating) mode.

PAGE 43

43 Pulse M ode In pulse mode, each individual photon that generates a detector effect is counted as a pulse. Before the detector can count the next pulse, the current pulse must effectively die away. The pulse height does not matter, as each pulse is individually counted.1 In the report made by Dugan et al. 200620Current M ode it was mentioned that each count is equally weighted regardless of the energy of the x ray being detected. The number of counts that is received in each pixel is directly proportional to the pixels intensity Although the pixel intensity increases with higher counts the detector in pulse mode does have its limits. For pulse mode detectors there is a preamp that processes the sig nal from the PMT. This preamp has an RC circuit for collecting charge from the PMT. The time constant for the RC circuit needs to be long enough to collect the photons and resulting charge from an interaction event, but short enough to distinguish betwee n individual radiation interaction event s If the count rate becomes too high, it becomes impossible to distinguish between individual radiation interaction event s, and pulse pileup occurs. In order to reduce the effects pulse pile up, current mode can be used the second detector mode i.e. current mode can be used. In the Dugan et al. 200620 report, whe n the detected count rate is high such that pulse pileup occurs, current (or integral) mode can be used. In current mode, individual pulses are not counted. Instead, the pulses are collected and integrated over a period of time.20 This reduces the problem of pulse pileup because each additional pulse will simply add to the area under the curve, allowing for the continued inclusion of these pulses without losing data or having to wait on the previous pulse to die. Usually current mode operation is only used when the radiation field is too high to count, but for RSD imaging, there is an added benefit to current mode. With regard to any drawbacks 1

PAGE 44

44 for current mode, the integration response time is considered to be large when compared to the time between individual events and as a result the ability to distinguish between individual interactions is lost. In Dan Shedlock s 2007 dissertation20 2, it was mentioned that for the RSD scanning systems built for Lockheed Martin Space Systems Co. and NASA it was found that the use of current mode significantly increase the contrast of deep void defects in SOFI without adversely affecting the contrast o f shallow void defects.2 I n pure count mode the weighting is fixed. But, if a spectrum is acquired pixel intensity can then be weighted by energy allowing the user to define or vary the importance of different energy groups.2 Energy weighting can be achieved by mapping varying voltage levels to intensity levels in pixels of an image. As a result, the pi xel intensity will be weighted toward the higher energy x rays as shown in Figure 31. Such spectral energy weightin g into the image contrast can be deemed as a form of collimation which help s in the detection of deeper features in a target. 2,20 Each of the spectral energy curves shown in the figure has the same total counts under the curve. Because in current mode the intensity is weighted by both energy an d count rate, the spectral curve with the higher energy x rays has a higher intensity value in the pixel. Dugan et al. 200620 also state s that higher energy x rays have a different scattering history compared to lower energy particles. In some cases these higher x rays may have penetrated deeper or have a higher probability of interacting with a defect. If this is the case, then weighting the signal strength based on energy can improve the ability to detect certain features.20

PAGE 45

45 Figure 3 1. Energy weighted current mode contrast for a single pixel for varying spectral curves 2,20

PAGE 46

46 CHAPTER 4 IMAGE CONTRAST STUDY Experimental Settings X R ay Generator The x ray generator used on the current BSX scanner is the Yxlon MXR 160/22. In the case of the image contrast study where foam (SOFI) was used as the target, the optimum settings for such examination appeared to be best at a tube voltage in the range of 55 kVp to 60 kVp. Both the NaI and YSO detector were employed in this study despite the fact that the literature indicates that YSO i s inferior to NaI, with a relative yield that i s approximately 38% of NaI. Later studies revealed that the determination of the relative light yield data is based on photons with energies in the MeV range. For our x ray generator settings, in the 55 kVp to 60 kVp range the average photon energy is approximately 30 keV and the average energy for detected backscatter photons from the foam block is also around 30 keV. Because of the low photon energy the YSO detector has performed better than the NaI detector for shallow or near surface defects This fact can be explained by curves that were generated showing the photon scattering to absorption ratio as a function of energy for NaI and YSO (F igure 4 1) Figure 4 1. Photon scattering to absorption ratio in NaI and YSO as a function of energy.

PAGE 47

47 Good detection efficiency is achieved for a low scattering to absorption ratio. Considering the range of 25 to 35 keV in Figure 4 1, the YSO detector has a lower scattering to absorption ratio compared to NaI. At about 33 keV there is a sudden drop in the scattering to absorption ratio for NaI and this can be attributed to the photoelectric absorption K shell edge in I at 33 keV. For energies higher than 33 keV the NaI displays a more favorable scattering to absorption ratio tha n YSO. Due to this behavior, there is a concern whether an increase in tube voltage would adversely affect the YSO detector performance or whethe r a decrease in tube voltage would adversely affect the NaI detector performance. To this end, measurements were performed on the poured foam block panel at a 60 kVp tube voltage and quantitative assessments were made of image contrast. Data Sets Three dat a sets were acquired and each data set corresponds to a different foam overlay thickness ( Table 4 1 ) The sample distance, which is defined as the distance between the surface of the SOFI block and the bottom of the detector sleeve, was kept the same, 2 inches for all the data sets. Table 4 1. Series used in the i mage c ontrast s tudy Series Foam Overlay 1 2 inches 2 4 inches 3 6 inches Several runs in each data set were made. This was done by gauging the image quality per run and then making slight changes to determine the conditions for the best image for contrast analysis. The changes encompassed either a change in sleeve collimator le ngth in one or more detector or looking at the image fr om a different detector. Tables 42, 43 and 44 indicate which detector s w ere used along with pertinent parameters that need to be considered

PAGE 48

48 Table 4 2. Series 1 poured block 2 inches f oam o verlay NaI YSO Date of Data Collection 10/21/2008 10/27/2008 Run 1 3 Count Mode Detector 4 Detector 8 Integral Mode Detector 9 Detector 10 Sleeve Collimator Length 0.8 cm 0.0 cm Table 4 3. Series 2 poured block 4 inches f oam o verlay NaI YSO Date of Data Collection 10/27/2008 10/27/2008 Run 5 5 Count Mode Detector 2 Detector 8 Integral Mode Detector 9 Detector 10 Sleeve Collimator Length 0.5 cm 0.5 cm Table 4 4. Series 3 poured block 6 inches f oam o verlay NaI YSO Date of Data Collection 10/2 6 /2008 10/2 6 /2008 Run 6 6 Count Mode Detector 1 Detector 8 Integral Mode Detector 9 Detector 10 Sleeve Collimator Length 1.0 cm 0.5 cm In the LABVIEW program two additional detector tabs were included. This was to facilitate the integral mode for two detectors. For instance if NaI detector 1 and YSO detector 5 were our detectors of interest in LABVIEW the detector tab labeled detector 9 corresponded to th e first detector under consideration in integral mode i.e. NaI detector 1 and the detector tab labeled detector 10 corresponded to the second detector under consideration in integral mode i.e. YSO detector 5. Additional Information All data sets were acquired at the same x ray generator settings and the exposure time per pixel was kept the same throughout. Each scan took on average an approximate time of 30 minutes. Table 45 summarizes the additional settings that were used to obtain each data set.

PAGE 49

49 Ta ble 4 5. Settings used for SOFI block scans X ray Voltage 60 ( kVp ) X ray Current 12.0 (mA) Focal Spot Size 5.5 (mm) Exposure time per pixel 0.150 (s) Scan Velocity 13.33 mm/s Beam Aperture 1.5 (mm) Target Material Poured Block Panel (SOFI) The poured block panel SOFI (Spray on Foam Insulation) used has the approximate dimensions of 12 inch x 12 inch x 2 inch thick block of foam. The poured block panel (SOFI block) has a series of holes with varying depths and varying diameters and is shown in Fi gure 4 2. The diameters of the holes are 0.250 in., 0.38 in. 0.5 in., 0.65 in. and 0.75 inches. The depths of the holes are 0.125 in., 0.25 in., 0.375 in., 0.5 in. and 0.65 inches. Figure 4 2. Photograph of poured block foam panel

PAGE 50

50 B ackscatter x ray scans were acquired over the entire region and the image contrast from the unprocessed scans was then compared with the contrast from the processed scans. This was done for both NaI detectors and YSO detectors while in count rate and current rate mode. The trend of the image contrasts were compared among three categories ; behavior and trend of contrast with increasing hole depth at constant diameter, behavior and trend of contrast with increasing hole diameter at constant depth, and behavior and trend of c ontrast with increasing hole diameter and hole depth. This latter category considers the holes that lie along the diagonal of the SOFI block. For the first category, three diameters (0.25 in., 0.5in., and 0.75 inches) were considered, and for the second ca tegory only two depths (0.25 in. and 0.5 inches) were considered. Factors Affecting Count Rate The count rates achieved during the BSX scanning process are heavily dependent on a number of parameters. The most obvious parameters are the generator voltage and current which affect tube x ray emission rate. B ut due to the complexity of the backscatter process and the varying sensitivity of the detectors changes in the x ray emission rate may not necessarily translate into identical changes in the detector co unt rate (E. Dugan, D. Shedlock, N. Sabri, C. Meng and G. Gueorguiev, Research to increase speed of backscatter x ray inspections, George C. Marshall Space Flight Center Final Report, Award Number NNM07AB25P Mod1, July October 2007). Count rates can al so be affected by the amount of detector collimation and by the distance of the detector from the target object surface. These latter parameters played a significant role in achieving the desired count rate for better counting statistics and, t herefore, th e collimation on some detectors was not necessarily the same for each scan.

PAGE 51

51 Image Contrast Contrast Contrast is affected by the changes in a target material electron density and/or scattering to absorption ratio. The relative contrast of the target material and its associated features is also determined by the multiple scatter photons, detector collimation, and feature geometric location.2For a quantitative comparison, the relative contrast of each flaw is calculated with respect to the surrounding background. In this study contrast is defi ned as the difference between the number of counts in a region of interest and background divided by background. For pencil beam CBI the change in count rate or signal intensity at the detector is directly proportional to the change in contrast in the imag e. Changes in the backscatter photon field intensity results in contrast changes in images. These changes are caused by differences in absorption and scattering cross sections along the path of the scattered photons. 2 Background Background Signal rast lativeCont ) ( Re A positive contrast indicate s the defect is brighter than background, likewise a negative contrast indicate s the defect is darker than the background. Since the count rate for the foam panel is lower over a flaw or defect indicating a negative contrast, the absolute value of the difference was found before dividing by background for this particular study Background is referred to the number of counts found over the area surrounding the defect. Equation 41 gives the equation for contrast and Fi gure 4 3 shows the regions defined as background. Figure 43 is an actual image taken from Series 1 2 inches foam overlay and NaI detector 4. Thus for each hole present in the SOFI block, four regions surrounding the hole were measured for background counts.

PAGE 52

52 || 1234 4 abscountaveragebkgd imagecontrast averagebkgd bkgdbkgdbkgdbkgd averagebkgd (4 1) Figure 4 3. Background r egions surrounding a hole or defect Methodology The primary goal of this study was to compare the image contrast obtained by detectors in both the count mode and current mode. This was taken a step further and the images were divided into two sections: Processed and Unprocessed (Raw Data) Images obtained directly from the detector and presented in the LABVI EW software are considered to be unprocessed (raw) images. Processed images are defined as those images that have been filtered using BKGD1 BKGD 2 BKGD 4 BKGD 3

PAGE 53

53 imaging s oftware called X ray Bat commonly used to improve the appearance of BSX images. The raw images were first flatte ned and the median filter was applied. Since the poured block panel contained holes of varying depths and diameters it made sense to observe the trends of image contrast found in: 1. Image contrast of increasing hole depth at constant hole diameter (Figure 4 4) 2. Image contrast of increasing hole diameter at constant hole depth (Figure 4 4) 3. Image contrast along the diagonal of the poured block i.e. contrast observed as hole depth and hole diameter increases The red line of interest indicated in Figure 4 5 shows the holes of interest (i.e. along the diagonal) Figure 44. Hole s with constant diameter (D) lie in the x direction and holes with constant depth lie in the y direction D=0.75 D=0.50 d=0.25 d=0.50 X dir. Y dir.

PAGE 54

54 Figure 45. Line along the diagonal of the block indicates the holes with increasing depth and increasing diameter. Results and Trends T he images that were collected from the NaI and YSO detector are shown in Figure 46 through Figure 4 11. For gray scale images, a decrease in signal (low scattering in a low density medium i.e. air) is indicated by an area of dark contrast (negative contrast) Alternatively, a n increase in signal is indicated by an area of light contrast (positive contrast) Thus, the dark regions identified in the Figure 4 5 are the holes (or cavities or voids) in the poured block (SOFI block)

PAGE 55

55 Figure 4 6. Raw images from NaI and YSO in count mode (Series 12 inch thick foam overlay ) Figure 4 7. Raw images from NaI and YSO in integrating mo de (Series 1 2 inch thick foam overlay)

PAGE 56

56 Figure 4 8. Raw images from NaI and YSO in count mode (Series 2 4 inch thick foam overlay ) Figure 4 9. Raw images from NaI and YSO in integrating mode (Series 2 4 inch thick foam overlay)

PAGE 57

57 Figure 4 10. Raw images from NaI and YSO in count mode (Series 3 6 inch thick foam overlay ) Figure 411. Raw images from NaI and YSO in integrating mode (Series 3 6 inch thick foam overlay)

PAGE 58

58 The ease of cavity distinction and image contrast decreases as the thickness of the foam overlay increases. For instance, Figure 4 6 (Series 1 with a 2 inch thick foam overlay) displays a clear image with a high percentage of image contrast. The ability fo r ease of cavity distinction translates to the magnitude of percentage of image contrast In contrast, Figure 48 (Series 2 with a 4 inch thick foam overlay) indicates a lower percentage of image contrast and consequently cavity distinction was much more di fficult to ascertain Apart from observing the trend of the magnitude of image contrast from series to series, it is just as important to consider the trend and magnitude of contrast within each series and with respect to change in cavity dimensions Image contrast changes throughout due to the varying diameter and depth of the holes. In general, the trend observed for each series was consistent and it indicates as the diameter of the hole and even the depth of the hole increases the magnitude of the relati ve contrast increases This increase in relative contrast can be attributed to the reduced scattering in a void (low density) region as oppose to a higher amount of scattering in a uniform target region (high density). Despite the consistent trend indicate d in all three series, each series had its own unique characteristics due to the varying thickness of the foam overlay. This resulted in the subsurface defects appearing to be deeper from series to series, and as a result affected the performance of both t he NaI and YSO detector The performance of either detector is indicated by the magnitude of image contrast percentage. In all cases (Series 1, Series 2, and Series 3) the comparison of two of the same detectors in different mode s count mode and integra l mode (NaIINT & NaICR and YSOINT & YSOCR) and two different detectors in the same mode (NaICR & YSOCR and NaIINT & YSOINT) were made For the raw images in S eries 1 the image contrast generally increased when the d iameter is constant and depth increases, generally increased when the depth is constant and diameter

PAGE 59

59 increases, and generally increased when both depth and diameter increased. For the varying depths at constant diameter, three diameters were considered (D=0.75, D=0.5 and D=0.25 inches). The trend can be observed in Figure s 412 to 415 for the constant diameter D of 0.75 inches (raw image). The trend in image contrast at varying depths and constant diameter D=0.5 inches and the trend in image contrast at varying depth and constant diameter D=0.25 inches can be observed in Appendix A. In the appendix it presents trends in image contrast for unprocessed (raw image) and processed image. Figure 4 12. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int for unprocessed images at constant diameter (D) of 0.75 inches

PAGE 60

60 Figure 4 13. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.75 inches Figure 4 14. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.75 inches

PAGE 61

61 Figure 4 15. Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.75 inches For the varying diameters at constant depth, two depths were considered (d=0.5 and d=0.25 inches). The trend for the case of constant depth (d=0.5 inches) and increasing diameter is observed in Figures 416to 419. The plots for the second case, constant depth of 0.25 inches (Series 1 unprocessed) is shown in Appendix A. Figure 4 16. Serie s 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant depth (d) of 0.5 inches

PAGE 62

62 Figure 4 17. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches Figure 4 18. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant depth (d) of 0.5 inches

PAGE 63

63 Figure 4 19. Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches Finally for the raw images in Series 1, Figures 420 to 423 indicates the generally increasing image contrast when the depth and diameter increase. For the raw images, image contrast was higher from detectors in the integral mode as compared to the count rate mode. In particular, the YSO detector in integral mode performed best and gave high magnitude of relative contrast. In comparison, for the same parameters o f increasing depth and diameter the YSO detector performed much better than the NaI detector in either mode count rate or integral. This is indicated by observing the axes in Figure 4 20 (NaI in both modes) and Figure 421 (YSO in both modes). The image contrast percentage is lower in Figure 4 20 compared to Figure 4 21.

PAGE 64

64 Figure 4 20. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at increasing depth (d) and diameter (D) Figure 4 21. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at increasing depth (d) and diameter (D)

PAGE 65

65 Figure 4 22. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at i ncreasing depth (d) and diameter (D) Figure 4 23. Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at increasing depth (d) and diameter (D)

PAGE 66

66 With respect to the processed images (graphs shown in Appendix A) in series 1 (2 inch foam overlay) for the cases with increasing depth or increasing diameter the magnitude of the contrast were very similar in both the count rate mode and integral mode s Processing the images generally caused the image contr ast to be about the same for images acquired in both detectors either in count mode or integral mode. Thus far, all the graphs indicate that the YSO detector has performed better than the NaI (either in count or current mode ). This can be attribute d to the fact that the conditions (2 inch foam overlay) and experimental settings (mentioned earlier) played a role in achieving low photon energy and therefore increasing the efficiency of the YSO. For the case (processed) of comparing at the same time inc reasing depth and diameter, image contrast for images obtained by detectors in integral mode were higher when compared to detectors in count rate mode. At shallow depths and small diameters the difference in image contrast for the two mode s was generally not that large but as the depth ge t s deeper and the diameter increases there was an increasing improvement in contrast for the integra l mode compared to the count mode In Series 2 (4 inch foam overlay), the image contrast for the raw images generally fo llowed the same trend as in Series 1. An exception for this series was the fact that the image contrast from detectors in integral mode was significantly higher than the image contrast from detectors in count mode. Also, image contrast from the YSO is now not always better than from the NaI. Series 1 did show a difference of contrast from the two m odes when compared to Series 2, but in Series 2 th is difference was much m ore pronounced. Figures 4 24 to 427 show the trend observed for the unprocessed images in the case of maintaining the diame ter constant at D = 0.75 inches.

PAGE 67

67 Figure 424. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.75 inches. Figure 4 25. Series 2: C omparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.75 inches.

PAGE 68

68 Figure 4 26. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.75 inches. Figure 4 27. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at const ant diameter (D) of 0.75 inches. Figures 4 28 to 431 show the trend observed for the unprocessed images in the case of maintaining the depth constant at d = 0.5 inches

PAGE 69

69 Figure 4 28. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant depth (d) of 0.5 inches. Figure 4 29. S eries 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches.

PAGE 70

70 Figure 4 30. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant depth (d) of 0.5 inches. Figure 4 31. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches. Figures 4 32 to 435 show the trend observed for the unproce ssed images in the case of increasing the depth and diameter. For all the other cases, the graphs depicting the trend are shown in Appendix A. As expected the detectors in integral mode continued to perform better

PAGE 71

71 that their counterparts in count mode. One obvious difference in this particular series is the fact that the NaI in integral mode appears to perform just as the YSO in integral mode for wider voids. Figure 4 32. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at increasing depth (d) and diameter (D). Figure 4 33. Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at increasing depth (d) and diameter (D).

PAGE 72

72 Figure 4 34. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at increasing depth (d) and diameter (D). Figure 4 35. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at increasing depth (d) and diameter (D).

PAGE 73

73 For the processed images in Series 2 (plots shown in Appendix A) the image contrast from either mode was similar for deeper wider voids in the SOFI. For shallow voids it appears that the YSO in integral mode gave better image contrast results. For Series 3 (6 inches of foam overlay) the trends observed in image contrast were not as consistent when compared to the earlier series. The fact that the thickness of the foa m overlay was quite large could cont ribute to some of the behaviors noted. Figure s 436to 439 considers the image contrast at a constant diameter (0.75 inches) and varying depths. The trend for unprocessed of image contrast at the other two constant dia meters (0.5 inches and 0.25 inches) can be shown in Appendix A. For the larger diameters, image contrast increased with increasing depth, as expected. At a smaller diameter, the image contrast was maintained constant despite the increasing depth. Figu re 4 36. Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.75 inches.

PAGE 74

74 Figure 4 37. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unpr ocessed images at constant diameter (D) of 0.75 inches. Figure 4 38. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.75 inches.

PAGE 75

75 Figure 4 39. Series 3: Comparing relativ e contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.75 inches. Detectors that operated in integral mode produced higher contrast, but this time it was the NaI detector that gave higher image contrast when compared to the YSO detector. Referring to Figure 4 1, this behavior can be explained by the fact that photons now have to traverse through a thicker region of foam before reaching the detector. The depth of subsurface features plays a significant rol e with regard to a detectors efficiency. Since the subsurface defect depth is now larger when compared to the other series (Series 1 and Series 2) then the NaI detector performance is now higher than the YSO detector performance. Higher energy x rays may have penetrated deeper or have a high probability of interacting with a defect. It is more than likely that the energy of the backscattered photons is higher than 33 keV and this result in a shift of detector performance from the YSO to the NaI. In this energy range, the scattering to absorption ratio is now lower for the NaI as compared to the YSO and therefore this indicates better detector performance.

PAGE 76

76 Figure s 440 to 443 show the trend of image contrast for increasing diameter at a constant depth of 0.5 inches. All other remaining graphs for observing the trend of image contrast for constant depth of 0.25 inches is given in Appendix A. Figures 4 40to 443 contain some erratic behavior. The magnitude of image contrast does not appear to follow a definite trend as shown in Series 1 (2 inches thick foam overlay) and Series 2 (4 inches thick foam overlay) It can be approximated that as the voids get wider the image contrast appears to improve. The NaI detector also seems to work best as compared to the YSO detector, under the conditions specified for Series 3 (6 inches thick foam overlay) Figure 4 40. Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant depth (d) of 0.5 inches.

PAGE 77

77 Figure 441. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches. Figure 4 42. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant depth (d) of 0.5 inches.

PAGE 78

78 Figure 4 43. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant depth (d) of 0.5 inches. Figures 4 44 to 447 shows the trend to follow a more consistent behavior. Following a drop in the magnitude of image contrast at Point 2 (diameter (D) of 0.38 inches) the image contrast improves significantly with the NaI detector (in either count or current mode) performing much better than the YSO. With the case of the processed images, Series 3 (6 inches thick foam overlay) were similar to Series 1 (2 inches thick foam overlay) and Series 2 (4 inches thick foam overlay) The image contrast (processed images ) has about the same magnitude when images were collected either by count or current mode. The trend observed remain about the same as was observed in the raw images. This trend for the processed images for Series 3 is shown in Appendix A.

PAGE 79

79 Figure 4 44. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at increasing depth (d) and diameter (D) Figure 4 45. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at increasing depth (d) and diameter (D ).

PAGE 80

80 Figure 4 46. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at increasing depth (d) and diameter (D). Figure 4 47. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at increasing depth (d) and diameter (D).

PAGE 81

81 CHAPTER 5 UF BSX IMAGING SYSTEM DEAD TIME Introduction Initially the purpose of this study was to determine the system dead time for the BSX system. This study ha s evolved to encompass the physics, statistical analysis of the effects of pile up and the individual testing of components that comprise the UF BSX i maging system to deduce the limiting factor that prevents the YSO detector from fully achieving its design ed maximum count rate limit of 2 million (cps). Research at UF explores mainly backscatter Radiography by Selective Detection (RSD) and this technique is employed by many industries today. As mentioned before, RSD uses a pencil beam source and it obtains data via a rastering technique. Despite RSD s unique capability for image acquisition, there have been some questions regarding count rate limits that may possibly cause image data to be lost. One of the recent questions being addressed in this particular s tudy is the magnitude of the BSX system up at high counting rates arises and the uncertainty about whether or not image data are being lost or jeopardized has also increased. In nearly all detector systems, there will be a minimum amount of time that must separate two events in order to be recorded as two separate pulses. In some cases the limiting time may be set by processes in the detector itself, and in other cases the limit may arise in the associated electr onics.14 Considerations in this study were made to include associations with both the detector and the electronics to create a combined system dead time. The minimum recorded event time to ensure separation of two pulses is usually called the dead time of the counting system. Because of the random nature of radioactive decay, there is always some probability that a true event will be lost because it occurs too quickly following a preceding event. These dead time losses can become

PAGE 82

82 rather severe when high counting rates are encountered. In order to determined the combined system dead time the UF BSX imaging system must first be identified as being either paralyzable or nonp aralyzable. After such identification an estimate of the system dead time and an an alysis of the effects can be achieved. Materials, Equipment, Settings and Measurements Two target materials were employed in the experimental setup for the calculation of the UF BSX system dead time; an uniform foam block of dimensions 30x30x11 (cm) and a uniform nylon block of dimensions 15x15x1 (cm). The equipment comprised of the UF BSX imaging system and can be seen in Figure 5 1 along with the image acquisition and processing i.e. LABVIEW counters and computer. Figure 51. UF BSX i maging e quipment s etup t he component specifications are all identified in Appendix B Experimental s ettings Tables 51 and 5 2 indicate the experimental settings used for the scans in each series of measurements Within each series the settings were held constant with the exception of x ray generator current. Typically most of the experimental settings were held constant from series to series except the aperture size that was changed in a case by case analysis. The voltag e of 60 kVp was also maintained throughout. There was one exception to this setting and as indicated in

PAGE 83

83 Table 5 3 the voltage settings was reduced to 55 kVp for Series 4 YSO Nylon. This reduction in voltage was to compensate for the sudden increase in aperture size of 2.5 mm to 4.0 mm (beam aperture size). Since the calculation of the system dead time required consistency, Series 4 YSO Nylon was re addressed and the voltage setting were reset to 60 kVp. Such results are presented in Series 5 YSO Nylon Table 51. NaI detectors f oam & n ylon t argets Series 1 (Foam) Series 2 (Foam) Series 1 (Nylon) Voltage (kVp) 60 60 60 Beam Aperture Size (mm) 1.5 2.0 1.5 Pixel Size (mm) 1.0 2.0 2.0 Collimation (cm) 0 {except detector 2 at 1.0cm} 0 {except detector 2 at 1.5cm} 0 {except detector 2 at 2.0cm} Fins (degrees) 0 0 0 Table 52. YSO detectors f oam & n ylon t argets Series 1 (Foam) Series 1 (Nylon) Series 2 (Nylon) Voltage (kVp) 60 60 60 Beam Aperture Size (mm) 2.5 2.0 2.0 Pixel Size (mm) 2.0 2.0 2.0 Sleeve Collimator length for YSO 5 is 1.7 cm for the Nylon Series For Nylon Series 2 and later peak output voltage on oscilloscope changed from 4 volts to 7 volts Table 53. YSO detectors c ontinuation of YSO s eries but with alterations Series 3 (Nylon) Series 4 (Nylon) Series 5 (Nylon) Voltage (kVp) 60 55 60 Beam Aperture Size (mm) 2.5 4.0 4.0 Pixel Size (mm) 2.0 2.0 2.0 Alterations include: a) Fast Filter baseline restore time constant* was reset to the fastest value by Dan Ekdahl, and b) YSO Fast Fi lter attenuator was turned off. Baseline restoration For the testing o f the YSO detectors in Series 3, Series 4 and Series 5 some alterations were made such as the restoration of the baseline in the fast filter amplifier. The primary purpose of baseline restoration is to return the true zero of the baseline between pulses in as short a time

PAGE 84

84 as possible. From Series 3 YSO Nylon and beyond, the fast filter baseline restore time constant was reset to its fastest value, an estimated 1.10x1007* seconds (refer to footnote made in Table 53). This reset not only allowed a faster time in returning to true zero of the baseline between pulses, but as a bonus, the use of baseline restoration greatly reduces the deleterious effects of any low frequency disturbances14 which may be present along with the signal. Baseline restoration can be explained with the use of an equivalent circuit that represents a baseline restorer as shown in Figure 52. Figure 52. Equivalent circuit of a baseline restorer In principle, the switch (S) is open during the duration of each pulse, and its closing restores the output voltage to zero, with a time constant given by the product of (R+Ro) and the coupling capacitance C, where R is the equivalent series resistance of the switch and Ro is the output impedance of the operational amplifier. To be effective, baseline restoration must take place near the end of the signal chain so that no further AC coupling takes place between the restorer and the point at which the pulses ar e analyzed or measured. Baseline restorers are therefore most commonly found at the output stage of linear amplifiers. For this study, it is worth

PAGE 85

85 mentioning that the restorer is located in the fastamplifier, which supplies pulses directly to the Quad single channel analyzer (SCA). Table 54. Fast a mplifier baseline restore time constant Resistance 1.1 x10 03 Capacitance 100 x10 12 Time Constant* F RXC = 1.1 x10 0 7 (seconds) Measurements The current settings on the Yxlon generator were varied over the range of 2 mA to 45 mA in small increments. The average of the net count rate (cps) for the eight detectors (4 NaI and 4 YSO) was obtained at each current setting (refer to Equation 51). The average of the net count rate was taken based on the premise that the target was uniform throughout. The pixel dwell time was set at 0.1 seconds for both the NaI Series and the YSO Series. Figure 53. Image acquired for a uniform nylon block

PAGE 86

86 Figure 53 helps to explain the derivation of Equation 51. Since the target material is uniform the expected net counts recorded (NC1, NC2, and NC3 NCR NC NC NC 1 0 33 2 1 ) should be close in magnitude throughout the entire block. To get the average number of net counts for each detector for the uniform block the average of the three net counts were taken (Equation 51). ( 51) Data Analysis Net Count Rate and Current Current is considered to be simply a scaling factor for the number of electrons that can be associated with a count rate. Because of this scaling factor behavior a plot of Net Count Rate (cps) versus Current (mA) can be likened to a plot of Recorded Count Rate (m) (cps) versus True Interaction Rate (n) (cps). With regard to the Foam series, only NaI data could be extracted and not the YSO data. This is due to the fact that the YSO is capable of recording much higher count rates than the NaI detector and the foam was not a good target material to achieve sufficiently high count rates to saturate the YSO detectors. The n ylon tar get material was much better suited for YSO detector analysis and by combining this change in target material with increased aperture size, saturation of the YSO detector was achievable. In this experiment it was important to push the UF BSX Imaging System detectors (NaI and YSO) to their limits in order to achieve high count rates. In order to determine if a system is paralyzable, the observed count rate must be seen to go through a maximum. Alternatively, if the system is nonparalyzable a definite plateau effect will be observed at high count rates before making such an analysis. Taking into consideration that the NaI detector can only record count rates about 2.5 times lower in magnitude than the YSO detector, the NaI detectors were turned off in order t o preserve the integrity of the NaI detectors while YSO testing was being done.

PAGE 87

87 Results NaI Series Foam The plots indicated in this section are obtained by plotting Net Count Rate (cps) versus Current (mA). There are eight detectors in total, Four NaI detectors and four YSO detectors. T he NaI detectors are numbered for ease of distinction, 1 through 4 and the YSO detectors are also numbered 5 through 8. Therefore, in the NaI series Foam there will be four plots indicating a different NaI detector. Figure 54. Series 1 NaI detector 1 f oam (1.5 mm beam aperture size )

PAGE 88

88 Figure 5 5. Series 1 NaI detector 2 foam (1.5 mm beam aperture). Figure 5 6. Series 1 NaI detector 3 foam (1.5 mm beam aperture).

PAGE 89

89 Figure 5 7. Series 1 NaI detector 4 foa m (1.5 mm beam aperture). In Figure s 5 4, 56, and 5 7, with the exception of NaI detector 2 shown in Figure 55, a plateau in the curves can be seen which implies that a higher current setting is required to not only achieve a maximum count rate but to also detect if there is a definitively a decrease in count rate as the current magnitude continues to increase. For the employed Yxlon MXR 160/22 x ray generator, at 60 kVp and with a 5.5 mm electron beam focal spot, the maximum achievable current is 45 mA. Because of the results obtained in this first series (Series 1 NaI Foam) and the limitation of the current settings, equipment settings need ed to be adjusted to facilitate a higher observed count rate. Such adjustments include, increasing the beam aperture size and also employ the use of a different target material of higher density (e.g. Nylon or Aluminum) than t he SOFI (Spray On Foam Insulation) block.

PAGE 90

90 Figure 5 8. Series 2 NaI detector 1 foam (2.0 mm beam aperture). Figure 5 9. Series 2 NaI detector 2 foam (2.0 mm beam aperture).

PAGE 91

91 Figure 5 10. Series 2 NaI detector 3 foam (2.0 mm beam aperture). Figure 5 11. Series 2 NaI detector 4 foam (2.0 mm beam aperture).

PAGE 92

92 As indicated in Figure s 58to 511 and Table 51, the aperture size of the beam was increased from 1.5 mm to 2.0 mm for the Series 2 NaI Foam. This increase in beam aperture size was t o make possible a higher observed count rate. Consequently a maximum count rate followed by a definite decrease in counts can be observed in Figures 58to 511 (plots of Net Count Rate (cps) versus Current (mA)) Such behavior indicates that the U niversit y of F lorida BSX (Backscatter X ray) imaging s ystem follows a paralyzable model. In an attempt to achieve an even higher count rate, but at a lower current the foam form (SOFI) target material was switched to a higher density material, a uniform nylon block. The higher density nylon target material combined with sufficient increased beam aperture size should be able to contribute to a sufficiently high observed count rate within the current range for the employed x ray generator (Yxlon MCG41) such that a maximum recorded count rate and then a definitive decrease can be easily observed within the current range from 2 mA to 45 mA. SOFI and Nylon Differences The Spray onFoam Insulation (SOFI) block that was used to conduct the initial NaI Series is a material that has a rough composition of carbon (C), nitrogen (N), chlorine (Cl), fluorine (Fl) oxygen (O), and hydrogen (H). The density of the materials is estimated to be 0.03 g/cm3; therefore, due to the extremely low density a photon will have signi ficantly fewer interactions, or a longer traveling time within the material before scattering and returning to the detector s (NaI or YSO) to be recorded. Alternatively the density of nylon (~1.14 g/cm3) is almost 40 times larger than that of SOFI (foam for m) and therefore it is reasonable to expect a higher quantity of photon interactions within the higher density target material shorter traveling time for the photons within the material before scattering and an increased proportion of backscattered photons returning to either detector (NaI or YSO)

PAGE 93

93 NaI Series Nylon Figure 512. Series 1 NaI n ylon (1.5 mm beam aperture) Figure 5 13. Series 1 NaI n ylon (1.5 mm beam aperture).

PAGE 94

94 Figure 5 14. Series 1 NaI n ylon (1.5 mm beam aperture). Figure 515. Series 1 NaI n ylon (1.5 mm beam aperture). In Figures 5 13, 514 and 515 the observed count rate is definitely seen to go through a maximum and then a decrease with the use of nylon as the target material. Such behavior indicates that the UF BSX S ystem follows a p aralyzable model.

PAGE 95

95 YSO Series Foam For the NaI detector the realistic design count rate limit is ~750,000 cps, but the predicted ideal count rate limit for the YSO detector is much higher, ~2.00E+06 cps. As a result the target used for most of the YSO series was nylon and the beam aperture size was larger as compared to the size used when performing tests on the NaI series. In Figure s 516, 517 and 518 the observable count rate was either still increasing or not definitively decreasing, even though the maximum current value was reached. This is explained by the small aperture size, choice of target material, and a potential problem in the electronics that limited the actual maximum count rate achievable by the YSO. Throughout the YSO seri es testing there was a concern : the predicted 2 million cps limit was not close to being achieved. Separate tests were performed to deduce what was causing this limitation. It was suggested to examine each component (indicated in Figure 51) separately and test that respective component to de termine where the bottleneck for the YSO is coming from. The testing of the individual components and the analysis is discussed later in this thesis. Figure 5 16. Series 1 YSO f oam (2.5 mm beam aperture).

PAGE 96

96 Figur e 5 17. Series 1 YSO f oam (2.5 mm beam aperture). Figure 5 18. Series 1 YSO f oam (2.5 mm beam aperture).

PAGE 97

97 Figure 5 19. Series 1 YSO f oam (2.5 mm beam aperture). The data provided in Series 1 YSO Foam (Figure s 516to 519) will not be used for the system dead time calculations. Instead data will be used when the target material chosen was nylon (higher density material) Since the YSO detector is capable of counting in a range approximately 2.5 times higher than that of the NaI detector a higher density material (nylon) was able to provide the measured count rate and to indicate the definitive trend observed (maximum count rate followed by a definitive decrease in counts) within the current range (2mA 45mA). YSO Series Nylon Figure s 520 to 523 indicates the trends observed in Series 1 Nylon for each YSO detector. The beam aperture size was 2.0 mm and the YSO Fast Filter amplifier peak output voltage was adjusted on the oscilloscope and it was set at an estimated 4 volts.

PAGE 98

98 Figure 5 20. Series 1 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 4 volts. Figure 5 21. Series 1 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscil loscope set at 4 volts.

PAGE 99

99 Figure 5 22. Series 1 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 4 volts. Figure 5 23. Series 1 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 4 volts. For the results indicated in Figure s 524 to 527, the beam aperture size remained at 2.0 mm, but the YSO Fast Filter peak output voltage on scope was res e t at an estimated 7 volts. This

PAGE 100

100 estimate of 7 volts was k ept the same for Series 3 YSO, Series 4 YSO (Figure s 528 to 531 and Figure s 532to 535 respectively ) and Series 5 YSO. The only changes that were made for Series 3 and Series 4 (YSO) were the aperture size as shown in Table 53 and the fast filter baseline restore time constant which was reset to its fastest value of 1.1x1007 (seconds) by Dan Ekdahl. Figure 5 24. Series 2 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts. Figure 5 25. Series 2 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts.

PAGE 101

101 Figure 5 26. Series 2 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts. Figure 5 27. Series 2 YSO n ylon (2.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts.

PAGE 102

102 An additional change was made to the scans for Series 4 and that was to reduce the voltage from 60 kVp to 55 kVp. For comparison purposes were slightly inconsistent when compared to other valid series. In experiments such as these it is prudent to keep variables constant and the lowered voltage might have created some inconsistencies At the time the scans (Series 4 YSO) were performed at 55 kVp it was thought that the reduced voltage mig ht be needed to compensate for the much large r beam aperture size of 4.0 mm values are being compared from series to series another series of scan s (Series 5 YSO) were performed at 60 kVp (just like for the previous series) and the dead time a nalysis was performed and shown in a later section. Figure 5 28. Series 3 YSO n ylon (2.5 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds).

PAGE 103

103 Figure 5 29. Series 3 YSO n ylon (2.5 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). Figure 5 30. Series 3 YSO n ylon (2.5 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds).

PAGE 104

104 Figure 5 31. Series 3 YSO n ylon (2.5 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007The next four figures represent the plot of Net C ount Rate (cps) versus Current (mA) for Series 4 for each of the four YSO detectors : (seconds). Figure 5 32. Series 4 YSO n ylon ( 4.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds).

PAGE 105

105 Figure 5 33. Series 4 YSO n ylon ( 4.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). Figure 5 34. Series 4 YSO n ylon ( 4.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds).

PAGE 106

106 Figure 5 35. Series 4 YSO n ylon ( 4.0 mm beam aperture) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007Judging by the graphs shown for the NaI Series and YSO Series, the plots have clearly indicated that the UF BSX imaging system is paralyzable. A counting system that follows a paralyzable response indicated a maximum being observed followed by a definite decrease in recorded counts at high current (mA) As a result, very high true interaction rates (n) (cps) result in a multiple extension of the dead period following an initial recorded count (m) (cps), and very few true events can be recorded. (seconds). For consistency and comparison, Series 5 YSO Nylon experimental results will now be presented. At the time of testing YSO detector 7 was not present since it was removed from the system for maintenance. Figures 5 36 to 538 indicate the plot of Net Count Rate (cps) versus Current (mA). The only difference between Series 4 YSO Nylon and Series 5 YSO Nylon is the x ray generator voltage setting at 60 kVp. 14

PAGE 107

107 Figure 5 36. Series 5 YSO n ylon ( 4.0 mm beam aperture and x ray voltage at 60 kVp) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds). Figure 5 37. Series 5 YSO n ylon ( 4.0 mm beam aperture and x ray voltage at 60 kVp ) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007 (seconds).

PAGE 108

108 Figure 5 38. Series 5 YSO n ylon ( 4.0 mm beam aperture and x ray voltage at 6 0 kVp) with the YSO fast filter amplifier peak output voltage on oscilloscope set at 7 volts and the baseline restore constant reset to its fastest value of 1.1X1007Consistent with the behavior of the earlier series, Series 5 YSO Nylon exper imental results have indicated that the UF BSX system does indeed follow a paralyzable model. Next, an interpretation of the results, properties of a counting system following a paralyzable behavior, will be given. The calculated system dead time ( ) value will include associations with both the detector and the electronics involved. (seconds).

PAGE 109

109 CHAPTER 6 DETERMINATION OF UF BSX SYSTEM DEAD TIME Interpretation of Results and Discussion Models for Dead Time Behavior 1. Non Paralyzable Model 2. Paralyzable Model The models indicated represent idealized behavior, one or the other of which often adequately resembles the response of a real counting system. Figure 61 (adapted from Knoll) 14 provides an illustration of two assumed models of dead time behavior for a radiation detector system. In Figure 6 1, six randomly spaced events in the detector system are indicated. At the bottom of the figure is the corresponding dead time behavior of a detector system assumed to be nonperiod of the detector system. Figure 61. Basi c a ssumptions of the two models 14

PAGE 110

110 True events that occur during the dead period are lost and assumed to have no effect whatsoever on the behavior of the detector system. In Figure 6 1, the nonparalyzable detector system would record four counts from the six true interactions. In contrast, the behavior of a paralyzable detector system is shown along the top line of Figure 6 assumed to follow each true interaction that occurs during the live period of the detector. True events that occur during the dead period, however, although still not recorded as counts, are assumed to extend t 6 1, the paralyzable detector system record three counts out of six true events. The following definitions were adopted (from Knoll)14 14n = true interaction rate (cps) to examine the response of the UF BSX System to a steady state source of radiation: m = recorded count rate (cps) Paralyzable Model In the paralyzable case, dead periods are not always of fixed length, so the argument used in the nonparalyzable case cannot be applied. Instead, the rate m is identical to the rate of of intervals between random events occurring at an average rate n is given by [P1 (T) dT] (refer to Figure 6 1). [P1 (T) dT] is the probability of observing an interval whose length lies within dT about T. 1 and infinity (P2 occurrence of such intervals is simply obtained by multiplying n (the true rate) by P2 dT ne dT T PnT ) (1 Equation 61). (6 1)

PAGE 111

111 ne dT T P P ) ( ) (1 2 (6 2) The rate of occurrence of such intervals is then obtained by simply multiplying the above expression by the true rate n nne m 14 (6 3) Equation 63 indicates that calculation of n (true interaction rate) can become cumbersome. The true interaction rate must be solved iteratively if n is to be calculated from measurements of m and knowledge of the system dead time ( ) For this project the analysis of the system dead time ( ) is not difficult since the values of m and n are being experimentally obtained. Since the UF BSX imaging system has been already identified as following a paralyzable response, then using the paralyzable model equation (Equation 6calculated by making the subject of the formula in Equation 63. The recorded count rate which was obtained experimentally is m and determination of the true rate (n) is explained in a later section. Recorded Count Rate versus True Interaction Rate Relationship between Observed Rate and True Rate Figure 62 gives a plot of the observed rate m versus the true rate n for both models nonparalyzable response and paralyzable response It is important to note that when the rates are low (operating at low current) the two models give virtually the same result, observed rate equals to the true rate (m=n) and it is this occurrence that allowed for the estimated calculation of the UF X ray Backscatter Imaging

PAGE 112

112 Figure 62. Variation of the observed rate m as a function of the true rate n for two models of dead time losses A nonparaly which represents the situation in which the counter barely has time to finish one dead period before starting another. For paralyzable behavior, the observed rate is seen to go through a maximum. Although the maximum on the paralyzable curve is not indicated it corresponds to the following an initial recorded count, and very few true e vents can be recorded.14 Plots of Observed Rate (m) versus True Rate (n) 14 In Figure 6 2 the line m=n holds true at considerabl y low rates for the two models. Consequently at very low current settings the current value can be used as a scaling factor to

PAGE 113

113 eva luate the true interaction rate. Table 6 1 indicates the calculation that uses the current (mA) to obtain the true interaction rate n. In almost all the cases, linearity was observed in all the figures given in Chapter 5 between a current from 0 (mA) t o 2 (or 3) (mA). Therefore it will be assumed that linearity is found between these two small values of current. Such linearity follows the equation m=n but it is important to factor in that even at a low current of 3 (mA) there will be some expected loss es For the NaI series, only Series 2 NaI Foam and Series 1 NaI Nylon were considered, since the recorded observable current values were 3 mA and 2 mA, respectively. With regard to Series 1 NaI Foam, the lowest measurement was recorded at 8 mA and t his current setting is too high to assume linearity between 0 mA and 8 mA. Table 61. Sample c alculation using Series 2 NaI f oam to obtain true rate (n) Current (mA) m (cps) n (cps) 0 0 0 3 m m 1 1 =n 4 1 m (4/3)* m 2 1 = n 5 2 m (5/3)* m 3 1 = n 3 For the YSO Series, Series 1 YSO Foam and Series 1 YSO Nylon were not used for system dead time calculations since the lowest current (mA) setting for these two series was considered too high to assume linearity. In Series 1 YSO Foam and Series 1 YSO Nylon the lowest current was at 7 mA. Losses and variation from linearity will occur within such a wide range (0 mA to 7 mA) and therefore future YSO series recorded started measurements at low currents of either 2 mA or 3 mA. In order to obse rve the behavior of the plot of Recorded Count Rate (cps) versus True Interaction Rate (cps) : select detectors from each series (NaI and YSO series) will be considered. All other plots for the remaining detectors in each series (NaI Series and YSO Series) will be shown in Appendix B

PAGE 114

114 Figure 6 3. Measured c ount r ate vs. True c ount r ate: Series 2 NaI 2 f oam (2.0 mm beam aperture). Figure 6 4. Measured c ount r ate vs. True c ount r ate: Series 1 NaI 2 n ylon (1.5 mm beam aperture). Figures 6 3 to 64 indicate the plot of Recorded Count Rate versus True Interaction Rate for NaI detector 2 for Series 2 NaI Foam and Series 1 NaI Nylon.

PAGE 115

115 Figure 6 5. Measured c ount r ate vs. True c ount r ate: Series 2 YSO 5 n ylon (2.0 mm beam aper ture). Figure 6 6. Measured c ount r ate vs. True c ount r ate: Series 4 YSO 8 n ylon (4.0 mm beam aperture and 55 kVp).

PAGE 116

116 Figure 6 7. Measured c ount r ate vs. True c ount r ate: Series 5 YSO 8 n ylon (4.0 mm beam aperture and 60 kVp). Figures 6 5 to 67 indicate the plot of Recorded Count Rate versus True Interaction Rate. Select detectors were chosen as a representative for each of the YSO Series applicable to the calculation of the UF BSX system dead time. All other plots for the other detectors in th e YSO Series are shown in Appendix B. Calculation Figure 62 represents an interpretation for the two models of dead time losses and it will The general approach will be using Equation 63 to 63 utilizes both the observed count rate as well as the true interaction rate, in an exponential relationship. maximum recorded count rate will be used. This quant ity is based on the paralyzable equation (system dead time) can be found. equation or the derivative should be equal. But, due to the uncertainty and inherent count losses

PAGE 117

117 Statistically, both values fall within the same region. Using the paralyzable model equation The paralyzable model is represented by Equation 63. For this method any value of observed count rate (m) and the corresponding value of true interaction rate (n) can be used, but for this analysis the maximum value i.e. mmax and the corresponding ncorr1. Better counting statistics higher counts will be considered The choi ce in using the highest point on the curve is justified by: 2. At low values the curved tend to follow the straight line equation m=n. To avoid this linear trend a higher value should be chosen to ensure it falls on the exponential curve representing the paralyzable model. Method 2: u sing In the paralyzable model graph the maximum (mmax axis. The symbol e in this equation simply represents the exponential of 1 which is calculated a s e1=2.71828. Therefore mmax point indicated on the paralyzable curve and is denoted by ( ncorr, mmax) In Figure 62, the axis (observed rat e) and the x axis (true rate) and this applies for a non paralyzable system. In a non paralyzable system an asymptotic value for the barely has time to finish o ne dead period before starting another. are presented (Table 62). It should be noted that zero. Table 62 shows the calculation to obtain the system dead time and Table 63 gives a summary of the system dead time calculated for Series 2 NaI Foam data. 14

PAGE 118

118 Table 62. Calculation of s ystem dead t ime Method m=ne ln(m max /n corr )/n corr 1/m max 1/[m e max (2.71828)] Summary of Calculate d UF BSX System Dead Time NaI Series A summary of the system dead time for the NaI Series is presented in Table 6 3. Series 2 NaI uses the uniform SOFI block as its target material, while the remaining NaI Series uses the found using the paralyzable equation. A complete compilation of the system dead time for each individual NaI Series is presented in Appendix B. As a reminder, Series 2 NaI Foam used a 2.0 mm beam aperture size and the Series 1 NaI Nylon used a 1.5 mm beam aperture size. It should be noted that the values obtained for the system dead time were approximated to two decimal places. Tab le 6 3. Summary of the UF BSX s ystem dead t NaI s eries Detector 1 (s) Detector 2 (s) Detector 3 (s) Detector 4 (s) Series 2 NaI Foam 5 45x10 07 2 17x10 5 36x10 09 07 2 02x10 5 47x10 09 07 2 .1 6x10 5 37x10 09 07 2 .1 3x10 09 Series 1 NaI Nylon 5.41x10 07 2.39X10 5 36x10 09 07 2.15x10 5 42x10 09 07 2 24x10 5 27x10 09 07 2 .1 5x10 09 YSO Series The system dead time values calculated from the experimental results found Series 3 YSO Nylon may not suitable for this study. In this Series, the YSO detectors did not yet achieve a maximum observed count rate. In order to ensure that a maximum count rate can be established, this problem was addressed in Series 4 YSO Nylon with the use of a larger size aperture for the beam opening. Figure 66 definitely indicated that an observed maximum rate was reached and it also indicates that the system is paralyzable. It is worth mentionin g that the baseline restore

PAGE 119

119 constant for the YSO amplifier was changed for Series 3 and Series 4 relative to Series 2. Table 6eries 2. T his consistency can be attribut ed to the fact that the Series 2 scans were performed with a magnitude (higher than Series 4 YSO Nylon) can be explained by the fact that the system dead time wa s being controlled by the baseline restore time constant. As mentioned before the experimental settings used for Series 4 YSO Nylon was not consistent with the earlier YSO series and therefore experimental results for a fifth series were made. A summary of the system dead time for the YSO Nylon Series is presented in Table 64. Table 6 4. Summary of the UF BSX s ystem dead t YSO s eries Detector 5 Detector 6 Detector 7 Detector 8 Series 2 YSO Nylon 3.79 x10 07 1.21 x10 3.69 x10 09 07 1.26 x10 3.69 x10 09 07 1.12 x10 3.53 x10 09 07 1.10 x10 09 Series 3 YSO Nylon 1.99 x10 07 7.28 x10 2.26 x10 10 07 8.31 x10 1.88 x10 10 07 6.72 x10 2.06 x10 10 07 7.31 x10 Series 4 YSO Nylon 10 1.64 x10 07 6.94 x10 1.72 x10 10 07 7.42 x10 1.74 x10 10 07 6.67 x10 1.95 x10 10 07 7.14 x10 Series 5 YSO Nylon 10 1.75 x10 07 7.08 x10 1.80 x10 10 07 7.00 x10 N/A 10 1.89 x10 07 7.92 x10 10 Series 2 YSO Nylon was under the influence of the baseline restore constant, Series 3 YSO Nylon did not fully achieve a maximum count rate, Series 4 YSO Nylon was obtained using a different voltage of 55 kVp, and Series 5 YSO Nylon was operated usi ng the same 60 kVp as in the other series. Nylon appeared to be much last series, the BSX system was not under the infl uence of the baseline restore constant and the experimental settings were maintained at constant values with the exception of current (mA). Current was the variable in this experiment and as mentioned before it was varied within 2 mA to 45 mA.

PAGE 120

120 (seconds) Now that the approximate magnitude of the UF BSX system dead time was calculated it would be best to use this system dead time value in the paralyzable equation. Such use will help with the comparison of the experimental measured count rate and the calculated measured count considered. For the YSO Series, this reasoning will also be applicable. Since Series 5 YSO Nylon has been refined and shown co Series 1 NaI 2 Nylon The UF BSX system dead time calculated for this particular series was found to be 5.36x10072.15x1009For the Series 1 NaI 2 Nylon 55% of the interpolations had less than 5% error, 18% of the interpolations had errors between 5% and 10%, and 27% of the interpolations had errors higher than 10%. The 27% of cases with errors higher than 10% was found for those cases when the true interaction rates (n) were considerably high. The reason for high errors can be attributed to the scaling f actor (current) used was based on the equation m=n at very low rates. (seconds) paralyzable equation to produce the corresponding recorded (measured) count rate (m). Since a plot of m versus n was already made for this particular series, then the n values found in this experim ent will be used to calculate a value of m and compare this calculated value to the experimental recorded count rate data. Series 5 YSO 8 Nylon The UF BSX system dead time calculated for this particular YSO series was found to be 1.89x10077.92x1010 (seconds). Using the same interpolation method as described for the NaI

PAGE 121

121 Series, 25% of the interpolations had less than 10% error. As a result, such high err ors have fact that the YSO detector is fully capable of recording much higher count rates than NaI and therefore the BSX system barely showed paralyzable behavior within the allowable current (mA) range for the employed x ray generator. Degradation of Uncertainty based on Count Statistics Earlier discussion has identified two models, paralyzable and non paralyzable. These two models predict the same firstorder losses and differ only when true events rates are high. This section tests this theory and explores the magnitude of losses for a varying range of true event rates. The complete behavior of the UF BSX system depends not only on the physical processes taking place in the detector itself, but on the pulse processing and electronics involved. Therefore, it is very important to consider an overall system dead time that includes both the detector and the associated electronics rather than just the dead time associated with the detector alone. For this analysis the NaI crystal dead time used is 2.3x1007 (seconds)14 and the YSO crystal dead time used is 4.00x1008 (seconds) ( YSO properties are shown in Appendix D). With regard to Series 3 YSO Nylon, the deg radation of uncertainty based on count statistics study was not performed because for this particular series it was difficult to calculate the system dead time. This was because the calculation of the system dead time depends on the use of the paralyzable model equation. The Series 3 YSO Nylon data did not achieve a definitive plateau alues found for the other series. Also, since Series 5 YSO Nylon was performed much later in this study, the statistical analysis of this particular series is absent in this section. Figure 6 8 shows the results of the degradation study for the NaI Ser ies and Figure 6 9 pertains to the YSO Series. To

PAGE 122

122 obtain plots shown the Figures 68 and 6 9, random values of n (true rate) were chosen that fell in the range of 18,000 (cps)
PAGE 123

123 BSXm n LossofU 1 1 (6 6) BSXm n ts LossOfCoun (6 7) Detectors crystalm lativeU 1 Re (6 8) Loss of uncertainty based on count statistics: crystal crystalm n LossofU 1 1 (6 9) crystal crystalm n ts LossOfCoun (6 10) Figures 6 8 to 611 indicate the loss of uncertainty and loss of counts for Series 2 NaI Foam and Series 1 NaI Nylon respectively. Figures 6 12to 615 indicate the loss of uncertainty and loss of counts for Se ries 2 YSO Nylon and Series 4 YSO Nylon. All the figures indicating the losses follow the similar trend, a s the true rate increases there is a greater loss of uncertainty and counts. The Figures also indicate that if an experimenter were to base their losses on the dead time constant of the detector crystal alone then, the perceived magnitude of loss is far less as compared to considering the entire system ( association of both detector and electronics) dead time. Therefore, to get a realistic idea of th e amount of losses at high counting rates the system dead time ( should be considered and not just the dead time of the detector crystal.

PAGE 124

124 Figure 6 8. Comparing loss of uncertainty obtained by the detector to the combination of detector and associated electronics : Series 2 NaI f oam. Figure 6 9. Comparing loss of counts obtained by the detector to the combination of detector and associated electronics : Series 2 NaI f oam.

PAGE 125

125 Figure 6 10. Comparing loss of uncertainty obtained by the detector to the combination of detector and associated electronics : Series 1 NaI n ylon. Figure 6 11. Comparing loss of counts obtained by the detector to the combination of detector and associated electronics : Series 1 NaI n ylon. The following figures pertain to th e YSO Series:

PAGE 126

126 Figure 6 12. Comparing loss of uncertainty obtained by the detector to the combination of detector and associated electronics : Series 2 YSO n ylon. Figure 6 13. Comparing loss of counts obtained by the detector to the combination of detector and associated electronics : Series 2 YSO n ylon.

PAGE 127

127 Figure 6 14. Comparing loss of uncertainty obtained by the detector to the combination of detector and associated electronics: Series 4 YSO n ylon. Figure 6 15. Comparing loss of counts obtai ned by the detector to the combination of detector and associated electronics : Series 4 YSO n ylon. Now that the UF BSX System dead time has been quantified and that the system follows a paralyzable model, certain analyses can be performed and the knowled ge of the paralyzable

PAGE 128

128 system can help with the explanation of behavior and anomalies especially at high counting rates. In particular, statistical analysis can be done to show the behavior of pile up at high count rates. This statistical analysis on pile up effects will be presented in Chapter 8.

PAGE 129

129 CHAPTER 7 DEGRADATION OF IMAGE QUALITY Image Quality Purpose This section is an extension of the image contrast study and it utilizes the fact that the UF BSX is a paralyzable system. At Lockheed Martin Space System Co., scans on the SOFI are constantly being performed. Observations, irregularities, and theories are continuously being swapped between Lockheed and the UF BSX research group such that the BSX scanning performance can be optimized. It was suggested by Warren Ussery that the study of image quality at high count rates should be considered. In the image contrast study, the count rates were generally well within the respective limits of both the NaI and YSO detectors, so now it is reasonable to look at image contrast and high count rates and determine the maximum count rate that can be used before image quali ty is degraded. Such determination can in effect eliminate wasted time and money since it does not make sense to exceed certain count rates and obtain poor image quality. Setup As a rule of thumb, based from experience, images typically with a contrast of less than 1% can be deemed as poor quality and that image cannot be used for representation or analysis purposes. The relative contrast in Chapter 4 will again be used to obtain image quality. The absolute value of the image contrast will not be used, but instead, as is done more normally, a negative value (or negative contrast) which shows a void as a dark region and a positive value (or positive contrast) which shows a void as a light region will be employed. A uniform nylon plate with a channel (void) wa s used for this analysis. Nylon is a better suited material for high count rate testing and will be used as the target material. The x ray generator settings were kept

PAGE 130

130 at 60 kVp and a 5.5 mm FOC. The variable, current (mA) was varied since it mimics the ma gnitude of the counting rate. Low current implies low count rates and a high current implies high count rates. A 2.0 mm beam aperture was employed for scanning and scans were performed 2 inches above the surface, first over the nylon plate without any addi tional nylon overlay and then second with a 14 mm nylon overlay on top of the nylon plate. Channel Contrast Analysis NaI Figures 7 1 to 74 show the relative contrast for both raw and processed images without a nylon overlay and Figures 75to 78 shows the relative contrast for the raw and processed images with a nylon overlay. As mentioned before contrast analysis was performed using the X ray Bat software. The processing of the images involved the application of the adaptive equalize to the r aw images. Figure 7 1. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 1 (without overlay)

PAGE 131

131 Figure 7 2. Trend observed for relative contrast as current (mA) increases for both proc essed and unprocessed images NaI detector 2 (without overlay) Figure 7 3. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 3 (without overlay)

PAGE 132

132 Figure 7 4. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 4 (without overlay) Figures 7 5 to 78 indicates that the relative contrast is lower with an overlay. This implies that with increasing thickness the image quality of a void or defect will be more difficult to ascertain. Figure 7 5. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 1 (with overlay).

PAGE 133

133 Figure 7 6. Tr end observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 2 (with overlay). Figure 7 7. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 3 (with overlay).

PAGE 134

134 Figure 7 8. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images NaI detector 4 (with overlay). The presence of an overlay certainly does affect the magnitude of the r elative contrast of a void. Adjustment of collimation might improve the contrast, but when it comes to image quality there is limit with respect to high count rate (current). In Figures 7 1to 78, the relative contrast started to decrease with increasing count rate, but then at even higher count rates it started to increase again, but with a change in sign. This behavior contradicts what was expected. Instead of a gradual degradation of image quality of the channel (void) in the nylon it appears that ther e was degradation then an improvement as the count rates get higher. Figure 79 shows the transition of image quality, without overlay over the current (mA) range from 1(mA) to 45 (mA).

PAGE 135

135 Figure 7 9. Image quality transition for NaI detector 3 (no overla y). Channel Contrast Analysis YSO Similar behavior can be observed for the channel contrast analysis using the YSO detector. The transition of negative contrast to positive contrast was noted and this can be especially observed in YSO detector 8. Figures 7 10to 712 and Figures 713to 715 indicate the relative contrast found for both the raw images and the processed images, without overlay and with overlay respectively. At the time of this channel contrast study YSO detector 7 was not available. F igure 7 10. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 5 (without overlay).

PAGE 136

136 Figure 7 11. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 6 (without overlay). Figure 712. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 8 (without overlay). Figures 7 10 to 712 indicate the values of the relative contrast obtained. In terms of image quality and appearance the YSO detector generally performed better than the NaI detectors, especially at high count rates. This trend can be seen by comparison of the NaI

PAGE 137

137 detectors (without overlay) with t he YSO detectors (without overlay). This fact is further shown when there was the presence of an overlay and channel contrast was still relatively acceptable as opposed to the channel contrast found for the NaI detectors. Figure 7 13. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 5 (with overlay). Figure 7 14. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 6 (with overlay).

PAGE 138

138 Figure 7 15. Trend observed for relative contrast as current (mA) increases for both processed and unprocessed images YSO detector 8 (with overlay). Figures 7 13 to 715 emphasize another important factor that need to be considered whe n it comes to explaining the change from negative contrast to positive contrast. Each individual detector can exhibit very different behavior based on its settings and locations. In this particular instance, YSO detector 8 shows significant changes in appe arance of the image as well as in the sign and magnitude of the contrast. Figure 716 shows the transition of image quality and appearance and the following section provides a hypothesis as to why there is an improvement of image quality at very high count rates. Figure 7 16. Image quality transition for the YSO detector 8 (without overlay).

PAGE 139

139 Unexpected Contrast Transition Hypothesis Before this channel contrast analysis was undertaken there was an expectation that at high count rates there will be a degradation of image quality and contrast. This analysis, has shown that as count rates gradually increases the image quality decreases, but as the count rate continue to increase, for some detectors the image quality then appears to improve. It is hypothe sized that this trend in image quality and contrast can be attributed to the UF BSX system following paralyzable model (refer to Chapter 5). Such behavior is indicated by the trends of the graph, True Interaction Rate (cps) versus Recorded Count Rate (cps ). All values of system dead times for this nylon plate channel contrast analysis are shown in Appendix C. Figure 717 gives an artistic representation of the trend observed if a detection system follows a paralyzable model: Figure 7 17. Representation of a paralyzable model. In Figure 7 17, from point A to B there is a gradual increase in recorded count rate as the true interaction count rate increases. But, from B to C the recorded count rate shows a decrease even though the true interaction count rate is still increasing. Such trend in recorded count rate is

PAGE 140

140 a classic representation of the behavior of a paralyzable counting system and it is as a result of this behavior that there is an observed transition or an inversion of negative contrast to positive contrast. For paralyzable behavior, the observed rate is seen to go through a maximum followed by a definite decrease. Consequently, there are always two possible true interaction rates ( X ) corresponding to a given observed rate.14Comparison For a counting system that is paralyzable there is the situation of low observed rates actually corresponding to very high true rates on the opposite side of the maximum. Because of this nonunique solution, it appears that the UF BSX system sees a lower count rate value and therefore the image quality and contrast value is affected. Not all the detectors were found to transition from negative contrast to positive contrast. A ll the NaI detectors, with the exception of NaI detector 1 displayed the inversion (change) of contrast and only YSO detector 8 showed a similar change. All the detectors that displayed transition behavior had something in common: the fact that the s e detectors (NaI 2, NaI 3, NaI 4, and YSO 8) were able to observe a maximum count rate and then a g radual decrease which is synonymous to paralyzable detector behavior. Figures 718to 721 show the NaI detector behavior and those detectors (NaI 1) that did not observe a definite maximum recordable count rate did not shown a change in contrast sign be cause the x ray generator current limit was already achieved. In Figure 7 18, although NaI detector 1 did achieve a maximum recorded count rate, the current limit on the Yxlon x ray generator (45 mA) was reached and thi s caused further operation of the UF BSX to be stopped. This prevented NaI detector 1 from showing a gradual decrease in recorded count at even higher true interaction rates (high current rates i.e.> 45 mA) and thus prevented the indication of an inversion (change) of contrast from negative contrast to positive contrast.

PAGE 141

141 Figure 7 18. Trend observed in NaI detector 1 that is attributed to the contrast changes at high count rates. In Figure 7 18, although maximum count was observed the current limit of the UF BSX imaging system was achieved first before the graph can indicate a decrease in value. Hence further changes in contrast value were prevented due to the lack of obtaining a non unique recorded count rate value i.e. low observed rate actually corresponding to high true rate on the oppos ite side of the maximum. Figure 7 19. Trend observed in NaI detector 2 that is attributed to the contrast changes at high count rates.

PAGE 142

142 Figure 7 20. Trend observed in NaI detector 3 that is attributed to the contrast changes at high count rates. Figu re 7 21. Trend observed in NaI detector 4 that is attributed to the contrast changes at high count rates. Figures 7 19 to 721 indicate the trend for NaI detectors 2, 3, and 4 respectively. These three detectors have displayed paralyzable model behavior a nd as a result, there were instances over a range of high count rates that nonunique recorded count rate values were observed. Such an interpretation used for the NaI detectors is just as applicable to the YSO detectors, and this can be observed in Figures 7 22to 724.

PAGE 143

143 Figure 7 22. Trend observed in YSO detector 5 that is attributed to the contrast changes at high count rates. Figure 7 23. Trend observed in YSO detector 6 that is attributed to the contrast changes at high count rates.

PAGE 144

144 Figure 724. Trend observed in YSO detector 6 that is attributed to the contrast changes at high count rates. In Figures 7 22to 723, paralyzable model behavior has not been observed. The current (mA) limit was reached before a definite maximum in measured coun ts was observed followed by a definite decrease. Figure 7 24, indicates paralyzable model behavior and because of this there was a change in contrast sign at increasing true count rates. In summary, for a paralyzable model there are always two possible true interaction rates corresponding to a given observed count rate. At high count rates when the count losses are great the calculated true count rate becomes very sensitive to small changes in the measured rate and the assumed system behavior. 14

PAGE 145

145 CHAPTER 8 STATISTICAL ANALYSIS OF PILE UP EVENTS Count Statistics Definition of Terms Throughout this analysis several terms that will be used need to be defined. These terms are taken from the text by Glen F. Knoll14 and are considered to be relevant in this statistical analysis of pile up events. An event is the consequence of a radiation interaction in the detector that should lead to a recorded pulse, in the absence of dead time or pile up. A count is a pulse that is actually registered by the recording system the U F BSX Imaging System. Because, of dead time and/or pile up, fewer counts are recorded than the number of true events.14 This defined as the minimum time that must separate two events so that they do not pile up. Thus events that reach the amplifier are randomly distributed as a Poisson distribution and these events True events are again assumed to occur at a rate n. D ue to pile up, the BSX system will perceive counts at a lower rate, m. Based on the fact that the UF BSX system behaves in a paralyzable manner this can help to deduce these observed counts (m) according to the number of true events that contribute to each count. Paralyzable System 14 The analysis and calculations found in this section were adapted from the Radiation Detection and Measurement textbook by Glenn F. Knoll.14 Most of the equations and graphs follow Knoll s s th e UF BSX system dead time.

PAGE 146

146 Pile Up Free From the Poisson distribution (Equation 81) the probability that exactly x events occur () () !xxxe Px x (8 1) () () !xn ne Px x (8 2) From Equation 72, it was previously shown that for a paralyzable system nmne (8 3) d that is free of pile up. So the probability that a given count is free of pile up in a paralyzable system is shown in Equation 84. (0)nPe (8 4) ithTwo (and only two) events will pile up under paralyzable conditions if the following followed by an event this way can be writte n as: Order PileUp P(i) 0 0(1)(_..0)(())((_..() (1). (1)(1)nn nnPPnoeventtPeventindtPnoeventtt Pendte Pee (8 5)

PAGE 147

147 Consequently the above analysis (Equation 85) can be extended to predict the probability that a count results from the pile up of three true events. In this case, for an initial event t=0, one additional event mu describe a two event pile up. Therefore, 0 ) 0 2(2)(_,,0)(()).(1) (2).(1 (2)(1)nnn nnPPnoeventtPeventindtP Pendtee Pee (8 6) In general, for a given count that is formed from the pile up of (x+1) events the probability (Equation 87 ) is ()(1)nnxPxee (8 7) As further proof to indicate that all counts have indeed been accounted for, the sum of the probabilities should be equal to 1 00()(1) 1n nxnn xxPxeeee (8 8) Each recorded count results from (x+1) tr ue events that come with a probability of P(x). To check the accounting of true events the average number of events per count is therefore, (Equation 89) 0 0 2(1)() (1)(1)x n nx x nnnxxPx xexe xeee (8 9)

PAGE 148

148 The expression shown in Equation 87 is consistent with the re sult for the paralyzable model. From Equation 89 and Equation 63: n nnn xe mne = the average number of events per count Probability Curves Plots of the Probabilities In this section plots of the probabilities that a recorded count is free of pile up [P(0)] or due to ithTable 8 1. Description of probability that a recorded count is caused by pile up. order pile up [P(i)] will be portrayed for each of the series that was considered in the determination of the system dead time. In these plots the x axis is the product of true event rate n and the UF BSX system dead ti Table 8 1 presents the description of probabilities that contribute to a count resulting from (x+1) events. Probability P(x) Description P(0) probability that a given count rate is free of pile up P(1) probability two (and only 2) events will pile up under paralyzable conditions P(2) probability that a count results from the pile up of three true events P(3) probability that a count results from the pile up of four true events NaI Series Probability Plots Series 1 NaI Nylon will be the representative series used to present the probability plots i.e plots describing probabilities that a recorded count is caused by pile up. Since the system dead time that were determined were very close in magnitude for each detector in the NaI series the trend observed in each case and the magnitude of the probabilities will be very similar. Figures 8 1to 8 4 show the probability plots for each NaI detector (1, 2, 3 and 4) in Series 1 NaI Nylon.

PAGE 149

149 Figure 8 1. Probability plot for Series 1 NaI 1 n ylon (1.5 mm beam a perture s ize) Figure 8 2. Probability plot for Series 1 NaI 2 n ylon (1.5 mm beam a perture s ize)

PAGE 150

150 Figure 8 3. Probability plot for Series 1 NaI 3 n ylon (1.5 mm beam a perture s ize) Figure 8 4. Probability plot for Series 1 NaI 4 n ylon (1.5 mm beam a perture s ize).

PAGE 151

151 YSO Series Probability Plots Series 5 YSO Nylon will be the series used to present the probability plots. Since this series represent a culmination of all the fine tuning with regard to experimental setup, resetting this particular series were consistent in magn itude for each detector, but not quite as close as when compared to the NaI series. Nevertheless, the trend of the probability plots will be very similar throughout the YSO series. Figures 85to 87 show the probability plots for each YSO detector (5, 6, and 8) in Series 5 YSO Nylon. Figure 8 5. Probability plot for Series 5 YSO 5 n ylon (4.0 mm beam a perture s ize).

PAGE 152

152 Figure 8 6. Probability plot for Series 5 YSO 6 n ylon (4.0 mm beam a perture s ize). Figure 8 7. Probability plot for Series 5 YS O 8 n ylon (4.0 mm beam a perture s ize). Theoretical Probability Plot s Figures 8 1 to 84 are the experimental probability plots for the NaI Series and Figures 85to 87 are the experimental probability plots for the YSO Series. In all the experimental

PAGE 153

153 probability that a given count is free of pile up decreases. Conseque probability that a given count is formed from the pile up of multiple events [P (1), P (2), P (3)] also increases. Figure 8 8 shows the theoretical probability plots (i.e. the contributions) and their shapes for a paralyzable syst em. Comparing the shapes of the theoretical plot to the experimental plots obtained, the trend observed is similar. Figure 8 8. Paralyzable s ystem t heoretical probability plot It is observed in the probability plots for the NaI and YSO series that t he x axis extends further as compared to Figure 88. In most cases the trend and shape of the graph is important a t and therefore extending the x axis beyond 1.0 in the theoretical plot i s not necessary. Another reason can be attributed t o the fact that a shorter range on the x axis will produce a more pronounced effect and will be able to display the expected behavior a bit more accurately. Sometimes the trend and behavior can be lost i f the range of the x axis is extended. With regard to the statistical analysis of the UF BSX Imaging system, it is the behavior and trend s to be known. Hence in this statistical analysis the x axis in the probability plots were extended. 14

PAGE 154

154 The probabilit It is safe to assume that this probability starts at the value of 1 and decreases steadily as the magnitude of the average number of true events increases. For the probability plot s representing the YSO up under the paralyzable condition appears to play a significant role in the UF BSX Imaging system. This is especially shown as the P (2) and P (3) increases steadily even at low count rates. This fact ties in with the fact that even at low current (e.g. 3 mA) some losses are expected. The probability that a count results from the pile up of only 2 events is obviously prominent a t a lower average number of true events, but shows a gradual decrease as higher count rates are achieved. Eventually, almost all the because P(x>3) is increasing This suggests that the system becomes greatly paralyzed at high count rates and that pileup signific antly overwhelms the BSX system Effect of PileUp on the Fraction of True Events Fraction of True Events that Escape Pile Up In the case of a paralyzable system, Equation 8 9 gives the average number of events per count as = e. The probability (per count) of recording a pile up free event is P (0) = e2(0) |n n epara nPe fe xe Therefore, the fraction of true events that escape pileup is given by Equation 810 (8 10) becomes |12eparafn (8 11) This result can be compared with the 1st order expression for the fraction of counts that are free of pile up (Equation 812 )

PAGE 155

155 n e P fn c 1 ) 0 ( (8 12) Plots of the fraction of true events that escape pile up ( fe|para) and fraction of counts that are free of pile up ( fcFraction of recorded counts that escape pile up = P (0) or f or P (0)) are shown in Figures 8 9 through 8 15: Fraction of true events that escape pileup = ( fc e|paraNaI Series ) Figure 8 9. The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile : Series 1 NaI 1 n ylon (1.5 mm beam aperture size)

PAGE 156

156 Figure 810. The fraction of recorded counts (upper curve) a nd true events (lower curve) that escape pile : Series 1 NaI 2 n ylon (1.5 mm beam aperture size) Figure 8 11. The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile up as a funct Series 1 NaI 3 n ylon (1.5 mm beam aperture size)

PAGE 157

157 Figure 8 12. The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile : Series 1 NaI 4 n ylon (1.5 mm beam aperture size) Y SO Series Figure 8 13. The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile up Series 5 YSO 5 n ylon (4.0 mm beam aperture size).

PAGE 158

158 Figure 8 14. The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile up Series 5 YSO 6 n ylon (4.0 mm beam aperture size). Figure 8 15. The fraction of recorded counts (upper curve) and true events (lower curve) that escape pile up Series 5 YSO 8 n ylon (4.0 mm beam aperture size).

PAGE 159

159 Theoretical Plot for the Fraction of True Event Rate Figure 8 16 shows the complete theoretical description for the fraction of true and recorded counts free of pile up versus number of true event s n F igures 8 9 through 815 only indicate the true events curves for the paralyzable system since the UF BSX system was found to follow a par alyzable model during the initial stages of this study. Figure 8 16. Fraction of recorded counts and true events that escape pile up. Rate of Observed Counts due to Pile Up 14 Based on the results calculated earlier another useful quantity can now be pre dicted rpu the rate at which counts due to pile up are observed. This rate is found by multiplying the overall counting rate (m) by the probability that a given count involves pile up: (1(0))purmP (8 13) Since P (0) is defined as the probability that a given count is free of pile up for the paralyzable model and substituting the equation for P (0), Equation 813 can be expanded into Equation 814:

PAGE 160

160 (0) (1)n n puPe rme (8 14) The expression in Equation 814 can be further examined in the low rate range by considering the approximations found in Equation 812 and also using that same expression to rewrite Equation 63. Thus ) 1 ( 1 n n m and n en Therefore to first order 2) 1 ( n r n n n rpu pu (8 15) Equation 815 is a simplified and widely used result and it shows that, in the limit of small pile up contributions, the rate of counts involving pile up is proportional to the system Consequently, the rate at which true events are recorded free of pileup (rpfr ) is given by the product of the counting rate m and the probability per count of escaping pile up, P (0): pf |para = m fe |para = me(8 16)

PAGE 161

161 2 21 | 1 1 |n pfpara n pfparaSince n m n then n re n n rne (8 17) The plots that indicate the variation between the true events recorded free of pileup (rpf) and the counts due to pile up (rpuNaI s eries ) are shown in Figures 8 17 through 823 : Figure 8 17. Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu NaI 1 n ylon 1.5 mm beam aperture size.

PAGE 162

162 Figure 8 18. Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu NaI 2 n ylon 1.5 mm beam aperture size. Figure 8 19. Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu NaI 3 n ylon 1.5 mm beam aperture size.

PAGE 163

163 Figure 8 20. Comparisons of the recorded true events free of pile up (rpf due to pile up (rpuYSO s eries NaI 4 n ylon 1.5 mm beam aperture size. Figure 8 21. Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu YSO 5 n ylon 4.0 mm beam aperture size.

PAGE 164

164 Figure 8 22. Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu YSO 6 n ylon 4.0 mm beam aperture size. Figure 8 23. Comparisons of the recorded true events free of pile up (rpf due to pile up (rpu YSO 8 n ylon 4.0 mm beam aperture size.

PAGE 165

165 Considering the NaI and YSO Series presented in Figures 817 to 823, over an average of true events the contribution of counts due to pile up increases rapidly, reaches a maximum and then falls to zero. Consequently the number of true events that are free of pile up is minimal compared to rpu and it evens decreases to zero at a very early stage in the overall counting process. Quantitative correction for pile up effects can be implemented as well as the use of active circuitry in the electronics that can discard pulses that are expected to be affected by pile up.

PAGE 166

166 CHAPTER 9 YSO DETECTOR COUNT LIMITATIONS YSO Detector YSO Description The YSO detector used in the UF BSX system is properly referred to as YSO ( Y2SiO5) or Yttrium Orthosilicate. It possesses several properties that make a very good detector candidate for the BSX scanning system. The YSO is rugged, not hydroscopic, faster than NaI, and produces more light output than NaI at X ray backscatter energies of i nterest. The YSO detector has a density of 4.45 g/cm3, a decay time constant of 40 ns a much smaller detection surface area (refer to Table 9 1 ), and a slightly higher quantum efficiency compared to the NaI detector for low energy x rays (10 keV 55 keV) As mentioned earlier, below 50 keV the Y2SiO5Table 9 1. Decay t ime c onstants and detection surface area crystal has a more favorable scattering to absorption ratio than the NaI detector, especially from about 16 keV to 33 keV; the lower the scattering to absorption ratio, the better the detection capabilities. Decay Time Constant Surface Area (cm 2 ) NaI Detector 230 20.3 YSO Detector 40 5.06 Both the NaI and the YSO detector have a design limit (refer to Ch. 6) on the number of counts that can be recorded. Compared to the NaI the YSO detector is capable of detecting a larger number of counts (refer to Table 9 2 ) : Table 9 2. Design c ount r ate l imit (cps) Count Rate Limit (cps) NaI Detector 7.5E+05 YSO Detector 2.00E+06 Even though the YSO detector possessed a high count rate limit it was observed throughout this study that the number of counts being recorded fell far below the calculated limit

PAGE 167

167 of 2 million (cps). The trends observed shows that only a maximum of approximately 1.6 million (cps) were being recorded and there was of interest to determine what was causing the unexpected reduced count limit. Individual Testing of Components Approach Recall the BSX system electronics setup in Figure 5 1. Each component affects the number of t he counts that can be recorded and as a result each component needed to be tested individually to deduce the location and cause of the YSO detector count limitation. The components that were considered included the fast amplifier, YSO detector, Ortec SCA, and the preamplifier. All the specifications for the components that have been identified are included in A ppendix D Fast A mplifier C omponent There are two primary functions that are conventionally provided by the linear amplifier component in the pulse processing chain: pulse shaping and amplitude gain. The linear amplifier accepts tail pulses as an input, and produces a shaped linear pulse with standard polarity and output pulse amplitude range. The amplification factor or gain required varies gre atly with application but is typically a factor between 100 and 5000. From the early beginnings of the UF BSX imaging system, the differential time constant of the fast amplifier for the NaI and YSO detectors was set at the same values. This value was set to be very long and was never changed, even though the scan rates of BSX operation have increased over the years. Therefore the differential time constant was changed to a minimum YSOFASTAMP (i.e. fastest). The idea behind this change was to improve the dead time response by narrowing the pulses with shorter fall times. Experimentally, this did improve the response by increasing the maximum rate from about 1.1 MHz to approximately 1.5 MHz on the YSO

PAGE 168

168 detectors, but clearly even though this change made an i mprovement there was another bottleneck from a limitation elsewhere in the system. Detector C omponent After the testing of a significant component in the BSX equipment setup with minimal improvement, the integrity of the YSO detector itself was now bein g questioned. To eliminate this as a potential source of the problem, the YSO detector was disconnected and was replaced with a plastic scintillator detector. The plastic scintillator detector gave slightly better results (~1.6 MHz) but the expected high c ount rate limit for the plastic scintillator detector (decay time constant of 1.8 ns) was still never achieved (Figure 91to 92) even though the count rate limit for the plastic is higher than the YSO detector. Figure 9 1. Recorded c ount r ate versus True i nteraction r ate: Series 1 Plastic n ylon (4.0 mm beam aperture size).

PAGE 169

169 Figure 9 2. Recorded c ount r ate versus True i nteraction r ate: Series 1 P lastic n ylon (no aperture) The theory behind such a slight improvement to the maximum number of count being achieved (i.e. from 1.5 MHz to 1.6 MHz) can be attributed to the shorter time constant of the plastic detector. The time constant of the plastic detector (1.8 ns) was greatly different from the time constant in the preamp (~0.5 s) and hence the detector did not affect the output pulse width of preamp If two time constants are similar the n there will be effect and cause the pulse fall time to drop faster. Consequently even though the YSO detector time constant (40 ns) is relatively sh ort it is presumed to have an effect with the time constant in the preamp (~0.5 s) and therefore affect the pulse fall time. Despite the maximum count rate of about 1.60E+06 cps with the plastic detector, this was considered to be good sign as far as fi guring out which component is causing the limitation. If the YSO detector was the problem, then it would have been extremely expensive to replace. Since the same effects (i.e. count limitation) were observed with the plastic detector, then the YSO detector s were not at fault and were therefore eliminated from the list of components causing the count limitation.

PAGE 170

170 SCA C omponent At the beginning of the dead time analysis study when the YSO count limitation was first identified it was suggested that the SCA (Ortec 850 QUAD Single Channel Analyzer) might be the cause of the count limitation. Preliminary calculations supported this theory and a design for a new SCA prototype by D. Ekdahl was made with the hope that higher count rates can be recorded. The reason s as to why the SCA might be the limiting component was found based on the magnitude of the output pulse width of the SCA. Measurements were made using the oscilloscope and the Ortec SCA output pulse width was found to be 0.43 s. SCA output pulse width = 0.43 s =0.43x1006According to Ortec SCA Specifications: s SCA pulse pair resolving time = 200 ns + SCA output pulse width = 0.2 s + 0.43 s = 0.63 s = 0.63x1006Therefore: s Maximum Achievable Count Rate =1/SCA pulse pair resolving t ime =1/(0.63E 06)=1587301 cps ~1.6 MHz) Further analysis was performed by measuring the pulse widths of the two electronic components (preamp and fast filter amplifier) before the SCA:

PAGE 171

171 Figure 9 3. Electronic components before the SCA and their corresponding pulse widths Preamp output pulse width = 0.3 s Input pulse width to SCA =0.18 s SCA output pulse width = 0.43 s There appears to be a significant mismatch in the magnitude of the input pul se width to the SCA and the SCA output pulse width (Figure 9 3). This discrepancy along with the fact that the reciprocal of the SCA pulse pair resolving time matches the observed maximum count rate (1.6 MHz) can lead to the hypothesis that the cause of th e YSO count limitation might be the Ortec SCA. Dan Ekdahls new design for the SCA was based on the Dallas Semiconductors DS1023 programmable delay line. This delay line can be used in 2 modes, one strictly as a programmable delay, and a 2nd mode as a pu lse generator. The basis behind this design was to try and create a fast and stable SCA, in effect to reduce the SCA pulse width such that the limitation of 1.6 MHz might be increased. The delay line will record only the leading edge of the signal such tha t there is ample time for the second signal to be recorded. The prototype of the new design is shown in Figure 9 4 and the circuit is shown in Figure 95. The prototype was inserted in the motherboard and it was only placed in the channel 5 configuration. Therefore any testing of this new SCA was monitored by observing the count rate for YSO detector 5.

PAGE 172

172 Figure 9 4. Pictures A and B show the new SCA prototype while pictures C and D show the prototype mounted onto the motherboard Figure 9 5. Circuit of the SCA prototype provided by D. Ekdahl After performing some scans on a uniform nylon block and monitoring the magnitude of the count rates being achieved it was concluded that the new prototype did not help to increase the measured count rate. This does not imply that the prototype should be disregarded, because it can help alleviate pileup effects when the system is being operated at higher counting rates. It was still inconclusive whether or not the SCA was the cause of problem and further analysis st ill

PAGE 173

173 needs to be performed. Future work will also include the analysis and reason behind the pulse width mismatch as indicated in Figure 9 3. Pre A mplifier C omponent The preamplifier (or preamp) serves as an intermediate amplification step. It is the first element in the signal processing chain and it provides an interface between the detector and the rest of the analysis electronics that follow. An electronic pulse generator is typically a common element used in most radiation instrumentation systems for initial setup and calibration of nuclear instrumentation. For convenience, a pulse generator can be used to provide a test pulse input to the preamplifier or used directly in place of the preamplifier output. Since the preamplifier component is being tested the pulse generator replaces the preamplifier (refer to Figure 9.6) Once the 10 MHz pulser replaced the preamplifier, the frequency of the pulser was varied while a scan was performed on the same nylon block to observe the number of recorded counts, as s hown in Figure 97. Figure 9 6. Equipment setup for testing the preamplifier.

PAGE 174

174 Figure 9 7. Frequency (MHz) versus Net c ount r ate (cps) The frequency range that was used for this testing was 1.5
PAGE 175

175 removal of the pulse transformer had no effect in any direction on the maximum count rate. In this particular experiment some considerations needed to be addressed. T he experimenter ha d the luxury of continually reducing the pulse width on the 10 MHz pulse generator. If the pulse width on the preamplifier can be adjusted then perhaps the expected count limit might be achieved. Charge s ensitive p reamplifier In retrospect, the idea of saturation is correct. Instead of considering the preamp transformer being saturated, perhaps the entire charge sensitive preamplifier configuration (Figure 9 8) should be considered as being saturated. Figure 9 8. Simplified circuit of a charge amplifier configuration In such a configuration, the rise time of the output pulse is kept as short as possible. Typically the decay time of the pulse is quite large to ensure full collection of charge from the detectors. I n order to address the saturation issue, several key components in the charge amplifier configuration should be changed. Since the decay time is usually large then action is required to reduce this decay time and in effect reduce the width of the pulse. Therefore,

PAGE 176

176 adjustment to the differential time constant such that the decay time is short can help reduce the saturation issue. Another component that can be considered for changing is the field effect transistor or FET. If the linear range of operation of the FET is increased then that can result in the ability of the preamp to resolve single events at high count rates. One other suggestion that can help address the issue of the YSO count limitations is to completely change the current preamp. The abil ity for the preamp to resolve single events at high rates is definitely hampered. Therefore, it will be desirous to employ a charge amplifier that is capable of resolving periodic charge at a factor of 5 to 10 times faster than the desired rate. In effec t, if the desired rate is approximately 2 MHz, then a chargesensitive preamplifier that is capable of resolving charge pulses of 10 MHz is needed.

PAGE 177

177 CHAPTER 10 SUMMARY, CONCLUSIONS AND FUTURE WORK Summary and Conclusions Contrast Study Pulse Mode versus Current Mode The ease of void detection and image contrast in foam decreases as the thickness of the foam overlay increases. Generally, the detectors that were in integral mode were more efficient when compared to the detectors in count mode e specially when increasing foam overlay was used. This is because the contrast is weighted by both energy and count rate, and higher energy x rays ha ve a higher contrast value. In most cases, when the images were processed they resulted in an image contras t of similar magnitude for either count mode or current mode. In cases where the count rate is required to be high and it is known that pulse pile up occurs, current (or integral) mode can be used. Also, current mode detector operation has shown that the magnitude of the contrast of deep voids is enhanced in SOFI using either NaI or YSO detectors. Knowledge of the UF BSX imaging system dead time is considered to be very important. Determination that the system is paralyzable and understa nding of the physics behind a paralyzable system allows for analysis of pileup and its effect on the system. At high counting rates, the magnitude of the system dead time is indeed crucial since it helps to determine pile up contributions towards the puls e height spectrum and recorded counts being achieved. Even though several series of scans were explored, the UF BSX system did not fail to deliver consistent results especially with regard to the trends observed in the statistical analysis of the pile up. were pretty much the same. There was much greater consistency for the slower NaI detectors

PAGE 178

178 than for the fast YSO detectors. The detectors in the BSX system are not the only cause and contributors to the system dead time. Collectively, the system dead time is based on detector type, detector position with respect to the x ray beam, and the combined effect of the electronics associated with the system. The most significant contribution towards the determination of the system dead time was the determination that the UF BSX system follows a paralyzable model. From there, based on the e determined. As mentioned before the system dead time was found for each series; series that considered different target materials and even beam aperture size. T he results were separated into two groups; corresponding to the two different detector types used. As a result, taking into in the range of 5.27x1007 NaI Series<5.47x1007 s. For the baseline restore constant and after the reset of the baseline restore constant respectively, 3.53x1007 YSO Series<3.79x1007 s and 1.64x1007 YSO Series<1.95x1007It should be first concluded that for the applicable series tested in this study the trends observed in all the statistical plots followed the same shape and behavior. In all the probability plots it is seen that a given count is free of pile up decreases Consequently a given count is formed from the pile up of multiple events (P (1), P (2), P (3)) also increases. The contributions that are formed from the pile up of (x+1) events can significantly shape the pulse hei ght spectrum and as a result, may affect the scanning results when it comes to the ability to distinguish a surface with a flaw and a surface without a flaw. s.

PAGE 179

179 With regard to the comparison of recorded true events free of pile up to the observed counts cause d by pile up, the results conclude that over an average of true events the contribution of counts due to pile up shows prominence even at low counting rates, increases steadily as the ue indicating that the BSX system itself can become incapacitated at extreme high counting rates. Alternatively, recorded true events free of pileup are significant at low count rates, but this quickly dissipates as the average number of true events incre ases. Channel Contrast Analysis At very high count rates the image quality and magnitude of relative contrast of a void in nylon is generally degraded. It was therefore expected that at exceedingly high count rates this degradation would have continued. I n the case of the UF BSX system this did not occur for all detectors. For some detectors, instead of continued image degradation the image quality and magnitude of the relative contrast were enhanced. Analysis has indicated that such behavior can occur when the detection system is known to follow the paralyzable model. Further testing needs to be explored especially using different target materials e.g. SOFI YSO Detector Count Rate Limitations Ever since the evolution of the UF BSX imaging system it was be lieved that the system was able to achieve and record a maximum count rate limit 2.0E+06 (cps) for the YSO detector s In this study, a discrepancy was found since even though the BSX system was allowed to operate under high counting rate conditions, the 2 million count rate limitation was not achievable. A maximum of 1.6 million (cps) was observed and this was reached due to changes made to the preamp and linear amplifier. After testing of each component in the BSX system possible saturation of the preamp was an indicator of the cause of the count limitation It was originally thought that the Ortec SCA was the limiting component of the system. This was further

PAGE 180

180 suspected when the SCA pulse pair resolving time was found to be 0.63 s and the reciprocal of the pulse pair resolving time equates to the maximum number of counts that were being achieved during the course of this study. But after testing of the charge sensitive preamplifier it was found that substituting the preamp with a pulse generator the expected design count limit was achieved. Therefore, changes need to be made to the preamp to minimize the saturation effects. Also, even though the preamp may be the possible problem it would be best to factor in that more than one component can simul taneously affect the count limitation. Future Work Transition Changes in Image Quality Further work needs to be performed in order to explore the inversion of contrast (negative to positive) when very high count rates are achieved. More experiments and fur ther analysis needs to be performed. Especially, testing at high count rates of different target materials, varying flaws and defects should be considered Count Rate Limitation More research, testing, and analyses are needed to help understand and to provi de an explanation for the YSO detector count limitation. Current work should be focused in the preamplifier component and perhaps address the probable saturation issue. Experiments that need to be conducted include changes made to the charge sensitive pr eamplifier configuration and to change some active components (e.g. FET) or reduce the time constant that determines the decay rate of the tail of the output pulse.

PAGE 181

181 APPENDIX A PLOTS OF THE OBSERVED TRENDS FOUND IN THE IMAGE CONTRAST STUDY Series 1 2 inches thick foam overlay Figure A 1. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.5 inches Figure A 2. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.5 inches

PAGE 182

182 Figure A 3. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.5 inches Figure A 4. Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.5 inches

PAGE 183

183 Figure A 5. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.25 inches Figure A 6. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.25 inches

PAGE 184

184 Figure A 7. Series 1: Comparing relative contra st of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.25 inches Figure A 8. Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.25 inches

PAGE 185

185 Figure A 9. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant depth (d) of 0.25 inches Figure A 10. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth (d) of 0.25 inches

PAGE 186

186 Figure A 11. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant depth (d) of 0.25 inches Figure A 12. Series 1: Compa ring relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant depth (d) of 0.25 inches

PAGE 187

187 Figure A 13. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.75 inches Figure A 14. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.75 inches

PAGE 188

188 Figure A 15. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.75 inches Figure A 16. Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.75 inches

PAGE 189

189 F igure A 17. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.5 inches Figure A 18. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for pro cessed images at constant diameter (D) of 0.5 inches

PAGE 190

190 Figure A 19. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.5 inches Figure A 20. Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.5 inches

PAGE 191

191 Figure A 21. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0. 25 inches Figure A 22. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.25 inches

PAGE 192

192 Figure A 23. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.25 inches. Figure A 24. Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.25 inches.

PAGE 193

193 Figure A 25. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant depth ( d) of 0.5 inches. Figure A 26. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant depth (d) of 0.5 inches.

PAGE 194

194 Figure A 27. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.5 inches. Figure A 28. Series 1: Comparing relative contrast of images obtain ed by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.5 inches.

PAGE 195

195 Figure A 29. Series 1: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant depth (d) of 0.25 inches. Figure A 30. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant depth (d) of 0.25 inches.

PAGE 196

196 Figure A 31. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.25 inches. Figure A 32. Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.25 inches.

PAGE 197

197 Figure A 33. Series 1: Comparing relative contrast of imag es obtained by NaI CR and NaI Int. for processed images at increasing depth (d) and diameter (D). Figure A 34. Series 1: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at increasing depth (d) and diameter (D).

PAGE 198

198 Figure A 35. Series 1: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at increasing depth (d) and diameter (D). Figure A 36. Series 1: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at increasing depth (d) and diameter (D).

PAGE 199

199 Series 2 4 inches thick foam overlay Figure A 37. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0. 5 inches. Figure A 38. Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.5 inches.

PAGE 200

200 Figure A 39. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.5 inches. Figure A 40. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.5 inches.

PAGE 201

201 Figure A 41. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant diameter (D) of 0.25 inches. Figure A 42. Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant diameter (D) of 0.25 inches.

PAGE 202

202 Figure A 43. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant diameter (D) of 0.25 inches. Figure A 44. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant diameter (D) of 0.25 inches.

PAGE 203

203 Figure A 45. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at constant depth ( d) of 0.25 inches. Figure A 46. Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at constant depth ( d) of 0.25 inches.

PAGE 204

204 Figure A 47. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at constant depth ( d) of 0.25 inches. Figure A 48. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at constant depth ( d) of 0.25 inches.

PAGE 205

205 Figure A 49. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for unprocessed images at increasing depth (d) and diameter (D). Figure A 50. Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for unprocessed images at increasi ng depth (d) and diameter (D).

PAGE 206

206 Figure A 51. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for unprocessed images at increasing depth (d) and diameter (D). Figure A 52. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for unprocessed images at increasing depth (d) and diameter (D).

PAGE 207

207 Figure A 53. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.75 inches. Figure A 54. Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.75 inches.

PAGE 208

208 Figure A 55. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.75 inches. Figure A 56. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.75 inches.

PAGE 209

209 Figure A 57. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0.5 inches. Figure A 58. Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.5 inches.

PAGE 210

210 Figure A 59. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.5 inches. Figure A 60. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter (D) of 0.5 inches.

PAGE 211

211 Figure A 61. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter (D) of 0. 25 inches. Figure A 62. Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter (D) of 0.25 inches.

PAGE 212

212 Figure A 63. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter (D) of 0.25 inches. Figure A 64. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at cons tant diameter (D) of 0.25 inches.

PAGE 213

213 Figure A 65. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant depth (d) of 0.5 inches. Figure A 66. Series 2: Comparing relative contrast of images obta ined by YSO CR and YSO Int. for processed images at constant depth (d) of 0.5 inches.

PAGE 214

214 Figure A 67. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.5 inches. Figure A 68. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.5 inches.

PAGE 215

215 Figure A 69. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant depth (d) of 0.25 inches. Figure A 70. Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant depth (d) of 0.25 inches.

PAGE 216

216 Figure A 71. Series 2: Comparing relative c ontrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.25 inches. Figure A 72. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.25 inches.

PAGE 217

217 Figure A 73. Series 2: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at increasing depth (d) and increasing diameter (D). Figure A 74. Series 2: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at increasing depth (d) and increasing diameter (D).

PAGE 218

218 Figure A 75. Series 2: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at increasing depth (d) and increasing diameter (D). Figure A 76. Series 2: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at increasing depth (d) and increasing diameter (D).

PAGE 219

219 Series 3 6 inches thick foam overlay Figure A 77. Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for un processed images at constant diameter (D) of 0.5 inches. Figure A 78. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for un processed images at c onstant diameter (D) of 0.5 inches.

PAGE 220

220 Figure A 79. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for un processed images at constant diameter (D) of 0.5 inches. Figure A 80. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for un processed images at constant diameter (D) of 0.5 inches.

PAGE 221

221 Figure A 81. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for un processed images at constant diameter (D) of 0.25 inches. Figure A 82. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for un processed images at constant diameter (D) of 0.25 inches.

PAGE 222

222 Figure A 83. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for un processed images at constant diameter (D) of 0.25 inches. Figure A 84. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for un processed images at constant diameter (D) of 0.25 inches.

PAGE 223

223 Figure A 85. Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for un processed images at constant depth (d) of 0.25 inches. Figure A 86. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for un processed images at constant depth (d) of 0.25 inches.

PAGE 224

224 Figure A 87. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for un processed images at const ant depth (d) of 0.25 inches. Figure A 88. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for un processed images at constant depth (d) of 0.25 inches.

PAGE 225

225 Figure A 89. Series 3: Comparing relative contrast of images obt ained by NaI CR and NaI Int. for processed images at constant diameter ( D ) of 0.75 inches. Figure A 90. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter ( D ) of 0.75 inches.

PAGE 226

226 Figure A 91. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter ( D ) of 0.75 inches. Figure A 92. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter ( D ) of 0.75 inches.

PAGE 227

227 Figure A 93. Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter ( D ) of 0.5 inches. Figure A 94. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter ( D ) of 0.5 inches.

PAGE 228

228 Figure A 95. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter ( D ) o f 0.5 inches. Figure A 96. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter ( D ) of 0.5 inches.

PAGE 229

229 Figure A 97. Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant diameter ( D ) of 0.25 inches. Figure A 98. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant diameter ( D ) of 0.25 inches.

PAGE 230

230 Figure A 99. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant diameter ( D ) of 0.25 inches. Figure A 100. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant diameter ( D ) of 0.25 inches.

PAGE 231

231 Figure A 101. Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant depth (d) of 0.5 inches. Figure A 102. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at constant depth (d) of 0.5 inches.

PAGE 232

232 Figure A 103. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.5 inches. Figure A 104. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.5 inches.

PAGE 233

233 Figure A 105. Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at constant depth (d) of 0.25 inches. Figure A 106. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at consta nt depth (d) of 0.25 inches.

PAGE 234

234 Figure A 107. Series 3: Comparing relative contrast of images obtained by NaI CR and YSO CR for processed images at constant depth (d) of 0.25 inches. Figure A 108. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at constant depth (d) of 0.25 inches.

PAGE 235

235 Figure A 109. Series 3: Comparing relative contrast of images obtained by NaI CR and NaI Int. for processed images at increasing depth (d) and increasing diameter (D). Figure A 110. Series 3: Comparing relative contrast of images obtained by YSO CR and YSO Int. for processed images at increasing depth (d) and increasing diameter (D).

PAGE 236

236 Figure A 111. Series 3: Comparing relative contrast of images obtained by NaI CR a nd YSO CR for processed images at increasing depth (d) and increasing diameter (D). Figure A 112. Series 3: Comparing relative contrast of images obtained by NaI Int. and YSO Int. for processed images at increasing depth (d) and increasing diameter (D).

PAGE 237

237 APPENDIX B UF BSX SYSTEM DEAD TIME CALCULATIONS NaI Series Plots of Recorded Count Rate vs. True Interaction Rate Figure B 1. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: NaI 1 f oam (2.0 mm beam aperture size). Figure B 2. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: NaI 3 f oam (2.0 mm beam aperture size).

PAGE 238

238 Figure B 3. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: NaI 4 f oam (2.0 mm beam aperture size). Figure B 4. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 1: NaI 1 n ylon (1.5 mm beam aperture size).

PAGE 239

239 Figure B 5. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 1: NaI 3 n ylon (1.5 mm beam aperture size). Figure B 6. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 1: NaI 4 n ylon (1.5 mm beam aperture size).

PAGE 240

240 YSO Series Plots of Recorded Count Rate vs. True Interaction Rate Figure B 7. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: YSO 6 n ylon ( 2.0 mm beam aperture size). Figure B 8. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 2: YSO 7 n ylon (2.0 mm beam aperture size).

PAGE 241

241 Figure B 9. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate f or Series 2: YSO 8 n ylon (2.0 mm beam aperture size). Figure B 10. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 4: YSO 5 n ylon (4.0 mm beam aperture size at 55 kVp).

PAGE 242

242 Figure B 11. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 4: YSO 6 n ylon (4.0 mm beam aperture size at 55 kVp). Figure B 12. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 4: YSO 7 n ylon (4.0 mm beam aperture size at 55 kVp).

PAGE 243

243 Figure B 13. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 5: YSO 5 n ylon (4.0 mm beam aperture size at 60 kVp). Figure B 14. Plot of r ecorded c ount r ate vs. t rue i nteraction r ate for Series 5: YSO 6 n ylon (4.0 mm beam aperture size at 60 kVp).

PAGE 244

244 UF BSX System Dead Time NaI Series Table B 1. System d ead t : Series 2 NaI f oam (2.0 mm beam aperture size) Paralyzable Equation NaI 1 5.45x10 07 2.17x10 09 5.4 6 x10 (s) 07 2.1 1 x10 09 (s) NaI 2 5. 36 x10 07 2. 02 x10 09 5. 37 x10 (s) 07 2. 06 x10 09 NaI 3 (s) 5.4 7 x10 07 2.1 6 x10 09 5.4 8 x10 (s) 07 2.1 2 x10 09 NaI 4 (s) 5. 37 x10 07 2.1 3 x10 09 5. 38 x10 (s) 07 2. 0 7x10 09 (s) Table B 2. System d ead t Series 1 NaI n ylon (1.5 mm beam aperture size) Paralyzable Equation NaI 1 5.41x10 07 2.39x10 09 5.48x10 (s) 07 2.12x10 09 (s) NaI 2 5.36x10 07 2.15x10 09 5.37x10 (s) 07 2.06x10 09 NaI 3 (s) 5.42x10 07 2.24x10 09 5.45x10 (s) 07 2.11x10 09 NaI 4 (s) 5.27x10 07 2.15x10 09 5.30x10 (s) 07 2.02x10 09 (s) UF BSX System Dead Time YSO Series Table B 3. System d ead t Series 2 YSO n ylon (2.0 mm beam aperture size) Paralyzable Equation YSO 5 3.79x10 07 1.21x10 09 3.79x10 (s) 07 1.22x10 09 (s) YSO 6 3.69x10 07 1.26x10 09 3.71x10 (s) 07 1.18x10 09 YSO 7 (s) 3.69x10 07 1.12x10 09 3.70x10 (s) 07 1.18x10 09 YSO 8 (s) 3.53x10 07 1.10x10 09 3.53x10 (s) 07 1.10x10 09 (s) Table B 4. System d ead t Series 3 YSO n ylon (2.5 mm beam aperture size, fast filter amplifier peak output voltage set at 7 V and the baseline restore constant reset to its fastest value of 110 ns) Paralyzable Equation YSO 5 1.99x10 07 7.28x10 10 2.46x10 (s) 07 6.37x10 1 0 (s) YSO 6 2.26x10 07 8.31x10 1 0 2.48x10 (s) 07 6.46x10 1 0 YSO 7 (s) 1.88x10 07 6.72x10 1 0 2.40x10 (s) 07 6.13x10 1 0 YSO 8 (s) 2.06x10 07 7.31x10 1 0 2.45x10 (s) 07 6.33x10 1 0 (s) Table B 5 Series 4 YSO n ylon (4.0 mm beam aperture size, fast filter amplifier peak output voltage set at 7 V and the baseline restore constant reset to its fastest value of 110 ns) Paralyzable Equation YSO 5 1. 63 x10 07 6 94 x10 10 2.4 5 x10 (s) 07 6.3 2 x10 1 0 (s) YSO 6 1 72 x10 07 7 42 x10 1 0 2.4 1 x10 (s) 07 6. 17 x10 1 0 YSO 7 (s) 1. 74 x10 07 6. 67 x10 1 0 2. 39 x10 (s) 07 6.1 1 x10 1 0 YSO 8 (s) 1 95 x10 07 7. 14 x10 1 0 2.4 4 x10 (s) 07 6.3 0 x10 1 0 (s)

PAGE 245

245 APPENDIX C DEGRADATION OF IMAGE QUALITY Channel Contrast Analysis System Dead Time for NaI and YSO Detectors Table C 1. UF BSX s ystem dead t ime for NaI d etectors c hannel in n ylon NaI 1 5.152E 072.800E 09 NaI 2 5.440E 072.341E 09 NaI 3 5.148E 072.221E 09 NaI 4 5.047E 072.267 Table C 2. UF BSX s ystem dead t ime for YSO detectors c hannel in n ylon YSO 5 2.50x10 07 8 95x10 09 YSO 6 2.29x10 07 8.16x10 YSO 7 10 Not Available YSO 8 2.07x10 07 7.10x10 10

PAGE 246

246 APPENDIX D ELECTRONICS SPECIFICATIONS YSO PMT Specifications

PAGE 247

247

PAGE 248

248 YSO Decay Time Constant Hitachi Chemical Co. Ltd.

PAGE 249

249 NaI PMT Specifications

PAGE 250

250

PAGE 251

251 Ortec Model 556 High Voltage Power Supply

PAGE 252

252

PAGE 253

253 Ortec Model 850 QUAD SCA

PAGE 254

254

PAGE 255

255 Charge Sensitive Preamplifier (D. Ekdahl) Description The charge preamplifiers used in the backscatter X ray system detectors are comprised of an active semiconductor element called a FET, an operational amplifier, biasing circuits, and a RC feedback network that essentially sets the rise and fall time constants. The simplified diagram shown illustrates how these components are connected and the solid and dashed arrows show the signal paths taken for both the integration and differentiation periods of the signal, respectively. The labels REF are simply voltage biasing used to set the charge preamplifiers operation level within a linear range and the current labeled Id is chosen to optimize the operation of the FET at its lowest noise level. In this diagram, signal current directions are drawn to reflect actual circuit pathways taken in the negative bias detector assemblies used for backscatter research. Selection of the particular components used is critical as noise, speed, sensitivity and dynamic range must all be considered. Operation of the Charge Preampl ifier An X ray event in the detector will produce a proportional current signal in the photomultiplier which will then appear at the charge preamplifier input. The detector used to sense the X ray even will naturally exhibit its own rise and fall time de pending on the detector type. Referring to the diagram, the path the charge event initially takes during the integration period is depicted by the solid arrows. In the absence of any signal the operational amplifier will set a precise voltage level on the input to the FET in order to maintain balance between its + and inputs. When a charge appears at the input to the FET, the FET will cause an unbalanced voltage at the amplifier + input, and the amplifier will rapidly respond with its output to rebalance its inputs again. In the diagram, this response is shown originating from the operational amplifiers

PAGE 256

256 output to charge the capacitor Cf. The speed that this occurs is referred to as the integration tau (time constant), and is the product of the charge pre amps input resistance Rin and Cf. If the operational amplifier and the FET are chosen very carefully, this FET amplifier combination will respond as fast or faster that the input signal, typically 210 nanoseconds from 10% to 90% of the peak value. Input r esistance of the charge preamplifier varies somewhat but is commonly 100 ohms or less. Cf for the backscatter preamplifiers is 2.7 pF (2.7 x 1012At the peak of the X ray event, the charge begins to taper off at a much slower rate as the illumination in the detector falls with e (exp) In practical systems there are multiple time constants associated with scintillation detectors so that the current does not immediately terminate. Additionally, there is a very small dark current in the pmt which is always present, however, the magnitude of these currents are much smaller than the decay of charge in the feedback capacitor, Cf, so the diagram shows the ideal path of the decay current looping through the Rf and Cf network, as denoted by the dashed arrows. As the voltage due to the stored charge begins to decay, this again appears as a signal at the input to the FET causing an unbalance to the amplifier + input, and so the amplifiers output responds once ag ain to rebalance the inputs. This output response is called the differential time and exhibits the time constant of RfCf. ). Special Considerations Overdrive and Pileup Under ideal conditions injecting periodic charge signals by the charge transfer method using a pulse generator, the charge preamplifier will resolve events at slightly better than 2 MHz rate. However, if there exists co incident events, even for a short period of time, the accumulated sum of the charge can cause saturation in the preamplifier. Saturation is caused by overdriving the preamplifier until one or more of the active components (e.g., FET or operational amplifier)

PAGE 257

257 exceed its linear range of operation, usually from a li mit set by biasing or the power supply. This can directly affect the detector deadtime since active components normally require extra time to recover from saturation. For this charge preamplifier design, limits are imposed on the signal in order to resolve the very low energies associated with backscatter X rays. Background events, considerably less than the imposed X ray backscatter rate and therefore have little effect on the detector deadtime. By the same token, pileup events due to coincidence at rates much less than 2 MHz will not adversely affect the detector performance, however, as these co incident events are integrated at higher rates, the ability for the pream plifier to resolve even single events will be severely hampered. Consequently, it may be desirous to employ a charge amplifier that is capable of resolving periodic charge pulses at a factor of 5 to 10 faster than the desired rate.

PAGE 258

258 REFERENCE LIST 1. Chris Meng, Computed i mage backscatter radiography: p roof of principle and initial development, M.S. Thesis, University of Florida (2008) 2. Daniel Shedlock. X ray backscatter imaging for radiography by selective detection and snapshot evolution, de velopment, and optimization, Ph. D. Dissertation, University of Florida (2007) 3. Nissia Sabri, An adapted modulation transfer function for x ray backscatter radiography by selective detection, M.S. Thesis, University of Florida (2007) 4. A. Jacobs, and J. Campbell, Landmine detection by scatter radiation radiography,Scientific and Technical Final Report, Contract DAAK 7086 K 0016, U.S. Army Belvoir Research, Development and Engineering Center, (1987) 5. J. Campbell, and A. Jacobs, Detection of buried land mines by Compton backscatter imaging, Nuclear Science and Engineering, 110 p. 417424 (1992) 6. Y. Watanabe, J. Monroe., S. Keshavmurthy, A. Jacobs, and E. Dugan, C omputational methods for shape restoration of buried objects in Compton backscatter imaging, Nuclear Science and Engineering, 122 p. 5567 (1996) 7. J. Wehlburg, S. Keshavmurthy, E. Dugan, and A. Jacobs, Geometric considerations relating to lateral migration backscatter radiography (LMBR) as applied to the detection of landmines," Proceeding of SPIE, 3079 p. 384393 (1997) 8. Z. Su, J. Howley, J. Jacobs, E. Dugan, and A. Jacobs., The discernibility of landmines using lateral migration radiography, Proceeding of SPIE, 3392 p. 878887 (1998) 9. C. Wells, Z. Su, J. Moore, E. Dugan and A. Jacobs, "Lateral migration radiography measured image signatures for the detection and identification of buried landmines, Proceeding of SPIE, 3710 p. 906916 (1999) 10. C. Wells, Z. Su, A. Allard, S. Salazar, E. Dugan and A. Jacobs, Suitability of simulated landmines for detection measurements using x ray lateral migration radiography, Proceeding of SPIE, 4038 p. 578589 (2000) 11. Z. Su, A. Jacobs, E. Dugan, J. Howley, and J. Jacobs, Lateral migration radiography application to land mine detection, confirmation and classification, Optical Engineering, 39(9) p. 24722479 (2000) 12. E. Dugan, A. Jacobs, Z. Su, L. Houssay, D. Ekdahl and S. Brygoo, Development and field tes ting of a mobile backscatter x ray lateral migration radiography land mine detection system, Proceeding of SPIE, 4742 p. 120131 (2002) 13. D. Shedlock, E. Dugan, A. Jacobs, B. Addicott, and D. Ekdahl, Backscatter Radiography by Selective Detection (RSD), Poster Presentation, University of Florida

PAGE 259

259 14. Glenn F. Knoll, Radiation detection and measurement, 3rd15. Semitracks 2008, Backscatter Imaging, edition, John Wiley & Sons, Inc. (1999) http://www.semitracks.com/reference/FA/die_level/imaging/backscatter/backscatter.htm. May 2008. 16. John R. Lamarsh and Anthony J. Baratta, Introduction to nuclear e ngineering, 3rd17. Steph anie Brygoo, X r ay l ateral m igration r adiography nondestructive f law detection m easurements and s imulation, M.S. Thesis, University of Florida (2002) edition, Prentice Hall (2001) 18. E. Dugan, A. Jacobs, L. Houssay and D. Ekdahl, Detection of f laws and defects using l ateral m igration x ray r adiography ," Proceedings of SPIE 48th19. D. Shedlock, E. Dugan, A. Jacobs, L.Houssay, Preliminary measurements supporting reactor vessel and large component inspection using x ray backscatter radiography by selective detection, Proceedings of 2006 International Congress on advances in nuclear power plants ICAPP, Reno, June, 2006. Annual Meeting, International Symposium On Optical Science and Technology, Penetrating Radiation Systems and Applications V,Vol. 5199, pp 4761, San Diego, August 38, 2003. 20. E. Dugan, A. Jacobs, D. Ekdahl, C. Meng, N. Sabri and D. Shedlock, Research into the f easibility of utilizing various m ini ature diodes and x ray detectors for x ray backscatter, NASA Final Report, Award Number NNL05AF19P, December 1, 2006.

PAGE 260

260 BIOGRAPHICAL SKETCH Kara Beharry was born and raised in Trinidad, an island that is part of the Caribbean. She obtained her first Bachelor of Science in physics earning Summa Cum Laude from South Carolina State University (SCSU) in 2005. Her second Bachelor of Science degree was obtained in nuclear engineering in 2006 at University of Wisconsin Madison. She received her Master of Science in nuclear engineering from the University of Florida in the fall of 2009. In spring 2010 she will be joining the nuclear engineering program at SCSU a s an instructor and researcher.