FUNCTIONAL MAGNETIC RESONANCE IMAGING OF OVERT LANGUAGE PRODUCTION IN APHASIA REHABILITA TION: THE CONTRIBUTION OF THE LANGUAGE NONDOMINANT HEMISPHERE By MEGAN E. GAIEFSKY 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 2003
Copyright 2003 by Megan E. Gaiefsky
To my family, whose sacrifices were freel y given, whose support was constant, and to whom I am blessed to be able to give my love to everyday.
iv ACKNOWLEDGMENTS I would like to thank my parents, sisters, and brother for their unwavering love and support throughout my many pursu its. I thank my mentor, Dr. Anna Moore, for her guidance, encouragement, and investment in this work. I thank th e team of people who helped make this project possible and with whom I have had the pleasure of working every day: Anna Moore, P h.D., Bruce Crosson, Ph.D., Rich ard Briggs, Ph.D., Leslie Gonzalez-Rothi, Ph.D., Keith White, Ph.D ., Kyung Peck, Ph.D., Kyundah Gopinath, M.S., Christina Wierenga, M.S., Katie Ri chards, M.S., David Soltysik, M.S., Diane Kendall, Ph.D., Christy Milsted, Didem Go ckay, Ph.D., M. Allison Cato, Ph.D., Tim Conway, M.S., Mary Watson and Rachel May.
v TABLE OF CONTENTS Page ACKNOWLEDGMENTS .................................................................................................iv LIST OF TABLES ............................................................................................................vii LIST OF FIGURES .........................................................................................................viii ABSTRACT Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…Â…viii CHAPTER 1 INTRODUCTION ........................................................................................................1 Anatomy of Language ..................................................................................................1 Aphasia .........................................................................................................................4 Functional Magnetic Resonance Imaging ....................................................................6 FMRI and Aphasia ........................................................................................................9 Hypotheses ..................................................................................................................12 2 METHODS .................................................................................................................14 Participants .................................................................................................................14 Procedure ....................................................................................................................16 Experimental Language Rehabilitation ...............................................................16 Imaging Procedure ...............................................................................................19 Imaging Analyses ................................................................................................21 3 RESULTS ...................................................................................................................26 4 DISCUSSION .............................................................................................................29 Increased Right Hemisphere Activi ty from Preto Posttreatment .............................30 Decreased Right Hemisphere Activi ty from Preto Posttreatment ............................30 No Change in Right Hemisphere Ac tivity from Preto Posttreatment ......................31 Conclusions .................................................................................................................31 Limitations ..................................................................................................................32 Implications and Future Directions ............................................................................32
vi REFERENCE LIST ...........................................................................................................35 BIOGRAPHICAL SKETCH .............................................................................................39
vii LIST OF TABLES Table page 1 Functional Activity and Correct Responses for Preand Posttreatment Scans........26 2 Independent Rater and C-Statistic Determin ations of Stable Baseline and Positive Treatment Response.................................................................................................27 3 Summary of Patterns in Functional Activ ation from Preto Posttreatment............28
viii LIST OF FIGURES Figure page 1 Anatomical landmarks for pars triangular is and pars opercularis without and with a diagnol sulcus (AHR: anterior horizontal ramus, AAR: anterior ascending ramus, DS: diagnol sulcus, PTR: pars tria ngularis, POP: pars opercularis)..........................3 2 Example of hemodynamic response..........................................................................8 3 Axial anatomic slices illustrati ng individual lesions for each subject......................24
ix Abstract of Thesis Presented to the Graduate School of th e University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science FUNCTIONAL MAGNETIC RESONANCE IMAGING OF OVERT LANGUAGE PRODUCTION IN APHASIA REHABILITA TION: THE CONTRIBUTION OF THE LANGUAGE NONDOMINANT HEMISPHERE By Megan E. Gaiefsky May 2003 Chair: Anna B. Moore, Ph.D. Major Department: Clini cal and Health Psychology Neural activity during language production in aphasia typically has been measured using functional magnetic resona nce imaging (fMRI) during silent production tasks. There are many potential disadvantages associated with the use of silent production tasks, including an inability to separate co rrect versus incorrect responses. We have modified a technique used in neurologically normal subjects that allows recording and analysis of overt response generation during fMRI. Six aphasic patients completed an event-related word generation (verbal fluency) task during fMRI at two time periods. The patients completed the first fMRI scan pr ior to beginning an experimental language rehabilitation treatment designed to improve language production th rough recruitment of right hemisphere mechanisms. The second fMRI scan was completed following completion of treatment, approximately 710 weeks after the initial scan. A tracing technique was used to define a region of in terest within the right hemisphere homologue of BrocaÂ’s area in order to precisely quantify the amount of neural ac tivity within this
x discrete region of the brain. Th is technique allowed pretreat ment and posttreatment scans to be compared to examine predicted changes in neural activity in the right hemisphere homologue of BrocaÂ’s area. Re sults suggest that for some nonfluent aphasic patients, a positive response to language rehabilitation was associated with a reorganization of function to the right hemisphere homologue of BrocaÂ’s area. Subsequently, implications for neural substrates of recove ry from aphasia are discussed.
1 CHAPTER 1 INTRODUCTION The advent of functional magnetic resonance imaging (fMRI) offers additional information about localization of brain f unction and provides a unique opportunity to examine the areas of the brain that are still functional following brain damage. In addition, fMRI allows a more comprehensive ex amination of the impact of a brain lesion and an analysis of functional perilesional ar eas. While language has not been the focus of the majority of fMRI studies, its role in th e localization of function debate has been both profound and persistent, last ing well over a century. Anatomy of Language While one of the first functions to be ascribed to a specific location, languagerelated functions dominated literatu re in the latter part of the 19th century, particularly due to the work of Paul Broca and Carl Wernicke. It was during this time that the Â“classical modelÂ” of language was deve loped and popularized, a model still commonly used in current research (Binder et al ., 1997; Broca, 1861/1997; Lichtime, 1885). This model proposes that there are two primary local es for language: 1) BrocaÂ’s area, a frontal and expressive language area for planning, speech production, and writing movements, and 2) WernickeÂ’s area, a posterior a nd receptive language area for analysis and identification of the components of language (f or review see Binder et al., 1997; Broca, 1861/1997; Finger, 1994/1997; Li chtime, 1885). More recent research continues to
2 provide support for the classical model via neuroimaging by showing localized activation in the inferior frontal gyrus (IFG; BrocaÂ’s area ), and in the superior temporal gyrus (STG; WernickeÂ’s area) (Binder et al., 1997; F oundas, Eure, Luevano, & Weinberger, 1998). Furthermore, additional studies have begun to add to the list of language-related functions that BrocaÂ’s area appears to be re sponsible for, namely syntactic processing, and feedback/reanalysis (Sakai, Hashimot o, & Homae, 2001). It must also be noted, however, that although language is predominantl y associated with th ese areas, languagerelated functions are not excl usively localized to these ar eas (Foundas et al., 1998). Of great importance to language res earch are consistent findings in right-handed people that language is predominantly a le ft hemisphere phenomenon. The demonstrated anatomical asymmetries in frontal and te mporal speech-language regions appear to consistently favor the left hemisphere across multiple measurement techniques including fMRI, magnetic resonance imaging (MRI), and positron emission tomography (PET) (Burton, Small, & Blumstein, 2000; Foundas et al., 1998; Holla nd et al., 2001; Musso et al., 1999). Recent research also indicates a developmental component to language lateralization, as left hemisphe re specialization (noted by ac tivation) for language appears to increase with age throughout childhood, while the amount of area acti vated in the right hemisphere during language tasks appears to be strongly negatively correlated with age (Holland et al., 2001). Subsequently, resear ch literature rese rves Â“BrocaÂ’sÂ” and Â“WernickeÂ’sÂ” areas only for the left hemisphe re, while the homotopic areas in the right hemisphere are referred to as Â“BrocaÂ’s homologueÂ” and Â“WernickeÂ’s homologue.Â” While not discounting the importan ce of WernickeÂ’s area as the receptive language locale, the focus of many recent studi es has been BrocaÂ’s area, its role in
3 speech-language production, and its right hemisphere homologue. BrocaÂ’s area is comprised of two anatomical structures: pars triangularis (PTR; BrodmannÂ’s area 45) and pars opercularis (POP; BrodmannÂ’s area 44) (Crosson et al., 2001; Foundas et al., 1998; Moore, Crosson, Gockay, Leonard, & Foundas, 2000; Moore, Loftis et al., 2001). Both regions are found on the lateral extent of th e left hemisphere and are defined by rami extending from the Sylvian Fissure. Specifically , according to Foundas et al. (1998), pars triangularis may be defined as Â“extending superiorly to the infe rior frontal sulcus, inferior to the anterior horizontal ramus (AHR), a nd caudally to the anterior ascending ramus (AAR).Â” Pars opercularis is adjacent to pars triangularis, and is Â“bounded superiorly by the inferior frontal sulcusÂ” and Â“bounded inferiorly by the anterior ascending ramus (AAR) and caudally by the precentral sulcusÂ” (Foundas et al., 1998). Without DS With DS Figure 1 . Anatomical landmarks for pars triangu laris and pars opercularis without and with a diagnol sulcus (AHR: anteri or horizontal ramus, AAR: anterior ascending ramus, DS: diagnol sulcus, PT R: pars triangularis, POP: pars opercularis). While the lateral-frontal region, known as BrocaÂ’s area, is predominantly associated with speech production, normal la nguage production appear s to involve both medialand lateral-frontal cortices (Abdullaev & Posner, 1998; Crosson, Rao et al., 1999; Crosson, Sadek et al., 1999). It has been sugge sted that the intentional, or initiation, component for language is primarily localized in the medial-frontal cortex, while the Sylvian Fissure
4 production component is thought to be primarily localized in the lateral-frontal cortex (Abdullaev & Posner, 1998; Crosson, Rao et al., 1999; Crosson, Sa dek, Bobholz et al., 1999; Crosson, Sadek, Maron et al., 2001). There also ap pears to be a temporal relationship between the two cortices, as not ed by event-related potential (ERP) studies, with the medial-frontal areas being activated prior to the la teral-frontal ar eas (Abdullaev & Posner, 1998). In summary, it appears that normal fluent language production requires synchrony of function between left hemisphere medialand lateral-frontal brain regions. Aphasia Aphasia is an acquired disorder re sulting from brain dysfunction in which one or more components of language (e.g. co mprehension, production, structure, or communicative intention) are disrupted (B roca, 1861/1997; Finger, 1994/1997; Nadeau, Rothi, & Crosson, 2000). Aphasia can follow st roke and traumatic brain injury, and may also be associated with diseases affecti ng brain substance and f unction (Nadeau et al., 2000). It is a persistent and e nduring condition affecting langua ge structures in the left hemisphere (Broca, 1861/1997; Finger, 1994/1997; Nadeau et al., 2000). Incidence rates approximate stroke occurrence at 300-500 people per every 100,000, and it is the leading health care problem in the U.S. that requires rehabilitative services (Nadeau et al., 2000). Twenty-five percent of stroke patients also suffer from an a ssociated aphasia (Nadeau et al., 2000). Nonfluent aphasia is a particular t ype of aphasia that is characterized by a disruption in language production. Language tends to be halting, agrammatic, and telegraphic. One type of nonflu ent aphasia is referred to as BrocaÂ’s aphasia given its association with dysfunction in BrocaÂ’s area. Following th e localization of language production dysfunction to BrocaÂ’s area, research efforts began to focus on rehabilitation
5 and recovery of language function in nonfluen t aphasics. Subsequently, debate ensued between proponents of Â“recove ry of functionÂ” and proponent s of Â“reorganization of functionÂ” (Cappa, 2000; Khatri & Hier, 1999; Musso et al., 1999; Rosen et al., 2000). Proponents of recovery of functi on believe that areas initially damaged by brain injury in the left hemisphere heal and subsequently reclaim their duties (Khatri & Hier, 1999; Rosen et al., 2000). In contrast, proponents of reorganization of func tion believe that the areas in the right hemisphere, homologous to the injured areas in the left hemisphere, assume the duties of the injured areas that may no longer be activated, or that may be unable to sustain activation (Cappa, 2000; Kh atri & Hier, 1999; Rosen et al., 2000). Work by Musso et al. (1999) provid es evidence that the right hemisphere homologue of WernickeÂ’s area assists in compensating for the loss of function following injury in WernickeÂ’s area, suggesting a reor ganization of receptive language to the right hemisphere following left hemisphere injury. Mu sso et al. (1999) further suggests that the reorganization of language to the right hemi sphere, which contribu tes to recovery of language in aphasic patients, results from a re-coordination of a network of areas, suggesting a concerted effort of support from many areas fo r more proficient language recovery. However, Thomas, Altenmuller, Marckmann, Kahrs, and Dichgans (1997) measured DC-potentials in nonfluent and fluent aphasics (characteri zed by dysfunction in BrocaÂ’s and WernickeÂ’s areas respectively) and found that reorganization of language function depended upon the type of aphasia. Their findings sugge st that nonfluent aphasics initially (2-4 weeks post injury) de monstrate a reorganizati on of function to the right hemisphere following injury to BrocaÂ’ s area, but follow-up studies (conducted when recovery was clinically evident) suggest a r ecovery of function, as noted by a complete
6 shift of laterality back towa rds the left hemisphere (Thomas et al., 1997). However, this study also suggests that fluent aphasics demonstrated an in itial (2-4 weeks post injury) reorganization of function to the right hemisphe re, that was maintained at follow-up, with no evidence of a shift of laterality back to wards the left hemisphere (Thomas et al., 1997). A third set of findings indicate that reco very of function by the left hemisphere, or reorganization of function to the right hemis phere, may be mediated by the severity of damage to the left hemisphere (Karbe et al ., 1998). The findings presented in Karbe et al. (1998) and Belin et al. (1996) suggest an inverse correla tion between the functional recovery of the tradit ional left hemisphere language ar eas, and perilesional areas, and right hemisphere reorganization. This furthe r suggests that rec overy of language production competency in the left hemisphere reduces the permanent compensatory functioning of the right hemis phere; moreover, more permanen t loss of functioning in the left hemisphere suggests increased compensa tory functioning in the right hemisphere (Karbe et al., 1998). In summary, previous st udies suggest evidence fo r both recovery of function in the left hemisphere and reorganizat ion of function to the right hemisphere, but that effort in one hemisphere appears to be inversely proportionate to functioning in the other. Functional Magnetic Resonance Imaging The rationale behind the use of f unctional neuroimaging resides in the specific information processing demands of the brain during task performance (Fiez, 2001). As the demands of the task are met, changes in neural activity occur in areas of the brain associated with task completion (Fiez, 2001). It has been suggested that imaging methods based on blood flow are more sensitive to ch anges in neuronal activity than other imaging methods, such as those that use cerebral me tabolism as the basis for measuring neuronal
7 activation (e.g. PET and SPECT, Nadeau & Crosson, 1995). Consequently, functional magnetic resonance imaging (fMRI) has become a popular choice among imaging techniques. FMRI is a non-i nvasive technique that measur es neural activity via the properties of the hydrogen atom and hemoglobin; the distribution of the magnetic signal is related to the distribution of water in th e brain tissue, as well as to the amount of oxygen being carried by the hemoglobin (Fiez, 2001). When blood flow to a specific area of the brain increases to meet the dema nds of a task, the number of oxygen-carrying hemoglobin molecules increases; likewis e, the percentage of deoxyhemoglobin (hemoglobin not carrying oxygen) in these ar eas decreases (Fiez, 2001; Rijntjes & Weiller, 2002). Blood oxygen level dependent (B OLD) fMRI images are created via the disruption in the strength of the hydrogen atom sÂ’ magnetic signal caused by the decrease in the percentage of deoxyhemoglobin that s ubsequently causes an increase in the local signal of the hydrogen atoms in the brain ti ssue (Fiez, 2001). Thus, fMRI images are created by the changing in blood flow to areas of the brain mediated by the changing percentages of deoxyhemoglobin in the brain tissue (Fiez, 20 01). Once functional neural activity has been measured, the functional images must be corr elated to anatomic images and registered in order to link the observed ch anges in blood flow to a specific area of the brain (Nadeau & Crosson, 1995). Blood flow changes during fMRI are known as hemodynamic responses (Figure 2) and they accompany neuronal changes. Once a hemodynamic response is initiated (typically 2-4 seconds following trial initiation) , it will typically peak 6-8 seconds after initiation and will resolve, or return to baseli ne levels, at 10-12 seconds after initiation (Fiez, 2001).
8 Time (seconds) Figure 2. Example of hemodynamic response In order to capture the hemodynamic response, two types of behavioral task designs may be used: a blocked design or an eventrelated design. A blocked design attempts to capture a stronger hemodynamic response by a lternating between Â“activeÂ” blocks and Â“controlÂ” blocks (Fiez, 2001). Â“ActiveÂ” blocks contain a number of stimuli with short interstimulus intervals, and Â“c ontrolÂ” blocks contain severa l control stimuli with short interstimulus intervals, or a lengthy period of rest. The rationale for using a blocked design is to provoke multiple hemodynami c responses that will produce stronger activation because there is no time for the hemodynamic response to return back to baseline before the presentation of the next stimulus, thereby causing an additive effect in the amount of measured response. In contrast, an event-related design attempts to capture a single hemodynamic response by alternating between a single Â“activeÂ” stimulus and a single Â“controlÂ” stimulus, or rest period, se parated by a lengthy interstimulus interval. The rationale for using an event-related desi gn is so that a single hemodynamic response 12345678910 R e s p o n s e S t r e n g t h
9 is initiated, peaks, and returns to baseline be fore the next Â“activeÂ” stimulus is presented. Event-related designs appear to maintain si gnificant advantages ove r blocked designs due to increased flexibility in design and anal ysis. For example, event-related designs provide the opportunity to anal yze response patterns via ove rt responses, examine task accuracy, and changes in behavior over time; advantages that cannot be obtained with block designed paradigms. FMRI and Aphasia The advent of fMRI marked a shift in the focus of brain injury research from an examination of which brain areas were injure d, to an examination of which brain areas were still functional (Khatri & Hier, 2000). FMRI surfaced amidst the recovery of function versus reorganization of function de bate. This imaging technique can answer many questions regarding recove ry of function that tradi tional methods cannot answer. Specifically related to aphasia, fMRI may help determine whether recovery of linguistic ability following a sustained period of aphasia is attributable to a recovery of function in the left hemisphere (Khatri & Hier, 2 000; Pizzamiglio, Galati, & Committeri, 2001; Rosen et al., 2000), a reorgani zation of function to the ri ght hemisphere (Cappa, 2000; Khatri & Hier, 2000; Pizzamiglio et al., 2001; Rosen et al., 2000), or whether linguistic recovery results from the coordination of efforts in both hemispheres. A number of fMRI research efforts have been pursued in attempts to end the debate between recovery of function and reor ganization of function, but results indicate support for both sides depending upon various mediating factor s. For example, one study found that left-sided brain damage acquire d in early childhood appears to induce a significant shift of activation associated with language pr oduction to the undamaged right hemisphere (Staudt et al., 2001). Another study found that in patients with language
10 deficits resulting from a left hemisphere in farct, there was a significant increase in the amount of activation in the right hemisphere areas homologous to th e language areas of the left hemisphere, however the decrease in the amount of activati on in the traditional left hemisphere language areas was not signifi cant, but activity was sh ifted away from the lesion site (Cao, Vikingstad, George, Johns on, & Welch, 1999). This rearrangement of activity to other areas of the le ft hemisphere that were anteri or and posterior to the lesion site also appeared to be associated with d ecreased activity in other intact regions of the left hemisphere (Cao et al., 1999). (This phenomenon of intrahemispheric rearrangement of activity to perilesional ar eas anatomically connected to the lesion site is known as diaschisis (Pizzamiglio et al., 2001; Rijntjes & Weiller, 2002)). In contrast, Gold and Kertesz (2000) and Rosen et al. (2000), f ound strong right hemisphere activation in the right inferior frontal gyrus (B rocaÂ’s homologue), with little activity detected in or near the damaged left inferior frontal gyrus (BrocaÂ’s area) in aphasic patients. Previous fMRI studies with aphasic patients have also used covert, or silent, response generation to measure neural activat ion (Rosen et al., 2000). Few studies have used an overt, or spoken, word generation t echnique with fMRI in healthy subjects (de Zubicaray, Wilaon, McMahon, & Muthiah, 2001) , and only one study has used this technique with nonfluent aphasic patients (Miura et al., 1999 ). Covert response generation methodology is problematic since th ere is no controlling for task adherence and completion. Thus, analyses and subsequent results may be corrupted due to subject unresponsiveness, or error, in spite of we ll-controlled debriefing processes. Covert language studies also assume that subvocal articulation will activate the same areas with the same intensity as over t vocalization, but findings comparing overt and covert
11 paradigms suggest that subvocal and overt articulations evoke different hemodynamic response patterns (de Zubicaray et al., 2001; Rosen et al., 2000). Also, since speech production disruption is the essence of nonfluent aphasia, covert responding is not a true measurement of the associated neural disr uption. Further, it has been reported that aphasic patients have complained that th e words they think and intend to say do not necessarily correspond with what th ey produce (Huang, Carr, & Cao, 2001). Previous fMRI studies with aphasic patients have used both blocked and eventrelated designs that have revealed notewor thy findings, but these designs have been associated with a covert response paradigm . There are many advantages to using an event-related paradigm requiring overt re sponses during fMRI (Burton et al., 2001; Caplan & Dapretto, 2001). Most notably is th e ability to determine whether subjects are responding appropriately to the presented stimuli. It has been suggested that a deficit in intentional processes can be better disti nguished by evaluating responses according to their correctness. As noted earlier, inte ntional processes underlie speech production by coordinating internal speech with motor m ovement initiations that facilitate speech production. To further assess neural activ ation, response based activation may be compared according to type of response (e.g., correct and incorrect ). Responses may be separated and coded as correct and incorrect/other responses, thus allowing the activity levels of each response set to be measured and subsequently compared. Response times may then be calculated and activation associ ated with actual speech production can be measured. The reason for the dearth of studies requiring overt responses is primarily due to movement during scanning solicited by overt responding. This movement potentially introduces motion artifacts into functional scan s that can corrupt the images. The present
12 study controls for motion artifact by precisely recording the times at which participants initiate verbal responses, and excluding images affected by motion from the modeled hemodynamic response. This technique exclude s the corrupted images thereby preserving the true hemodynamic responses associated w ith overt responding (the excluded images now become part of the error term). Thus , by using a more controlled methodology, the present study will provide a better measure of neural activity associated with language production, as measured by fMRI. Hypotheses The present study proposes that if pa tients show significant change in language abilities following treatment, then a change in neural activation dur ing an fMRI language production task will be observed. Specifically, these changes in neural activity will be evident in medial-frontal (e.g. pre-supplemen tary motor area) and lateral-frontal (e.g. BrocaÂ’s area and right hemisphere homologue to BrocaÂ’s area) region s. Changes in these areas are believed to be associated with im proved language function. In this first analysis of these data, I will present findings from only the right hemisphere homologue of BrocaÂ’s area. There are three potential mani festations of observable change: 1) an increase in functional activity from pretreat ment to posttreatment scans indicative of a reorganization of function to the right he misphere homologue of BrocaÂ’s area, 2) a decrease in functional activity from pretreat ment to posttreatment scans suggesting initial reorganization of function to the right hemisphe re and then possible lateral shift back to the left hemisphere (further analyses of perile sional areas in the left hemisphere should be examined for a potential recove ry of function by these areas), 3) no right hemisphere functional activity, or no change in right he misphere activation sugge sting that activity
13 associated with language production did not re organize to the right hemisphere, or that activity in the ri ght hemisphere homologue of BrocaÂ’s area was stable.
14 CHAPTER 2 METHODS Participants Nine nonfluent aphasic patients (4 men, 5 women) with middle cerebral artery vascular accidents participated in this study. All participants fulfilled the following inclusion criteria: documented left hemisphe re middle cerebral artery (MCA) cerebral vascular accident (CVA), demonstrated nonf luent aphasia, premorbidly right-handed, native English speaker, height a nd weight compatible with th e scanning environment, and concurrent enrollment in the Intention trea tment for nonfluent language production, explained below. Participants were excluded if they met a ny of the following exclusion criteria: claustrophobia, metal in the body (e.g. pins, plates, rods, screws, non-removable dental work, non-removable body piercings, shrapnel, cardiac pacemaker or other implanted device), internal birth control devi ce, tattooed eyeliner, artificial limb/joint, metallic fixation device, premorbid diagnosis of learning disability, diagnosis of psychiatric disorder or psychiatric hospitaliz ation, prior treatment for alcohol or drug abuse, and pregnancy or possible pregnancy. Additional exclusion criteria included unconsciousness > 5 minutes, seizures or fainti ng spells not associated with stroke, and neurological disorders other than stroke. Nine patients were enrolled in both the treatment and imaging components of the study. However, one patient was unable to complete the imaging component due to discomfort duri ng scanning, and excessive movements during
15 scanning corrupted two patientsÂ’ images. Also, during data analyses, a fourth patientÂ’s pretreatment images could not be analyzed and subsequent comparisons to posttreatment images could not be made. Subsequently, pre and posttreatment comparisons could be made for only five patients (3 men, 2 women; mean age = 59.8 years, SD = 11.76 years, range 46-70 years; mean number of mont hs post-stroke = 60.8 months, SD = 62.06 months, range = 8-160 months). Subject one (S01) was a 46-year-old male who suffered a left middle cerebral artery stroke that affected the temporal, frontal, and parietal lobes extending into subcortical areas. S01 was enrolled in this study at 48 months post-stroke. Subject two (S02) was a 48-year-old female who suffered a left middle cerebral artery stroke affecting the temporal, frontal a nd parietal lobes with some subcortical area extension. S02 was enrolled in this study 8 months post-stroke. Subject three (S03) was a 67-year-old male who suffered a left middle cerebral artery stroke with damage to the frontal and temporal lobes, as well as insula involvement. S03 was enrolled in this study 160 months post-stroke. Subject four (S04) was a 70-year-old male who suffered a left middle cerebral artery stroke involving the temporal lobe and possible fr ontal operculum involvement. In addition, the posterior half of the insula, an terior parietal lobe and lenticulostriate endzone appear to have been affected. S 04 was enrolled in this study 76 months poststroke. Subject five (S05) was a 68-year-old female who suffered a left middle cerebral artery stroke affecting the fr ontal, temporal, and parietal lobes with some subcortical involvement. S05 was enrolled in this study 12 months post-stroke.
16 Procedure Experimental Language Rehabilitation All patients were enrolled in a language rehabilitati on treatment. Although the purpose of this study is not di rectly related to the langua ge treatment study, a brief explanation of the tr eatment study is warrant ed as it sets the c ontext for the imaging study. The language treatment was designed to ta rget the medial-front al cortex with the goal of transferring function from left hemisphe re medial cortex to the right hemisphere homologue. It has been suggested that the ar ea of the brain primarily responsible for the intentional component of language is the left medial-frontal cortex, whereas left lateralfrontal areas are believed to be responsibl e for speech production. The middle cerebral artery (MCA) innervates lateral-frontal cortices, thus a left MCA infarct disrupts the normal functioning of the left lateral-frontal ar eas, but not the left me dial-frontal regions. While left medial-frontal regions remain in tact, the normal interface of the left medialfrontal and left lateral-frontal regions during language produ ction may be disrupted as a result of a left MCA infarct. Subsequently, nonfluent aphasia is believed to result from damage to left lateral-frontal areas associated with left middle cerebral artery vascular accidents. Attempts at language production are believed to engage the intact left medial regions that then attempt to act in concert wi th left lateral regions. Left lateral regions, however, have been rendered dysfunctional by the stroke. It is im portant to note that while left lateral areas have been rendered dysfunctional by the stroke in these five patients, right medial and right lateral regions are still intact. Given this disruption in the left hemisphere, the target of the present tr eatment is the right me dial-frontal cortex. Medial-frontal cortex was chosen as the targ et area based upon prev ious research which indicates that right hemisphere lateral regi ons begin to show so me neural activation
17 during language production in patients with nonfluent aphasia (Belin et al., 1996). As noted above, it has been suggested that activation of medial regions temporally precedes activation of lateral regions. Treatment incl uded an object-naming task that was paired with a non-meaningful movement produced by the left hand. By pairing a naming task, requiring an overt response, with a left-hande d gesture, it is believed that the right hemisphere medial-frontal mechanisms ar e more easily engaged, thus helping to stimulate the right lateral-frontal area to take up the functions of the damaged left lateralfrontal area, namely that of speech production. The Intention treatment (described in further detail in Richards, Singletary, Rothi, Koehler, & Crosson, 2002) was divide d into three phases. All three phases of treatment were designed to target right hemisphere intentional mechanisms via complex movements/gestures generated by the subjec tÂ’s left hand. Progressi on from phase one to phase two, and then to phase three, is marked by transition from an internally generated movement prompted by external cues (e.g. a to ne and flashing star), to an uncued, selfinitiated complex movement sequence. By using a progressive two-cue to no-cue approach, the subject may gradually learn to pair a movement sequence with the initiation of language. The final movement/gestural seque nce is a meaningless circular gesture. This gesture is both internally generated a nd generalizable to inte ractions outside of treatment. The movement sequence is neithe r word-related, nor sym bolic in nature. The same circular gesture is used for every word and is believed to be unrepresentative of any action familiar to the subject. Treatment wa s conducted individually for each patient across all phases.
18 Phase one (10 sessions): Each subj ect was seated at a desk directly facing a computer monitor with their head and body facing straight ah ead. To begin the trial, the therapist pressed a mouse button. A one-inch by one-inch star subseque ntly appeared at the center of the computer screen and a 1000 Hz tone sounded. To initiate the presence of the line drawing, each subject was instructed to lift the lid on a small box located to his/her left with their left ha nd and was further instructed to press a button located within the box. The button press caused the tone and st ar to disappear. Afte r a two-second delay, a black-and-white drawing appear ed at the center of the comp uter screen and a timer was initiated. If the subject named the picture correctly, the therapist pressed the mouse button that terminated the trial, stopped th e timer, and the line drawing was removed from the screen. If the subject provided an in correct response to the drawing, the therapist provided the correct name for the picture while simultaneously making the circular gesture described above with hi s/her left hand. The subject wa s instructed to repeat the corrected picture name aloud while also making the same circular gesture. This process was repeated for each of the 50 drawings, and the same 50 drawings were used for each session during this phase of treatment. Phase two (10 sessions): The same subject positioning and treatment procedure were used in phase two as were used in pha se one, except that the tone was eliminated from this phase, and a different set of 50 lin e drawings (not used in phase one), were depicted. Incorrect responses were corrected using the same procedure described above in phase one. Phase three (10 sessions): The same subject positioning and treatment procedure used in phases one and two were used in phase three. The trial was initiated the same way
19 it was initiated in phases one and two, however the subject was instructed to perform the same meaningless circular gesture performed in the previous phases w ith his/her left hand prior to the presentation of the line drawings. Response in structions, and correction of incorrect responses, were the same as descri bed above in phases one and two. Fifty line drawings different from those presented in pha ses one and two, were used in phase three. Two sets of line drawings were used : one set for patients receiving the balanced treatment (both high and low frequency words) , and one set for patients receiving the low frequency treatment. The balanced set of line drawings contai ned 15 high frequency words (21-717 occurrences per million), 15 medium frequency words (4-20 occurrences per million), and 20 low frequency words (0-30 occurrences per million), in order to provide a balanced set of words and prev ent subjects from obtaining ceiling effects during treatment. The low frequency set of line drawings contained 50 low frequency words. Frequencies were based on Francis and KuceraÂ’s Â“Frequency Analysis of the English LanguageÂ” (Francis & Kucera, 1982). Imaging Procedure Prior to beginning language treatment, patients completed a baseline, or pretreatment scan. Within two weeks of comple tion of the treatment, patients completed a posttreatment scan. All images were acquire d on a 3 Tesla GE Signa scanner using a dome-shaped RF quadrature head coil (MRI Devices). Functional imaging parameters were as follows: single shot spiral scan, gr adient echo pulse sequence, TE = 18ms, TR = 1660ms, FA = 60 degrees, FOV = 200mm, matr ix = 64 x 64, 32 slices with whole brain coverage, and slice thickness = 4.0mm. Struct ural imaging parameters were as follows: 3D spoiled GRASS sequence, TE = 6ms, TR = 23ms, FOV = 240mm, matrix size = 256 x 192 and slice thickness = 1.3mm.
20 A modified verbal fluency task was used in an event-related paradigm. The task used in the present study was modeled after the word generation task used in Crosson, Sadek, Maron et al., (2001). Prior to their pl acement in the magnet, participants provided written informed consent. The functional task was then explained and demonstrated for each participant. The participant was informed that they would be listening to a compact disc (CD) over a set of headphones while they were in the scanner. Participants were instructed to listen to the CD for a category, and to produce one exemplar of a member of that category. For example, if participants heard the category Â“farm animals,Â” they might respond with Â“cow.Â” Following their response, they were instructed to relax and await the presentation of the next categor y. Participants were instructed to give only one response for each category and were reminded to stay still in the scanner to minimize motion artifact. Participants were inst ructed to respond with Â“noÂ” if they were unable to hear a category, unable to understand the name of the category, or if they were unable to produce a response after attempting initiation, in order to avoid coding a response as Â“incorrect/otherÂ” when stimuli were not audible or were uninterpretable. FMRI stimuli were presented to subjects via a magnac oustic digital audio system and non-magnetic headset with microphone. Sound attenuation processes were performed prior to the functional scans to ensure that particip ants could hear the stimuli. Five runs with nine active events and variab le rest intervals were used. Interstimulus intervals (ISIs) equal to 21.58 seconds, 23.24 seconds, 24.9 seconds, and 26.56 seconds were randomized throughout each run. Prior to their participation in the fMRI task, response latencies for language production were measured for each patient. Interstimulus intervals were determined using patientsÂ’ mean response times plus 1.2 standard
21 deviations, thus allowing the patients to res pond to stimuli according to their own level of language production ability. Variable rest leng ths between category presentations allowed time for both the participantÂ’s response, and time for the hemodynamic response to return to baseline level. Overt responses were recorded to a laptop computer via the magnacoustic microphone. The same stimuli were presented to all part icipants, and runs were counterbalanced across subjects and acr oss pretreatment and posttreatment scan type for each participant. Following the fMRI procedure, participants were debriefed and offered an opportunity to view anatomical scan s of their brain. At th e conclusion of their pretreatment scans, participants were aske d to return after completing the Intention treatment, which was typically four w eeks in duration. The procedure for the posttreatment scan was identical to the pro cedure for the pretreatment scan. The same stimuli presented during the pretreatment s can were presented dur ing the posttreatment scan in a randomized order. Imaging Analyses A commercial software package (Cool Edit 2000 ) was used to record patient responses directly to a laptop computer. S canner noise was also recorded during the recording of the patientsÂ’ responses. The commercial software package was used to reduce the amount of scanner noise in the reco rded responses so that patient responses could be heard. Patient responses were then coded off-line as correct or incorrect . Recorded responses were also analyzed to determine the time at which the response was initiated. By determining the time at wh ich the response was initiated, the image acquisition number for each response may be determined.
22 Imaging data were analyzed usin g AFNI (3dDeconvolve) to derive functional maps based on response type. A thresholdi ng procedure was used to minimize large vessel effects by setting voxels in which the standard deviatio n of the acquired time series exceeded 5% of the mean signal, to zero. Smoothing of the images was not performed. Deconvolution analyses were conducted on the correct responses. Since speech-related movements greatly impact the integrity of the images, the first two images after response initiation were not modeled in the subjectÂ’s hemodynamic response. Regions of interest, specifically th e right hemisphere homologue of BrocaÂ’s area and its components: pars tr iangularis homologue and pars opercularis homologue, were traced using Localization of Functional Activity (LOFA) software. The tracing techniques, based on those of Foundas et al . (1998) noted above (see Figure 1) and described in Moore, Loftis et al. (2001), were employed si nce the anatomical landmarks for BrocaÂ’s area, specifically pars triangularis (PTr) and pars opercularis (POp), and the right hemisphere homologues of these areas, ar e considerably similar. For each subject, functional and anatomic images were converted to 1 mm3 voxels and deformed into atlas space (Talairach & Tournoux, 1988) to normaliz e each brain to a standard size and orientation. The deformation pro cess uses 10 landmark points to fit brains to atlas space. These points include: the midline posterior, s uperior margin of the anterior commisure (AC); the inferior margin of the posterior commisure (PC); two midsagittal points in the interhemispheric fissure; the left-most and ri ght-most points in the brain; the superiormost and inferior-most points in the brain; and the anterior and pos terior-most points in the brain.
23 The PTr and POp homologues were defined by an operator (M.E.G.) using LOFA. The operator used a cursor and the m ouse to draw the homologues of the PTr and POp (which comprise BrocaÂ’s homologue), and when present, the diagnol sulcus (DS). The PTr homologue was identified by findi ng and tracing the AHR and AAR. Tracings of the AHR were not initiated until a clearl y defined ventral/rostra l border was apparent. The AHR was traced from the rostral border to the intersection with the inferior portion of the AAR. The tracing was continued to in clude the AAR, which is the rostral boundary for the homologue of POp. Tracings for the homo logue of POp followed the extent of the Sylvian Fissure from the AAR to the ca udal boundary of the POp homologue, the anterior subcentral sulcus. C ontiguous tracings in cluded the DS when present. The rami for the homologues of PTr and POp were traced together on 6 to 10 contiguous sagittal slices. Within subjects, the same number of pretreatment and posttreatment images were traced. Once the homologues of PTr and POp we re traced, tracings were Â“thickenedÂ” to 3 mm on either side of the sulci to captu re the sulcal banks, and merged with correct response functional data to capture only activit y within the region of interest. Contiguous activation clusters with a product moment correlation of 0.40 or greater and a total volume of 25 Âµl or greater were identified in the homologues of PTr and POp, and in the DS when present. Many different techniques have been used to evaluate the results of fMRI. However, no standardized way of quantif ying, and subsequently qualifying, functional activity has been developed. One of the most seemingly contr oversial facets involved in the quantification of activity involves gr ouping images. By collapsing across subjects, activated voxels and activity clusters may be more eas ily localized and evaluated.
24 However, when working with patients who ha ve sustained structural brain damage, it is essential that each patient be evaluated individually. As illustrated in Figure 3, the anatomy of each brain is subs tantially different, suggesting that collapsing data across patients disregards important i ndividual differences such as lesion size and location, and corrupts the investigation and understanding of which structures may still be functional in individual subjects, as some structures may have sustained more damage in some subjects than in others. Consequently, data obtained for each subject were analyzed using withinsubject comparisons. Right Left Right Left Right Left Subject 01 Subject 02 Subject 03 Right Left Right Left Subject 04 Subject 05 Figure 3. Axial anatomic slices illustra ting individual lesions for each subject Cluster analyses were performed on activity, captured by the merged tracings and correct response function, on each set of partic ipantÂ’s images at each time period.
25 Conservative parameters were used to disti nguish activity believed to be associated with language production. Only activated voxels with an R2 > .16 were captured for analysis. Resulting cluster volumes for pr etreatment and posttreatment act ivation in the region of interest were compared within su bjects using chi-square analyses.
26 CHAPTER 3 RESULTS Raw time series data were deconvol ved with verbal res ponse initiation input vectors, specifically correct response input vectors. Anat omical tracings of the homologues of PTr and POp were executed acco rding to rules described previously in order to quantify changes in activity within th e region of interest. Cluster analyses were then performed on the resulting ar eas of activation within the re gion of interest in order to determine the number of active voxels. Chi square analyses were performed on raw number of active voxels within th e region of interest for each s ubject, in order to evaluate change in functional activity from pretreatment to posttreatm ent. Results are depicted in Table 1 and are summarized below. Table 1 Functional Activity and Correct Responses for Preand Posttreatment Scans Subject Pre treatment activity (%) Post treatment activity (%) 2 YatesÂ’ correction Number of pre treatment correct responses Number of post treatment correct responses 1 1.69 4.91 74.79* 73.82* 24/45 32/45 2 0.00 2.25 98.82* 96.82* 42/45 44/45 3 0.00 0.00 0.00 N/A 16/45 16/45 4 2.03 3.90 16.28* 15.67* 9/18 9/18 5 43.40 13.24 304.08* 302.78* 12/18 12/18 Note: * indicates p < 0.001; bold indicates treatment responder Individual subject performance during tr eatment was scored and evaluated by an independent, blind rater (a doctoral level speech-pathologist). The independent rater evaluated subject performance at each phase of treatment and rated each performance as showing a Â“positive treatment effect,Â” Â“a negative treatment effect,Â” or Â“no effect.Â”
27 Ratings were used to further classify subjec ts as either a Â“treatment responder,Â” or a Â“treatment non-responderÂ” based on their trea tment performance ratings. The C-Statistic was employed as a second method of examin ing response to treatment. The table below summarizes these results (Table 2). Table 2 Independent Rater and C-Statistic Determin ations of Stable Baseline and Positive Treatment Response Subject Stable Baseline Treatment Responder Independent Rater C-Statistic Independent Rater C-Statistic S01 X X X X S02 X X X X S03 X X S04 X X X S05 X X X X S01Â’s performance was rated as showing a stable baseline and positive treatment effects at each phase. Subse quently, S01 was classified as a Â“treatment responder.Â” S01 also showed significantly more functional activation in th e right hemisphere homologue of BrocaÂ’s area following treatment ( 2 = 74.79, p < 0.001; YatesÂ’ correction = 73.82, YatesÂ’ p < 0.001). S02Â’s performance was rated as showing a stable baseline and positive treatment effects at each phase. S02 was classified as a Â“treatment responder.Â” Furthermore, S02 showed significantly more activation in right BrocaÂ’s area homologue at posttreatment when compared to pretreatment activity ( 2 = 98.82, p < 0.001; YatesÂ’ correction = 96.82, YatesÂ’ p < 0.001). S03Â’s data indicated a stable baseline, and positive treatment effects for the first two phases of treatment, however, pha se three performance was rated as Â“a questionable positive treatmen t effect,Â” suggested by a decrease in the performance arc from phase two. This subj ectÂ’s performance suggests that a positive treatment effect was initially observed, but th at treatment effects were not consistently
28 maintained. Subsequently, this subject could not be classified as either a Â“treatment responderÂ” or a Â“treatment non-responderÂ” by the independent rater. Moreover, S03 did not show a significant change in functional activity in the right hemisphere homologue of BrocaÂ’s area from pretreatment to posttreatment ( 2 = 0.000087, p > .05). S04Â’s performance indicated an uns table baseline and no treatm ent effect for phase one. However, positive treatment effects were indicated in phase two and phase three performances. Subsequently, this subject was classified as a Â“treatment responder.Â” S04 also showed a significant incr ease in the amount of functiona l activation in right BrocaÂ’s homologue at posttreatment when comp ared to the pretreatment scan ( 2 = 16.28, p < 0.001; YatesÂ’ correction = 15.67, YatesÂ’ p < 0.001). S05Â’s performance suggested a stable baseline, but no positive treatment effect for phase one. However, phases two and three were rated as showing positive treatment effects. Thus, S05 was also classified as a Â“treatment responder.Â” An examination of func tional data revealed a significant decrease in right BrocaÂ’s area from pret reatment to posttreatment ( 2 = 304.08, p < 0.001; YatesÂ’ correction = 302.78, YatesÂ’ p < 0.001). In summary, four subjects were classified as treatment responders, and of those, three showed a significant increase in right hemisphere activity in BrocaÂ’s homologue, while one subject showed a significant decrease in activity, and another unclassifiab le subject, showed no change in the amount of activity in this area (Table 3). Table 3 Summary of Patterns in Functional Activ ation from Preto Posttreatment Increased Activation in BrocaÂ’s homologue Decreased Activation in BrocaÂ’s homologue No Change in Activation in BrocaÂ’s Homologue S01 S02 S05 S03 S04
29 CHAPTER 4 DISCUSSION An event-related word generation task requiring overt responses during fMRI was used to examine functional activity associat ed with language production in five patients with nonfluent aphasia resulting from left MCA stroke. Patients were instructed to provide one exemplar of a member of a cat egory presented to them auditorally during scanning. Overt responses were coded as correct and incorrect , based on response appropriateness for each presented category. Subsequently, time of correct response initiation was noted for each response and th e image acquisition number associated with the response was determined. Correct response input vectors were deconvolved with raw time series data. Anatomical tracings of the region of interest (ROI) were performed and merged with the deconvolved data, and cluste r analyses were performed to determine the raw number of active voxels cap tured within the ROI analys is. The same procedure was used for both pre and posttreatment scans. Fo llowing the pretreatment scan, participants completed the Intention treatment designed to ta rget right medial-front al cortex in order to engage the intentional component of the traditionally la nguage non-dominant hemisphere. By engaging the right medial-fr ontal cortex, the treat ment was designed to align medial and lateral cortical activ ation for language production within one hemisphere. It was proposed that signifi cant changes in language abilities following treatment would be represented by changes in measured functional act ivation in the right lateral-frontal region (BrocaÂ’s homologue ) from pre to posttreatment scans.
30 Increased Right Hemisphere Activity from Preto Posttreatment Findings suggest that those patients whose performance reflected positive treatment effects at each phase of treatment (S01 and S 02) also showed a concurrent increase in functional activity in the right hemisphere homologue of BrocaÂ’s area during an fMRI word generation task. S04, who initially showed no treatment effect at phase one of the treatment, appeared to respond positively to treatment at phases two and three. Imaging data for S04 also showed a significant in crease in functional activity in BrocaÂ’s homologue following treatment. These findings suggest that the re habilitative language intervention appears to have successfully e ngaged right hemisphere regions for language production. Of note, lesion locus and extent is similar among these three subjects. Specifically, there is extensive frontal, tempor al, and parietal damage with extensions into subcortical structures. Decreased Right Hemisphere Activity from Preto Posttreatment One subject (S05), who initially show ed no treatment effect at phase one of the treatment, appeared to respond positively to tr eatment at phases two and three. In contrast to the aforementioned treatment responders, im aging data for S05 revealed a significant decrease in functional activity in BrocaÂ’s homologue following treatment. These findings are difficult to interpret, but not impossible. While they pe rformed similarly in treatment, S04 and S05 differed considerably in both stro ke severity and extent of damage (see Figure 3 above). S05 also differed considerably from S01 and S02 in stroke severity and extent of damage (also notable in Figure 3) . Specifically, S05 sustained less extensive damage from the stroke. While subcortical structures were largely compromised in S01, S02, and S04, these structures appe ar relatively intact in S05. It is al so possible that S05 experienced preservation of pe ri-lesional cortex that may not have been preserved in the
31 other four subjects. Thus, it is possible that while a reorgani zation of function to the right hemisphere homologue of BrocaÂ’s area was observed in the three subjects discussed above, a recovery of function may have occurr ed for S05 suggesting that left hemisphere functioning was restored. Further examination of the left late ral-frontal areas is required in order to more clearly understand whethe r a recovery of func tion by left lateral perilesional areas has occurred before furthe r conclusions can be made for this subject. No Change in Right Hemisphere Activi ty from Preto Posttreatment Findings also indicated that for one subj ect (S03), who showed variable treatment effects across time, no significant change in right hemisphere activity was observed in BrocaÂ’s homologue. While no substantial c onclusions can be made based on a single subjectÂ’s performance, it appears that inconsis tent treatment effects may be associated with no significant changes in activation in the region of interest in the right hemisphere. Further examination of S03Â’s pre and posttreatment functional activity, particularly in the right medial areas, is warranted to examine wh ether treatment effects were evident in this area since it was the target area for treatment. Conclusions This study documents successful implementation of a behavioral paradigm employing overt responses with aphasic patient s. Use of overt responses during an fMRI behavioral paradigm offers valuable insight into activation associated with language production by allowing examination of response type and verification of task compliance. In addition, region of interest analyses may also provide valuable information that may help to determine which areas are still func tional in patients who have suffered brain dysfunction. This study also suggests that fMRI provides valuable information about areas of function that may potentially influe nce rehabilitative effo rts. Of particular
32 importance are results from this study that s uggest that the locati on and extent of the lesion is an important factor in language rehabilitation, specifically in terms of implications for a reorganizati on of function to the homologous right hemisphere regions. Limitations It is important to note that while fMRI ma y offer further insight s as to the location of activation following language rehabilitati on, functional imaging is, for the moment, unable to distinguish between activity in a lesion due to incomplete infarction and activity due to new neurons and connections (Rijntjes & Weiller, 2002). Cao et al. (1999) also described difficulties in distinguishi ng between the unknown pr operties associated with the reorganization of function to the right hemisphere, namely whether a reactivation of preexisting right hemisphere mechanisms occurs, and/or determining whether a recruitment of new language area s occurs. It has been suggested that a reorganization of function to the right hemis phere is more consistent with the hypothesis that a reactivation of a preexis ting language network area occurs to compensate for left hemispheric dysfunction (Cao et al., 1999; Rosen et al., 2000). Implications and Future Directions It appears that the innovative techniques developed in association with this study, such as: requiring overt responses from nonfluent aphasic patients during fMRI, deconvolving imaging data based on response type (e.g. correct response input vectors), and excluding the first two moti on affected images associated with participant response initiation from the modeled hemodynamic re sponse; in congruence with an innovative approach to treatment and language rehabil itation, provide an appropriate measure of language production in an aphasic population.
33 In addition to measurement innovati on, the findings from this study suggest that lesion location and extent impacts neural s ubstrates of rehabil itation. Chronic aphasic patients with larger and more extensive left hemisphere damage, who show poorer recovery, do not appear to show spontane ous reorganization of function to the right hemisphere. However, the results of this study suggest that chronic ap hasic patients with large lesions, who have poor r ecovery of function by the left hemisphere areas, but who respond to a treatment targeting the right he misphere medial-frontal areas, appear to show good reorganization of func tion to the right hemisphere. Finally, the results of this study also suggest that chroni c aphasic patients with less severe and less extensive lesions, who show good recovery in response to similar types of treatment, may initially reorganize function to the right hemisphere , but may then recover function in left hemisphere peri-lesional areas. Further inves tigation of left hemisphere peri-lesional areas is necessary before any substantial co nclusions can be made about recovery of function in left hemisphere peri-lesional areas in less severe left MCA stroke patients. In summary, findings from this study suggest that support can be found for both the recovery of function and reorganization of function positions. Data to support these positions may be examining two different samples of patients, with studies documenting recovery of function enrolling aphasic patients with less severe and less extensive left hemisphere lesions, and studies documenting reorganization of function enrolling aphasic patients with more severe and more extensive left hemisphere lesions. Future directions driven by the fi ndings from this study will pursue further region of interest analyses fo cusing on activity in the left lateral-frontal peri-lesional areas, as well as on activity in bilateral medial-frontal regi ons. Analyses of left lateral-
34 frontal peri-lesional areas ma y offer further insight into lesion size association with recovery of function and/or reorganization of function. Analyses of the bilateral medialfrontal regions may offer furthe r explanation of functional acti vity that is more closely associated with the targeted areas of treat ment and may provide a more pure measure of treatment effects associated with the Intention treatment. The implications of measuring func tional activity associated with rehabilitation are also of great importance, specifically in regard to predicting treatment effectiveness for individual patients. This research may he lp direct rehabilitative efforts on a patientby-patient basis by examining lesion size and patterns of functional activation prior to treatment. This approach to treatment ma y better aid in devel oping patient-specific treatments that will most ut ilize areas that are still functional in each patient, thereby maximizing treatment effectiveness and maintaining s ubsequent recovery.
35 REFERENCE LIST Abdullaev, Y. G., & Posner, M. I. (1998). Event-related brain potential imaging of semantic encoding during processing single words. NeuroImage, 7, 1-13. Belin, P., Van Eeckhout, P., Zilbovicius, M ., Remy, P., Francois, C., Guillaume, S., Chain, F., Rancurel, G., & Samson, Y. (1996) . Recovery from nonfluent aphasia after melodic intonation therapy: A PET study. Neurology, 47, 1504-1511. Binder, J. R., Frost, J. A., Hammeke, T. A ., Cox, R. W., Rao, S. M ., & Prieto, T. (1997). Human brain language areas identified by functional magnetic resonance imaging. The Journal of Neuroscience, 17, 353-362. Broca, P. (1997). On the speech center. In L. Benjamin, Jr. (Ed.), A history of psychology: Original sources and contemporary research (pp. 77-81). New York: The McGraw-Hill Companies, Inc. (O riginal work published 1861). Burton, M. W., Noll, D. C., & Small, S. L. (2001). The anatomy of auditory word processing: Indivi dual variability. Brain and Language, 77, 119-131. Burton, M. W., Small, S. L., & Blumstein, S. E. (2000). The role of segmentation in phonological processing: An fMRI investigation. Journal of Cognitive Neuroscience, 12, 679-690. Cao, Y., Vikingstad, E. M., George, K. P., Johnson, A. F., & Welch, K. M. A. (1999). Cortical language activation in stroke pati ents recovering from a phasia with functional MRI. Stroke, 30, 2331-2340. Caplan, R., & Dapretto, M. (2001). Making sense during conversation: An fMRI study. NeuroReport, 12, 3625-3632. Cappa, S. F. (2000). Neuroimagi ng of recovery from aphasis. Neuropsychological Rehabilitation, 10, 365-376. Crosson, B., Rao, S. M., Woodley, S. J., Rosen, A. C., Hammeke, T. A., Bobholz, J. A., Mayer, A., Cunningham, J. M., Fuller, S. A., Binder, J. R., Cox, R. W., & Stein, E. A. (1999). Mapping of semantic, phonological, and orthographic verbal working memory in normal adults with fMRI. Neuropsychology, 13, 171-187.
36 Crosson, B., Sadek, J. R., Bobholz, J. A., Gockay, D., Mohr, C. M., Leonard, C. M., Maron, L., Auerbach, E. J., Brows, S. R ., Freeman, A. J., & Briggs, R. W. (1999). Activity in the paracingulate and cingulat e sulci during word ge neration: An fMRI study of functional anatomy. Cerebral Cortex, 9, 307-316. Crosson, B., Sadek, J. R., Maron, L., Gockay, D., Mohr, C. M., Auerbach, E. J., Freeman, A. J., Leonard, C. M., & Briggs, R. W. (2001) . Relative shift in activ ity from medial to lateral frontal cortex during internally versus externally guided word generation. Journal of Cognitive Neuroscience, 13, 272-283. de Zubicaray, G. I., Wilaon, S. J., McMahon, K. L., & Muthiah, S. (2001). The semantic interference effect in the picture-word paradigm: An event-related fMRI study employing overt responses. Human Brain Mapping, 14, 218-227. Fiez, J. A. (2001). Neuroimaging studies of speech: An overview of techniques and methodological approaches. Journal of Communication Disorders, 34, 445-454. Finger, S. (1997). Cortical localization and cer ebral dominance: The work of Paul Broca. In L. Benjamin, Jr. (Ed.), A history of psychology: Orig inal sources and contemporary research (pp. 77-81). New York: The McGraw-Hill Companies, Inc. (Original work published 1994). Foundas, A. L., Eure, K. F., Luevano, L. F., & Weinberger, D. R. (1998). MRI asymmetries of brocaÂ’s area: The pars triangularis and pars opercularis. Brain and Language, 64, 282-295. Francis, W. N., & Kucera, H. (1982). Frequency analysis of English usage: Lexicon and grammar. Boston: Houghton Mifflin Co. Gold, B. T., & Kertesz, A. (2000). Right hemis phere semantic processing of visual words in an aphasic patient: An fMRI study. Brain and Language, 73, 456-465. Hertz-Pannier, L., Chiron, C., Jambaque, I ., Renaux-Kieffer, V., Van de Moortele, P., Delalande, O., Fohlen, M., Brunelle, F., & Le Bihan, D. (2002). Late plasticity for language in a childÂ’s non-dominant hemis phere: A preand pos t-surgery fMRI study. Brain, 125, 361-372. Hickok, G., Love, T., Swinney, D., Wong, E. C., & Buxton, R. B. (1997). Functional MR imaging during auditory word perception: A single-trial presentation paradigm. Brain and Language, 58, 197-201. Holland, S. K., Plante, E., Byars, A. W., Stawsburg, R. H., Schmithorst, V., & Ball, W. S. (2001). Normal fMRI brain activation pattern s in children performing a verb generation task. NeuroImage, 14, 837-843.
37 Huang, J., Carr, T. H., & Cao, Y. (2001). Comp aring cortical activa tions for silent and overt speech using event-related fMRI. Human Brain Mapping, 15, 39-53. Karbe, H., Thiel, A., Weber-Luxenberger, G., Herholz, K., Kessler, J., & Heiss, W. D. (1998). Brain plasticity in poststroke aphasi a: What is the contribution of the right hemisphere? Brain and Language, 64, 215-230. Khatri, P., & Hier, D.B. (2000). Imaging aphasia: The coming paradigm shift. Brain and Cognition, 4, 60-63. Lichtime, L. (1885). On aphasia. Brain, 7, 433-484. Miura, K., Nakamura, Y., Miura, F., Yama da, I., Takahasi, M., Yoshikawa, A., & Mizobata, T. (1999). Functional magnetic resona nce imaging to word generation task in a patient with BrocaÂ’s aphasia. Journal of Neurology, 246, 939-942. Moore, A. B., Crosson, B., Gockay, D., Leonard, C. M., & Foundas, A. L. (2000). Defining pars triangularis. Society for Neuroscience Abstracts, 30, Part 1, 1247. Moore, A. B., Loftis, C., Crosson, B., Gockay, D., Leonard, C. M., & Foundas, A. L. (2001). Localization of functional activity in pars triangularis. Journal of the International Neuropsyc hological Society, 7, 197. Musso, M., Weiller, C., Kiebel, S., Muller, S. P., Bulau, P., & Rijintes, M. (1999). Training-induced brain plasticity in aphasia. Brain, 122, 1781-1790. Nadeau, S. E., & Crosson, B. (1995). A guide to the functional imaging of cognitive processes. Neuropsychiatry, Neuropsychology, and Behavioral Neurology, 8, 143-162. Nadeau, S., Rothi, L. J. G., & Crosson, B. (2000). Preface. In S. Nadeau, L. J. G. Rothi, & B. Crosson (Eds.), Aphasia and language: Theory to practice. New York: Guilford Press. Pizzamiglio, L., Galati, G., & Committeri, G. (2001). The contribution of functional neuroimaging to recovery af ter brain damage: A review. Cortex, 37, 11-31. Richards, K., Singletary, F., Rothi, L. J. G., Koehler, S., & Crosson, B. (2002). The activation of intentional m echanisms through utilization of nonsymbolic movements in aphasia rehabilitation. Journal of Rehabilitation Research and Development, 39, 445454. Rijntjes, M., & Weiller, C. (2002). Recovery of motor and language ab ilities after stroke: The contribution of functional imaging. Progress in Neurobiology, 66, 109-122.
38 Rosen, H. J., Petersen, S. E., Linenweber, M. R., Snyder, A. Z., White, D. A., Chapman, L., Dromerick, A. W., Fiez, J. A., & Corbetta , M. (2000). Neural co rrelates of recovery from aphasia after damage to left inferior frontal cortex. Neurology, 55, 1883-94. Sakai, K. L., Hashimoto, R., & Homae, F. (2001). Sentence processing in the cerebral cortex. Neuroscience Research, 39, 1-10. Staudt, M., Grodd, W., Niemann, G., Wildgr uber, D., Erb, M., & Krageloh-Mann, I. (2001). Early left periventricular brain lesi ons induce right hemispheric organization of speech. Neurology, 57, 122-125. Talairach, J., & Tournoux, P. (1988). Co-Planar Sterotaxic Atlas of the Human Brain . New York: Thiem Medical Publishers. Thomas, C., Altenmuller, E., Marckmann, G., Kahrs, J., & Dichgans, J. (1997). Language processing in aphasia: Changes in latera lization patterns durin g recovery reflect cerebral plasticity in adults. Electroencephalography and C linical Neurophysiology, 102, 86-97.
39 BIOGRAPHICAL SKETCH Megan Elizabeth Gaiefsky was born in La Mirada, California on July 6, 1978. She graduated from Alta Loma High School in 1996 and received her Bachelor of Arts degree in psychology from Trinity University, Sa n Antonio, Texas, in May 2000. From May 2000 through September 2000, Megan Gaiefsky was employed as a research associate in the Department of Neurology at Harbor-UCL A Medical Center Research and Education Institute in Torrance, California. In Septem ber 2000, she was asked to pursue a research assistant position in the Me dical Department at Brookha ven National Laboratory in Upton, New York. She was employed in the Medical Department at Brookhaven National Laboratory from September 2000 through July 2001. In August 2001, Megan Gaiefsky enrolled in the doctoral program in the Depart ment of Clinical and Health Psychology at the University of Florida. Her clinical and re search interests are in the area of clinical neuropsychology.