<%BANNER%>

Broca's Area Thalamocortical Circuitry

Permanent Link: http://ufdc.ufl.edu/UFE0045211/00001

Material Information

Title: Broca's Area Thalamocortical Circuitry Effects of Anterior and Posterior Thalamic Lesions Investigated Using Diffusion-Weighted Tractography
Physical Description: 1 online resource (136 p.)
Language: english
Creator: Ford, Anastasia A
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2013

Subjects

Subjects / Keywords: broca -- language -- thalamus -- tractography
Psychology -- Dissertations, Academic -- UF
Genre: Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Language is a complex cognitive skill attributed almost exclusively to humans. Early language models attribute almost all language processing to two cortical zones,namely Broca’s and Wernicke’s areas. Recent studies implicate a number of othercortical as well as subcortical structures to be involved in languageprocessing. The present manuscript focuses on structural connectivity betweenBroca’s area and the thalamus, specifically ventral anterior nucleus andpulvinar. We demonstrate that Broca’s area shares direct connections with bothof these thalamic nuclei and present a Broca’s area—thalamocortical (orcorticothalamic) network involved in linguistic processing. In addition, wepresent models of anterior and posterior exemplar thalamic lesions andinvestigate their effects on this circuitry. Our results indicate that theexemplary anterior thalamic lesion eliminates about half of the en passagepathways traveling between Broca’s area, passing though ventral anteriornucleus and thence to pulvinar, while affecting only twenty percent of pathwaysconnecting Broca’s area with pulvinar. Posterior thalamic lesions result inless than fifteen percent connectivity reduction for Broca’s area—ventralanterior nucleus pathways. In contrast, the lesion eliminates more than half ofBroca’s area—pulvinar pathways and more than seventy percent of Broca’sarea—ventral anterior nucleus—pulvinar en passage pathways. We conclude thatdifferential effects of anterior/posterior thalamic lesions on circuitryconnecting Broca’s area and the thalamus may explain differences in languagedeficits in patients who had suffered damage to the dominant thalamus. However,our finding that both types of lesions affect a single structural networkconnecting Broca’s area, ventral anterior nucleus, and pulvinar may explainlargely overlapping set of core language deficits present in patients withdominant thalamic lesions.
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 Anastasia A Ford.
Thesis: Thesis (Ph.D.)--University of Florida, 2013.
Local: Adviser: White, Keith D.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2013
System ID: UFE0045211:00001

Permanent Link: http://ufdc.ufl.edu/UFE0045211/00001

Material Information

Title: Broca's Area Thalamocortical Circuitry Effects of Anterior and Posterior Thalamic Lesions Investigated Using Diffusion-Weighted Tractography
Physical Description: 1 online resource (136 p.)
Language: english
Creator: Ford, Anastasia A
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2013

Subjects

Subjects / Keywords: broca -- language -- thalamus -- tractography
Psychology -- Dissertations, Academic -- UF
Genre: Psychology thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Language is a complex cognitive skill attributed almost exclusively to humans. Early language models attribute almost all language processing to two cortical zones,namely Broca’s and Wernicke’s areas. Recent studies implicate a number of othercortical as well as subcortical structures to be involved in languageprocessing. The present manuscript focuses on structural connectivity betweenBroca’s area and the thalamus, specifically ventral anterior nucleus andpulvinar. We demonstrate that Broca’s area shares direct connections with bothof these thalamic nuclei and present a Broca’s area—thalamocortical (orcorticothalamic) network involved in linguistic processing. In addition, wepresent models of anterior and posterior exemplar thalamic lesions andinvestigate their effects on this circuitry. Our results indicate that theexemplary anterior thalamic lesion eliminates about half of the en passagepathways traveling between Broca’s area, passing though ventral anteriornucleus and thence to pulvinar, while affecting only twenty percent of pathwaysconnecting Broca’s area with pulvinar. Posterior thalamic lesions result inless than fifteen percent connectivity reduction for Broca’s area—ventralanterior nucleus pathways. In contrast, the lesion eliminates more than half ofBroca’s area—pulvinar pathways and more than seventy percent of Broca’sarea—ventral anterior nucleus—pulvinar en passage pathways. We conclude thatdifferential effects of anterior/posterior thalamic lesions on circuitryconnecting Broca’s area and the thalamus may explain differences in languagedeficits in patients who had suffered damage to the dominant thalamus. However,our finding that both types of lesions affect a single structural networkconnecting Broca’s area, ventral anterior nucleus, and pulvinar may explainlargely overlapping set of core language deficits present in patients withdominant thalamic lesions.
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 Anastasia A Ford.
Thesis: Thesis (Ph.D.)--University of Florida, 2013.
Local: Adviser: White, Keith D.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2013
System ID: UFE0045211:00001


This item has the following downloads:


Full Text

PAGE 1

1 POSTERIOR THALAMIC LESIONS INVESTIGATED USING DIFFUSION WEIGHTED TRACTOGRAPHY By ANASTASIA A. FORD A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Anastasia A. Ford

PAGE 3

3 To my family and friends for their continuing love and support

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank Drs. Crosson, Mareci, and White for their help in conceptualization of this project and my dissertation committee for their inputs on revisions of this manuscript.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF FIGURES ................................ ................................ ................................ .......... 7 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 EA AND THE THALAMUS: CORTICAL AND SUBCORTICAL REGIONS INVOLVED IN LANGUAGE PROCESSING ................................ .......... 12 Language Processing ................................ ... 12 ................................ ................................ .................... 12 ................................ ................................ ................... 13 Thalamus and Its Role in Language Processing ................................ ..................... 22 Electrophysiological Studies of Thalamic Language Function .......................... 23 Lesion Effects on Thalamic Language Function ................................ ............... 24 Neuroimaging Evidence of Thalamic Language Functions ............................... 27 Thalamocortical Language Circuitry? ................................ ............................... 30 The Present Study ................................ ................................ ............................ 40 2 METHODS ................................ ................................ ................................ .............. 44 Participants ................................ ................................ ................................ ............. 44 Image Acquisition and Processing ................................ ................................ .......... 44 Acquisition Parameters ................................ ................................ ..................... 44 Image Processing ................................ ................................ ............................. 45 Regions of Interest ................................ ................................ ........................... 46 Cortical regions of interest ................................ ................................ ......... 46 Subcortical Regions of Interest ................................ ................................ .. 46 Exclusion masks ................................ ................................ ........................ 47 Thalamic lesion models ................................ ................................ ............. 47 Tractography analysis ................................ ................................ ................ 47 Quantitative tractography measures ................................ .......................... 49 3 ................................ ............ 52 Ventral Anterior Nucleus Pathways ................................ ................ 52 Pulvinar Pathways ................................ ................................ .......... 55 Ventral Anterior Nucleus Pulvinar Projections ............................. 57 ............................ 60

PAGE 6

6 4 EFFECTS OF ANTERIOR AND POSTERIOR THALAMIC LESIONS ON .......................... 73 Anterior Thalamic Lesion Effects ................................ ................................ ............ 73 and Ventral Anterior Nucleus ................................ ................................ ........ 73 Anterior Thalamic Lesion and Pulvinar ................................ ................................ ................................ .. 75 with Ventral Anterior Nucleus and Pulvinar ................................ ................... 77 Summary of Anterior Thalamic Lesion Effects on Connections between d Thalamus ................................ ................................ .......... 79 Posterior Thalamic Lesion Effects ................................ ................................ .......... 81 and Ventral Anterior Nucleus ................................ ................................ ........ 82 and Pulvinar ................................ ................................ ................................ .. 84 Posterior Th with Ventral Anterior Nucleus and Pulvinar ................................ ................... 86 Summary of Posterior Thala mic Lesion Effects on Connections between ................................ ................................ .......... 88 Conclusions: Differential Effects of Anterior and Posterio r Thalamic Lesions on ................................ ................................ 89 5 DISCUSSION ................................ ................................ ................................ ....... 112 ........................... 112 Thalamocortical Circuitry And Corresponding Language Deficits ...................... 117 Anterior Thalamic Lesion Effects ................................ ................................ .... 117 Posterior Thalamic Lesion Effects ................................ ................................ .. 121 Study Limitations and Future Directions ................................ ............................... 124 LIST OF REFERENCES ................................ ................................ ............................. 129 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 136

PAGE 7

7 LIST OF TABLES Table page 2 1 Participant Demographics. ................................ ................................ ................. 51 3 1 Region of Interest Surface Area Measures. ................................ ....................... 63 3 2 Nucleus of the Thalamus. ................................ ................................ ................... 63 3 3 Anterior Nucleus of the Thalamus. ................................ ................................ ..... 64 3 4 ............... 64 3 5 ....... 65 3 6 Nucleus and Pulvinar. ................................ ................................ ......................... 65 3 7 Tract Edge Weights for Pa Anterior Nucleus and Pulvinar. ................................ ................................ ........... 66 4 1 Anterior Thalamic Lesion Effects on Pathways Co Ventral Anterior Nucleus of the Thalamus (Tract Volumes). ............................... 91 4 2 Anterior Thalamic Lesion Effects on P Ventral Anterior Nucleus of the Thalamus (Tract Edge Weights). ...................... 92 4 3 Pulvinar (Tract Volumes). ................................ ................................ ................... 93 4 4 Anterior Thalamic Les Pulvinar (Tract Edge Weights). ................................ ................................ ........... 94 4 5 Anterior Thalamic Lesion Effects on Pat Ventral Anterior Nucleus of the Thalamus, and Pulvinar (Tract Volumes). ......... 95 4 6 Anterior Thalamic L Ventral Anterior Nucleus of the Thalamus, and Pulvinar (Tract Edge Weights). ................................ ................................ ................................ ............ 96 4 7 and Ventral Anterior Nucleus of the Thalamus (Tract Volumes). ........................ 97 4 8 and Ventral Anterior Nucleus of the Thalamus (Tract Edge Weights). ............... 98

PAGE 8

8 4 9 and Pulvinar (Tract Volumes). ................................ ................................ ............ 99 4 10 and Pulvinar (Tract Edge Weights). ................................ ................................ .. 100 4 11 Ventral Anterior Nucleus, and Pulvinar (Tract Volumes). ................................ 101 4 12 Ventral Anterior Nucleus, and Pulvinar (Tract Edge Weights). ......................... 102 4 13 Thalamocortical Circuitry. ................................ ................................ ................. 103

PAGE 9

9 LIST OF FIGURES Figure page 2 1 Lesion models rendered from two clinical cases. ................................ ............... 51 3 1 Projections connecting pars opercularis (a) and pars triangularis, (b) with ventral anterior nucleus in participant 1. ................................ ............................. 67 3 2 Pathways connecting pars opercularis (a) and pars triangularis, (b) with ventral anterior nucleus of the thalamus ................................ ............................. 68 3 3 Projections connecting pars opercularis (a) and pars triangularis, (b) with pul vinar in participant 1 ................................ ................................ ....................... 69 3 4 Pathways connecting pars opercularis (a) and pars triangula ris, (b) with pulvinar ................................ ................................ ................................ .............. 70 3 5 Projections among opercularis (a), pars triangularis (b), ventral anterior nucleus and pul vinar in participant 1 ................................ ................................ 71 3 6 Pathways connecting pars opercularis (a) and pars triangularis (b) with ventral anterior nucleus of the thalam us and pulvinar ................................ ........ 72 4 1 Effects of an anterior thalamic lesion on pathways connectin and the thalamus ................................ ................................ .............................. 104 4 2 Anterior thalamic lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) and ventral anterior nucleus ................................ ....... 105 4 3 Anterior thalamic lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) and pulvinar ................................ ............................... 106 4 4 Anterior thalamic lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) with ventral anterior nucl eus and pulvinar ................. 107 4 5 Effects of a posterior thalamic lesion on pathways connectin and the thalamus ................................ ................................ .............................. 108 4 6 Posterior thalamic lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) and ventral anterior nucleus. ................................ ...... 109 4 7 Posterior lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) with pulvinar. ................................ ................................ ............. 110 4 8 Posterior thalamic lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) with ventral anterior nucle us and pulvinar .................. 111

PAGE 10

10 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy POSTERIOR THALAMIC LESIONS INVESTIGATED USING DIFFUSION WEIGHTED TRACTOGRAPHY By Anastasia A. Ford May 2013 Chair: Keith White Major: Psychology Language is a complex cognitive skill attributed almost exclusively to humans. Early language models attribute almost all language processing to two cortical zones, ber of other cortical as well as subcortical structures to be involved in language processing. The thalamus, specifically ventral anterior nucleus and pulvinar. We demonstra te that thalamocortical (or corticothalamic) network involved in linguistic processing. In addition, we present models of anterior and posterior exemplar t halamic lesions and investigate their effects on this circuitry. Our results indicate that the exemplary anterior thalamic lesion eliminates about half of the en passage pathways hence to pulvinar. Posterior thalamic lesions result in less than fifteen percent connectivity ventral anterior nucleus pathways. In contrast, the lesion

PAGE 11

11 pulvinar pathways and more than seventy ventral anterior nucleus pulvinar en passage pathways. We conclude that differential effects of anterior/posterior thalamic lesions on cir cuitry in patients who had suffered damage to the dominant thalamus. However, our finding area, ventral anterior nucleus, and pulvinar may explain largely overlapping set of core language deficits present in patients with dominant thalamic lesions.

PAGE 12

12 CHAPTER 1 INVOLVED IN LANGUAGE P ROCESSING hemisphere. Gross anatomy defining this area can be encapsulated by pars triangular i s and pars opercularis. More specifically, the ventral border is defined by the Sylvian fissure and dorsal border by the inferior frontal sulcus. Anterior border is defined by a coronal plane through the anterior margin of the anterior horizontal ramus of the Sylvian fissure. Inferior precentral sulcus defines the posterior border, while the anterior ascending ramus of the Sylvian fissure defines the border between pars triangularis and pars opercularis. Prominent individual differences exist in the size, s hape, and relative within the left hemisphere is larger than that of its right hemisphere homologue (Galaburda, 1980; Amunts et al., 1999). Interestingly, this hemispheric asymmetry is also present in great apes, indicating that left hemisphere dominance in vocalization developed as early as five million years ago (Cantalupo & Hopkins, 2001; Corballis, 2003). (BA) 44 and 45. BA 44 and 45 share many common features. In particular, both areas demonstrate prominent pyramidal cells in layers III and V, no clearly defined border between la yers II and III, and low cell density in layer VI. Defining characteristic that separates these areas is that layer IV in BA 44 is very thin and this area is thus called

PAGE 13

13 However, Amunts and colleagues have demonstrated that a previously assumed association between sulcal patterns and cytoarchitectonic borders does not always hold (Amunts et al., 1999). In particular, they s howed that in general there is significant individual variability in the location and extent of BA 44 and 45. Although BA 44 is typically located within pars opercularis it may encroach on pars tri a ngularis. Similarly, BA 45 is found within pars triangular is but may extend into pars opercularis and occupy parts of the middle frontal gyrus ( Amunts et al., 1999; Rajkowska, 1995; Goldman Rakic, 1995). Because cytoarchitectonic criteria cannot presently be evaluated directly in live human brains, studies of la nguage functions in living participants infer locations within rostral/caudal border between BA s 45 and 44 faces uncertainty for living participants due to in di vidual stru ctural variations and only partial correlation between sulcal landmarks and cytoarchitectonic borders. Differing approaches have been taken across studies to manage these uncertainties. Identification of BA 44 and BA 45 in discussions below will not necess arily be c ytoarchitectonically neither accurate nor consistent across studies. Functionally, BA 44 and 45 can be differentiated based on the types of language processes that they suppor t (Binder et al., 1996; Cabeza, 2000; Nyberg, 2000; Friederici, 2002, 2004; Amunts et al., 2004). In particular, BA 45 has been implicated to be involved in semantic processes, while BA 44 was shown to support syntactic and phonological processing (Amunts et al., 2004; McDermott et al., 2003 ; Devlin et al., 2003 ). Tasks that involve category member generation based on pre specified semantic

PAGE 14

14 category also tend to activate BA 45 to a greater extent than BA 44. Amunts and colleagues examined differences in activation patterns between BA 45 and 44 using a task that required participants to generate either members of a specified semantic category (i.e. animals) or over learned categories (days of the week, months of the year) (Amunts et al., 2004). To localize resulting activation the authors used p robabilistic cytoarchitectonic maps of BA 44 and 45 (Amunts et al., 1999). They found that both BA 44 and 45 were activated during both semantic fluency conditions, however BA 45 was activated to a higher extent when activation from category member generat ion was contrasted with over learned categories. This finding implies that BA 45 may be involved in processing of verbal fluency tasks with high semantic load. A recent meta analysis study examining functional imaging studies employing similar semantic flu ency tasks also identified BA 45 as a region showing consistent activation during semantic processing (Costafreda et al., 2006). In an functional magnetic resonance imaging (f MRI ) study that asked participants to judge whether a presented sentence was sema ntically correct or incorrect, Hagoort and colleagues observed an increased activation of BA 47/45 during processing of semantically odd sentences (Hagoort et al., 2004). The authors postulated that an increased level of activation in this area indicates that it is involved with establishing the (i.e. words with double meaning) also show an incre ased activation of BA 45 (Rodd et al., 2005 ), which is attributed to higher semantic unification demands because a coherent interpretation of this type of sentence requires additional contextual processing

PAGE 15

15 and reference (Hagoort et al., 2004 ). Further evid ence of this can be seen from studies employing complement coercion sentences (Husband et al., 2011). Complement coercion sentences are grammatically correct and semantically plausible sentences that require additional semantic operations necessary for sen tence comprehension. An most likely interpretation would be the latter one. Thus, for each sentence the participant has to recruit additional linguistic resources indica tive of higher semantic unification load in order to correctly interpret a complement coercion sentence. FMRI data demonstrate that processing of these types of sentences results in higher levels of activation within BA 45 as compared with control sentence s (Husband et al., 2011). Based on the presented evidence we can conclude that tasks involving higher semantic load that requires additional linguistic processing preferentially recruit BA 45 within the left hemisphere. In addition to semantic processing, BA 45 may be involved in processing of syntax. Horwitz and colleagues asked participants to recall an incident from the past and describe this incident while undergoing positron emission tomogr aphy (PET) (Horwitz et al., 2003 ). In addition to English, half of the participants were fluent in American Sign Language (ASL) and were asked to sign the story. Controlling for articulatory and gestural components, BA 45 showed an increase in activation during story telling. This activation was located within the cor e portion of BA 45 (as determined

PAGE 16

16 by application of probabilistic cytorarchitectonic maps provided by Amunts et al., 1999) for both verbal and gestural discourse. This observation implies that BA 45 may be involved in both semantic and syntactical processi ng. In order to create a coherent story participants must ensure that content of the words in the sentences match semantically and conveys a consistent story line. At the same time, combination of words within the sentences has to follow grammatical rules in order to create correct sentence structure. Interestingly, in this study activation of BA 44 was observed only during oral and hand control motor tasks, where participants were asked to produce laryngeal and oral articulatory movements and sounds devoid of linguistic content or to move their hands and arms in a random pattern. Thus, in this study activation of BA 44 can be attributed to complex, non linguistic laryngeal and limb motor and sound production components associated with speech and sign produc tion (Horwitz et al., 20 03 ). Heim and colleagues designed an fMRI experiment to examine how semantic, 2008). In this study the participants were asked to perform thr ee different tasks. The first task tested semantic fluency and required participants to overtly generate examples of 6 semantic categories (birds, mammals, food, weapons, tools, and toys). Phonological fluency task required participants to generate nouns s tarting with b, f, k, m, sh, t. Lastly syntactic fluency task required participants to generate nouns of masculine, feminine, or neuter genders (a common feature of German language). The results showed consistent activation of both BA 44 and 45 during all three tasks. More specifically, BA 45 activation was of equivalent extent for all three verbal fluency tasks. BA 44 activation on the other hand was higher for phonological fluency task as compared with two other

PAGE 17

17 conditions. This finding lead the authors t o conclude that BA 44 preferentially supports phonological processing while also being involved (to a lesser extent) in semantic and syntactic processing. Equivalent activation of BA 45 on all three tasks may indicate that this area is involved in controll ed retrieval of lexical information during speech production. The authors extended their hypothesis further in a follow up study (Heim et al., 2009). In this new study, participants engaged in the same tasks involving semantic and phonological fluency, ho wever the data were analyzed using dynamic causal modeling (DCM). In this modeling scheme BA 44 and 45 were represented as input regions that were driven by each of the two tasks, while motor cortex was represented as an output region for cortical language production network. Authors explicitly tested three different activation models based on the activation patterns observed in the first study. Model 1 can be described as follows. If BA 45 is involved in semantic and phonological retrieval during speech pr oduction then it will receive driving inputs during both of these tasks. Due to anatomical connectivity between BA 45 and 44, activation of BA 45 will then spread into BA 44 and activate it to a small extent. If, in addition, BA 44 is preferentially involv ed in phonological processing then this area will show an additional increase in activation during phonological fluency task. Thus Model 1 interprets varying degrees of activation within BA 44 during phonological processing as domain preferentiality of thi s area. Model 2 represents previously found activation patterns in BA 44 and 45 in terms of domain specificity. In particular, this model assumes that semantic processing will activate BA 45, while phonological processing will activate BA 44. Since two are as are connected, activation in one region may drive activation in the other. Previous

PAGE 18

18 observation of higher activation in BA 44 during phonological fluency task may indicate that BA 45 drives activation in BA 44 to a larger extent than the drive resulting from BA 44 to BA 45. The last model, Model 3, assumes that both areas are equally involved in semantic and phonological processing and the activation drive produced by BA 45 is greater than that of BA 44. Resulting data analysis showed that Model 1 receiv ed the most support. In particular, authors show that both semantic and phonological fluency tasks drive BA 45 and it in turn drives BA 44. During phonological fluency task BA 44 inhibits activation of BA 45. This pattern of activation/inhibition results i n equivalent activation levels in BA 44 and 45 during semantic fluency task and an increased activation in BA 44 during phonological fluency task. The authors further hypothesize that inhibition imposed by BA 44 onto BA 45 during phonological processing ma y serve two functions. First, it may suppress lexical search within BA 45 once articulatory processing of the to be pronounced item is finalized within BA 44. Another explanation for this inhibition may be to narrow lexical search within BA 45 only to lexi cal items that fit appropriate phonological criterion as determined by BA 44. The main conclusion that we can draw from this series of experiments is that BA 44 is preferentially involved in phonological processing but also supports semantic and syntactic processing, though to a lesser extent. BA 45 on the other hand may be involved in word retrieval processing regardless of its search criteria (semantic, phonological, or syntactic). These findings seem to somewhat contrast results of Amunts and colleagues described earlier in this paper (Amunts et al., 2004). In their study Amunts and colleagues found activation of BA 44 and 45 during both semantic fluency conditions (category member generation and overlearned category generation). However, contrast

PAGE 19

19 betwee n the two conditions revealed higher levels of activation within BA 45 during category member generation. Differences in tasks and contrasts employed in these studies may be responsible for different patterns of activations that were observed. In addition to semantic and phonological processing involved in speech grammatical rules dictating the order of words in a sentence that enable successful interpretation of its meaning. In particular, some scholars believe that our ability to successfully process hierarchical sentence structures characterized by recursive embeddings is a unique feature of human language facult y (Friederici et al., 2004 ). To requiring hierarchical reordering of the arguments due to a non canonical sentence structure, Friederici and colleagues designed an fMRI experiment in which participants had to judge correctness of finite state grammar and phrase state grammar se ntences (Friederici et al., 2004 ). Finite state grammar is fully determined by local transitional probabilities between a finite number of items. An example of a sentence following finite conventional English subject verb object sequence. Phrase state grammar allows generation of phrase structures by rec ursive rules (i.e. embedded sentences). An example of a phrase whether it was the boy or the store that was robbed. Comparison of functional activations during processing of these two types of grammar reveals that violations of

PAGE 20

20 either type of grammar results in activation of frontal operculum (defined here as cortex located 2 cm posterior to posterior extent of BA 44, roughly corresponding to BA 6) whereas violations of only phrase Importantly, while both BA 44 and 45 show activation under this contract, the peak of activation was located in BA 44. Friederici a nd colleagues conclude that these patterns of activation suggest that frontal operculum (BA 6 as designated by the authors of the present manuscript (in particular BA 44) may be playing different roles in syntactic processing. Frontal oper culum (BA 6) may be involved in processing ungrammaticalities independent of the structure of a sentence. It may be checking an hand, specifically BA 44, activates exclusive ly during processing of structural hierarchies. Sentences with hierarchical structures possess greater syntactic load findings where higher semantic unification demands (specifically BA 45), sentences requiring more effort during syntactic unification also In addition to differential involvement in syntax and phonology, BA 44 was also shown to be part of the human mirror neuron system by support ing imagery of motion and action understanding processing (Binkofski et al., 2000; Rizzolatti & Craighero, 2004). In particular, when participants where asked to imagine their own hand and arm m determined by probabilistic cytoarchitectonic mapping (Binkofski et al., 2000; Amunts et al., 1999). Activation of this area during imagery of abstract movement may indicate that

PAGE 21

21 BA 44 is important for execution, recognition, and imagery of skilled forelimb movements. In language discourse, hand and arm movements typically serve as communicative gestures conveying linguistic information. In addition, BA 44 may be involved in recogn ition of abstract motor behavior that is relevant for communication. This hypothesis is further supported by studies that show BA 44 activation during observation, imagination, and imitation of forelimb movements (Grafton et al., 1996; Iacoboni et al., 199 9 ). A recent magnetoencephalograhy (MEG) study demonstrated (Nishitani and Hari, 2002). This activation may represent a link between sender and receiver of action related messa ges. Communicative disorders may involve partial impairment in neural mechanisms underlying comprehension and expression in discourse. Asperger syndrome patients show lower as compared with healthy volunteers during observation a nd execution of orofacial movements (Nishitani et al., 2004). Activation of ventral BA 44 is also observed during action imitation (Molnar Szakacs et al., 2005). Repeated transcranial magnetic stimulation (rTMS) applied over pars opercularis arrests imitat ion of finger movements while leaving execution of internally generated finger movements preserved (Heiser et al., 2003). Verbal working memory processing has also been attributed to BA 44 (Cabeza and Nyberg, 2000; Chein et al., 2002). Specifically, in the ir meta analysis of 30 functional imaging studies, Chein and colleagues showed that this area shows consistent activation in studies where task difficulty levels were manipulated (Chein et al., 2002).

PAGE 22

22 area is involved in various aspects of speech production and comprehension. These functions are not posterior ing roughly to BA 45 is differentially involved in retrieval of lexical semantic knowledge from declarative memory, semantic unification, regulation of top down linguistic processing involved in verbal retrieval, and processing of verbal fluency with high semantic load (Amunts et al., 2004; Hagoort et al., 2004; Ullman 2004). Posterior port control high level speech programming and production, processing of phonology and syntax, imitation, action understanding, and working memo ry ( Heim et al., 2008, 2009; Hagoort et al., 2004 ; Rizzolatti & Craighero, 2004; Cabeza and Nyberg, 2000). Thalamus and Its Role in Language Processing Our present knowledge about thalamic involvement in language processing stems from studies of language f unction by applying electrical stimulation during neurosurgery, from lesions effecting language deficits in patients with dominant thalamic infarcts or hemorrhages, and from functional neuroimaging studies. We will discuss evidence provided each of these t ypes of studies below. It is crucial to point out, however, that our understanding of the role of specific thalamic nuclei in language processing remains incomplete. It is premature to conclude which parts of the thalamic structures are or are not relevant to language. Most evidence supporting a thalamic role in language functions concerns more rostral nuclei (ventral anterior or ventral lateral) or posterior thalamus, namely pulvinar.

PAGE 23

23 Electrophysiological S tudies of Thalamic Language Function A number of studies applying electric stimulation to the dominant thalamus observed effects of this stimulation on language production and comprehension (Shaltenbrand 1965, 1975;Ojemann & Van Buren 1967; Ojemann, 1977). In particular, stimulation applied to the ventra l lateral (VL) nucleus of the thalamus was shown to produce slowing or silencing of speech (Schaltenbrand 1975). The patient may report that during stimulation he/she feels paralyzed and is unable to speak (Shaltenbrand 1965, 1975). Ojemann and Van Buren s howed that stimulation of this region also prolonged the expiratory phase of respiration during speech (Ojemann & Van Buren 1967). This effect, however, could be overcome voluntarily by the patient, suggesting that this region may be involved in coordinati on of respiration with speech production. Stimulation of VL nucleus of the thalamus during object naming typically results in anomia (the patient was able to speak but was not able to name the picture of an object presented to him/her) (Ojemann, 1977). Int erestingly, when the stimulation was applied more posteriorly, close to the anterior pulvinar patients produced omission and misnaming errors (Ojemann, 1983). Ventrolateral thalamus also seems to be involved in verbal short term memory (Johnson & Ojemann, 2000). When stimulation to this region is applied during encoding of to be remembered items patients tend to produce less errors during subsequent recall. However, if stimulation is applied during recall it results in more recall errors (Johnson & Ojemann 2000). Based on these results Johnson and Ojemann proposed nucleus enhances input of category specific verbal information into memory, while at the

PAGE 24

24 same time suppressing retrieval of other lexical items. Central portion of the ventral lateral nucleus is believed to integrate motor aspects of speech production, including respiration (Jo hnson & Ojemann, 2000). Stimulation of ventral anterior (VA) nucleus of the thalamus typically results in monosyllabic yells and exclamations (Schaltenbrand 1975). Specifically, Schaltenbrand observed that patients produced short sentences, syllable and wo rd repetitions (Schaltenbrand 1971, 1975). The author termed this spontaneous language output as recollection of the utterances although they remain conscious during th ese episodes. Schaltenbrand believes that this finding may indicate that the thalamus may be involved in timing of when speech should be released. In normal discourse, thalamus may be activating specific cortical networks that store contextually appropriat e lexical items. Pulvinar is another thalamic nucleus that has been implicated to be involved in language processing. Electric stimulation of this area (specifically the anterior superior pulvinar) results in anomia characterized by either misnamings or fa ilure to name an object while retaining ability to speak (Johnson and Ojemann, 2000). In addition, pulvinar stimulation has been shown to result in detrimental effects to short term verbal memory (Ojemann and Fedio, 1968). Pulvinar stimulation at the time of encoding and also during recall of lexical items results in greater amount of recall errors. These findings suggest that pulvinar may be involved in lexical retrieval mechanisms in language processing. Lesion Effect s on Thalamic Language Function Studie s of patients with ischemic or hemorrhagic lesions to the dominant thalamus provide additional knowledge of thalamic involvement in language. However,

PAGE 25

25 results of these studies and interpretations about roles of specific nuclei in language processing should be done with caution since lesions affect multiple thalamic nuclei as well as the white matter surrounding them. Lesions to the anterior portions of the thalamus involving VA and VL nuclei of the thalamus have been shown to result in language deficits inc luding difficulties with speech production, diminished vocal volume, aphasia, and category specific naming deficits (Graff Radford et al., 1985; Alexander and LoVerme, 1980; Nadeau & Crosson, 1997; Raymer et al., 1997). Raymer and colleagues presented a ca se of a patient who had suffered an infarction in the tuberothalamic artery territory, affecting VA and portions of VL nucleus. The patient presented with fluent but paraphasic speech (made a lot of substitutions during discourse), intact repetition and co mprehension. However, the patient had difficulties with oral and written picture naming, as well as with oral naming to auditory definition (Raymer et al., 1997; Crosson, 1999). These results indicate that the patient had impaired retrieval of items based on semantic information, while lexical processing remained intact. Moreover, the patient was able to match auditory and written forms of objects to correct pictures. This implies that phonologic, orthographic, and semantic systems were intact (Crosson 1999 ). Thus, damage to the anterior portions of the thalamus seems to impair mechanisms involved in lexical retrieval based on semantic information. Lesions to the dominant pulvinar also tend to result in language deficits (Van Buren 1975; Alexander and LoVerm e, 1980; Crosson et al., 1986, 1997). Specifically, patients with these lesions exhibit object naming difficulties, aphasia, and category specific naming deficits, while their speech remains fluent and comprehension and

PAGE 26

26 repetition are relatively spared. Cr osson and colleagues provide a detailed account of two such cases (Crosson et al., 1986, 1997). The first case presented in 1986 describes a patient who had suffered a hemorrhagic infarct of the left thalamus, specifically dorsal pulvinar and posterior two thirds of the lateral nucleus (confirmed postmortem). Examined acutely, the patient presented with fluent speech with frequent paraphasias and word finding difficulties. He made a significant amount of errors in comprehension of single words that could no t be attributed to poor working memory. Repetition was relatively preserved, however reading included significant amount of semantic substitutions. During a second assessment performed six weeks after the initial assessment most language functions had impr oved; however, the patient still suffered from word finding difficulties and experienced semantic substitutions during reading. Thus, although some language deficits resolved within the six week period following the hemorrhage, aphasia and semantic paraphr asias did not. As the authors pointed out the important in disruption of object naming during electric stimulation (Johnson and Ojemann, 2000). Thus convergent evidence suggests that pulvinar may be involved in lexical semantic processing. A second case of the dominant thalamic lesion involving the pulvinar describes a patient who presented with circumscribed anomia for medical items and conditions (Crosson et al., 1997) When examined chronically, the patient demonstrated a significant category specific naming deficit for man made medical items and medical conditions. Performance on all other semantic categories examined was normal. tems or conditions did not stem from unfamiliarity

PAGE 27

27 did not improve when phonemic cues were provided to ease recall. This indicates that the hemorrhage impaired lexical a ccess mechanisms making the patient unable to access medical items based on semantic information provided through auditory and visual channels. Because aphasia in this patient was circumscribed to a very specific semantic domain, the data suggested that th e pulvinar may be involved in recruiting particular cortical regions involved in accessing lexical items based on semantic content. Further evidence of this hypothesis is discussed in the next section of this chapter. Based on the evidence presented in les ion studies described in this manuscript we can deduce that patients suffering from aphasia following dominant thalamic lesion present with an overlapping set of language deficits regardless of the anterior posterior location of the lesion. In particular, core symptoms of thalamic aphasia typically include fluent but paraphasic speech, relatively spared repetition and comprehension, with word finding and object naming difficulties. It is therefore possible that thalamic nuclei implicated in language process ing (ventral anterior nucleus and pulvinar) may be part of a functional network subserving similar aspects of language processing. Damage to any portion of this network (either anterior or posterior thalamus) may then result in similar language deficits, a s observed in the cases described here. Neuroimaging E vidence of Thalamic Language Functions In addition to electric stimulation and lesion studies, neuroimaging literature provides further support for thalamic involvement in language processing. Kraut and colleagues implemented an interesting fMRI paradigm that provides valuable insights into the role of the thalamus in language (Kraut et al., 2002). Participants were

PAGE 28

28 presented with word pairs that either bind together to form a new object from features word pai rs binded together to form an object signal changes were observed in pre supplementary motor area (pre SMA), ventral occipito temporal gyri, and left thalamus. In contrast, on trials were word pairs did not form a new object cortical activation pattern rem ained the same but thalamic activation was not observed. Thus, thalamic activation was only present when the object was activated from its features. These findings may indicate that thalamus may be involved in coordinating cortical areas storing lexical re presentations based on semantic information. This mechanism may be supported by synchronization of activity in cortical regions storing semantically appropriate lexical information via an oscillating rhythm. This conjecture is at least partially supported by electroencephalography (EEG) recordings of the cortex and thalamus during semantic memory recall (Slotnick et al., 2002). A patient undergoing neurosurgery to alleviate seizures was presented with word pairs that either combined into a meaningful lexic al item or did not (for example, words Kraut and colleagues, where two words could bind together to activate a third lexically distinct but semantically related item, t he present task required the participant to judge whether a word pair was semantically related and could be formed directly from the two s observed that within 1 2 seconds post stimulus onset there was a significant drop in low frequency rhythm power within the thalamus and cortex. This event was followed by an increase in thalamic fast rhythm (20 60 Hz)

PAGE 29

29 with a corresponding localized incre ase in the same frequency band within the cortex. In cats, low frequency rhythms appear to be controlled by spatially widespread inhibitory projections from nucleus reticularis of the thalamus thought to keep the cortex in relative state of inhibition (Hun tsmann et al., 1999). Decrease in low frequency power may be indicative of global cortical disinhibition during semantic recall. This disinhibition is then followed by spatially specific increase in fast rhythm frequency that may be mediating feature bindi ng during recall by activating cortical zones that store contextually appropriate lexical items. This mechanism may therefore allow us to recognize an object from its features as well as aid in object naming tasks. Both of these linguistic tasks have been attributed to the pulvinar (Crosson et al., 1997; Kraut et al., 2002). Therefore, pulvinar may be involved in recruitment of specific cortical regions that store semantic information and/or corresponding lexical items during language comprehension and prod uction. Based on the evidence derived from electrical stimulation, lesion, and neuroimaging studies discussed above we can conclude that thalamus plays a crucial role in language production and comprehension mechanisms. Damage to thalamic nuclei frequently results in profound language deficits that may persist chronically. Clinical evaluations of thalamic aphasia patients indicate that regardless of the anterior posterior gradient of lesion location most patients exhibit characteristic language deficits tha t include fluent but paraphasic speech, and anomia with spared comprehension and repetition. This observation may indicate that thalamic nuclei implicated in language processing (VA and pulvinar) may be part of the same thalamocortical circuitry involved i n speech production and comprehension.

PAGE 30

30 Thalamocortical Lang uage Circuitry? The evidence discussed thus far in the present manuscript makes it clear that neural networks supporting language processing expand further than originally proposed by classical lan guage models (Binder et al., 1997; Assaf et al., 2006 ; Crosson, 2007). In addition, recent studies show that functions believed to be supported advanced. Specifically, Broc brain, but rather a sophisticated neural cluster involved in semantic, phonological, syntactic processing, action understanding and imitation and working memory (Amunts et al., 2004; Hagoort et al., 2004; Ullman 2004; Heim et al., 2008, 2009; Rizzolatti & Craighero, 2004; Cabeza and Nyberg, 2000). Similarly, thalamus is no longer believed to be a simple sensory relay station, but rather an important contributor to many cognitive processes including la nguage (Crosson et al., 1997; Johnson and Ojemann, 2000; Kraut et al., 2002). Our knowledge of the neural system and its organization suggests that substrates supporting similar functions are organized into structural as well as functional networks. Commun ication between these regions is plausibly enabled through shared circuitry to support processing of a given function. processes, and reason that they may therefore be structurally c onnected. When it comes to language production, we know that stimulation of the VA nucleus of the thalamus results in spontaneous speech, whereas lesions affecting this region often result in dysarthria (Shaltenbrand 1965, 1975;Graff Radford et al., 1985; Alexander and be involved in control of orofacial and laryngeal articulation during speech production,

PAGE 31

31 as well as speech programming (Horwitz et al., 2003). Lexical sema ntic retrieval also particular, object naming and category member generation were shown to activate VA nucleus and pulvinar lesions often result in semantic paraphasias and word finding difficulties further implicating the thalamus in lexical semantic search mechanisms (Raymer et al., 1997; Crosson et al., 1997, 1999). Based on this evidence, we believe that it would be such that they could be members of the same circuit. Our prior work (Ford et al., in of the thalamus. It could be plausible given the overwhelming evidence of its involvement in language processing, that the pulvinar may also be part of this structural network. Recent advances in structural imaging have introduced promising methods to inv estigate neural connectivity. Of particular interest is diffusion weighted magnetic resonance imaging (DW MRI). First introduced by Basser and colleagues, this technique measures MR signal decay attributed to displacement of hydrogen protons in the directi on of an applied diffusion gradient (Basser et al., 1994). DW MRI assumes that maximum diffusion displacement will correspond to longitudinal axis of a neural bundle offering the path of least resistance to the flow of water (Behrens et al., 2003). Followi ng this assumption a number of modeling methods, collectively known as diffusion tractography, were developed to trace white matter fiber bundles. Application of this technique has provided many valuable insights into structural organization of

PAGE 32

32 functional networks within the brain and diffusion tractography remains the only noninvasive method presently available to study structural connectivity in vivo. One of the first papers to showcase diffusion tractography capabilities was a study by Catani and colleag ues demonstrating virtual dissection of major white matter faciculi in the human brain (Catani et al., 2008). Results of the study show three dimensional visualization of white matter pathways, including superior and inferior branches of the longitudinal f asciculus, cingulum, and fornix, faithful to the classical descriptions of these tracts derived from postmortem studies. In addition to providing information about individual trajectories of known anatomical pathways, tractography also offers insights abou t the presence of networks that have not been previously described. Specifically, Catani and colleagues showed that the arcuate fasciculus (Catani et al., 2005). The indi rect pathway runs parallel and lateral to the classical (direct) arcuate fasciculus connecting temporal and inferior frontal lobes. Two segments territory with the inferio r parietal lobe, and a posterior segment connecting the inferior of structural organization of the arcuate fasciculus, Catani and colleagues also examined hemispheric asymmetries of this pathway (Catani et al., 2006). They found more than half of participants and distributed bilaterally in less than a third of participants. Leftwar d lateralization has been attributed to development of language processing in our species (Geschwind and Levitsky, 1968). Extensive connectivity

PAGE 33

33 between frontal and temporal language areas in the left hemisphere may support more efficient and computationa lly intense linguistic processing. Further investigation of the arcuate fasciculus revealed that this pathway has two distinct termination regions within the temporal lobe presumably supporting different language functions (Glasser & Rilling, 2008). In par ticular, arcuate fasciculus projections from the inferior frontal cortex terminate in the posterior superior temporal gyrus (STG) and in the middle temporal gyrus (MTG). These regions within the left hemisphere overlap with previously published fMRI activa tions associated with phonological and lexical semantic processing, respectively (Glasser & Rilling, 2008). Thus, tractography provides valuable information about structural organization of functional networks supporting language processing. In addition, this network approach may be extended to explain aphasia syndromes. The most frequently reported aphasia associated with interruption of pathways connecting inferior frontal and temporal cortical regions is conduction aphasia. Patients with this syndrome p resent with an inability to repeat lexical items while their comprehension and spontaneous speech are largely preserved. Glasser and Rilling proposed that conduction aphasia symptoms may be explained by interruption of the pathways connecting inferior fron tal cortical areas with the STG. This portion of the arcuate fasciculus is thought to support phonological processing and since the MTG portion of the pathway remains comprehension remains largely unaffected. Another type of aphasia resulting from damage to the frontotemporal network is transcortical motor aphasia. Patients with this disorder present with limited spontaneous speech, i mpaired naming and intact

PAGE 34

34 repetition. Glasser and Rilling postulate that these findings could explain these language deficits by interruption of the pathways connecting inferior frontal cortex and spontaneous speech. In addition, preserved repetition could result when the STG connections with the frontal lobe are spared enabling intact phonological processing (Glasser and Rilling, 2008). In addition to the arcuate fasciculus, tractography studies identified additional and temporal lobes (Makris et al., 2009). These pathways are analogous to the ventral and dorsal pathways found in non human primates and have been la fasciculus (SLF) with posterior temporal lobe (Brodmann area 40). The ventral (EC) and the uncinate fasciculus (UF) with anterior STG (Parker et al., 2005; Frey et al., 2008). These pathways may support functional networks involved in syntact ic processing (Friederici et al., 2006). Specifically, the ventral pathways connect the frontal operculum and anterior STG via UF pathway (Friederici et al., 2006). Friederici and colleagues previously implicated the frontal operculum to be involved in loc al phrase structure building (finite state grammar) (Friederici, 2004). In addition, anterior STG was also shown to be involved in similar syntactic processing (Friederici, 2002). Therefore, the ventral pathways connecting these areas may support a network involved in local phrase syntactical processing (Friederici 2006, 2009). On the other hand, the dorsal

PAGE 35

35 within the left hemisphere was shown to support processing of hierarch ical structure processing (phrase state grammar), while posterior STG activates during processing of syntactically complex sentences (Friederici, 2004; Bornkessel et al., 2005). Diffusion tractography studies also identified neuroanatomical differences bet ween human and non human primates implicated in the evolution of language (Rilling et al., 2008) by examining white matter pathways in humans, chimpanzees, and macaques. Rilling and colleagues demonstrated that the dorsal pathway consisting of the AF and S LF fibers is very prominent in humans and terminates predominantly in the temporal lobe. Termination regions within the temporal lobe include the STG and MTG previously implicated in phonological and lexical semantic processing respectively (Glasser and Ri lling, 2008). In contrast, the dorsal pathway in chimpanzees terminates predominantly in the parietal lobe and its terminations within the STG are less prominent than that of humans. Of the four chimpanzee datasets examined, only one showed dorsal pathway termination in the MTG (Riling et al., 2008). In addition, dorsal pathways in macaque datasets were much weaker than those of both chimpanzees and humans and connectivity between frontal and temporal lobes was dominated by the ventral pathways running in t he vicinity of the EC. These results demonstrate prominent inter species differences in organization of structural connectivity between frontal and temporal lobes. Human data suggests that the dorsal pathways connecting these regions are very prominent and more robust than the ventral pathways. In addition, termination points within the temporal lobe are much more widespread, including regions within STG and MTG, which appears to be a unique feature of the human anatomy. Further investigation of strength of connectivity between ventral lateral

PAGE 36

36 prefrontal cortex and temporal lobe indicates that MTG has a higher probability of connectivity with pars triangularis and pars orbitalis (BA45, 47) than pars opercularis (BA44). Rilling and colleagues hypothesize that this network may support lexical semantic processing by transmitting word meaning from the MTG to the inferior parietal lobe and BA 45/47 to enhance sentence comprehension and construction (Rilling et al., 2008). In addition, other scholars propose that s yntactic processing may be supported by a structural network connecting BA 44 and posterior STS/STG (Friederici 2009). These regions have been shown to be involved in syntactic unification associated with comprehension of sentences with high syntactic dema nd (Friederici et al., 2004). Processing of syntactically complex sentences (or phrase state grammar) appears to be unique to human language faculty (Fitch & Hauser, 2004). It is therefore not surprising that this pathway is robust in human data and weak i n non human primates (Rilling et al., 2008). In conclusion, the evidence gathered from the above mentioned tractography studies indicates that the dorsal pathways connecting frontal and temporal lobes underwent significant structural changes in order to s upport advanced linguistic functions in the course of the evolution of language. In addition to allowing investigation of cortical connectivity patterns, diffusion tractography has also been used to investigate connections between the cortex and subcortica l structures (Behrens et al., 2003; Johansen Berg et al., 2005; Draganski et al., 2008). Behrens and colleagues were among the first to employ diffusion tractography to investigate cortical connectivity with subcortical regions (Behrens et al., 2003). In this study, connectivity between prefrontal, temporal, parietal, and occipital cortical regions and the thalamus was investigated and quantified. The authors showed

PAGE 37

37 that connectivity patterns between the cortical regions examined and the thalamus followed previously described corticothalamocortical projections in macaques derived using tracer injections (Behrens et al., 2003). In addition, the authors parcellated the thalamus based on connectivity strength with the cortical zones and the resulting parcellat ions closely resembled known nuclear divisions of the thalamus (Behrens et al., 2003). In a follow up study, Johansen Berg and colleagues correlated resulting thalamic parcellations with histologically defined thalamic nuclear volumes and thalamic activati ons derived from functional MRI studies (Johansen Berg et al., 2005). Thalamic parcellations defined using diffusion tractography correlated with both location and volumes of histological thalamic nuclei. Portions of the thalamus showing high probability o f connection to sensorimotor and premotor areas as determined by tractography overlapped with clusters of thalamic activations during fMRI motor tasks. Thalamic activations elicited during executive and memory tasks fell within thalamic regions showing hi gh probability of connection with the prefrontal cortex (Johansen Berg, et al., 2005). Another tractography study examined connectivity between a single thalamic nucleus and multiple regions within the prefrontal cortex in macaques and humans (Klein et a. 2010). Klein and colleagues examined connectivity between mediodorsal (MD) thalamus and regions within the frontal lobe including dorsal prefrontal cortex (DPC), dorsolateral prefrontal cortex (DLPC), and ventrolateral prefrontal cortex (VLPFC). The resu lts show high fidelity between tractography based connectivity patterns and tracer studies in macaque datasets. In addition, in both macaque and human datasets each cortical region showed a distinct peak connectivity location within

PAGE 38

38 the mediolateral nucleu s of the thalamus indicating that regions within MD exhibit preferential connectivity with the frontal lobe (Klein, et al., 2010). A number of tractography studies also examined cortical connectivity with the basal ganglia extending non human primate trac er literature to in vivo human data (Draganski et al., 2008). Draganski and colleagues parcellated basal ganglia based on connectivity strength with medial, dorsolateral, and orbitofrontal prefrontal cortex, as well as premotor and motor cortex (Draganski et al., 2008). Their findings show that the more anterior portions of the caudate, putamen and the pallidum exhibit stronger connections with medial/orbital and dorsolateral prefrontal cortex, while posterior portions of these subcortical structures conne ct with premotor and motor cortex. These results agree with prior findings resulting from tracer injections in non human primates. Interestingly, although most animal studies suggest that the basal ganglia circuits (channels) remain segregated throughout t he basal ganglia, Draganski and colleagues show striatal regions with overlapping connections to different cortical areas (Draganski et al., 2008). In particular, connections from obritofrontal, medial and dorsolateral prefrontal cortex converge within sim ilar regions within the striatum. These results suggest that the striatum may be mediating reward based learning by integrating reward representation from orbito striatal network with optimal contextual behavioral output within the dorsolateral prefrontal striatal circuitry. Data from our lab also supports existence of integrative basal ganglia circuitry. findings indicate that projections from both pars opercularis and par s triangularis project to an overlapping region within the anterior one third of the putamen and the ventral

PAGE 39

39 anterior nucleus of the thalamus (Ford et al., in review). Convergence of these inputs may serve particular functional significance for language pr ocessing. We hypothesize that once the most appropriate semantic response has been enhanced through interactions within the pars triangularis basal ganglia circuitry, this selection is relayed to the pars opercularis network to strengthen activation for th e corresponding lexical phonological representation. This ensures that the desired semantic response is articulated using appropriate phonemes during discourse. The overlap of these circuits at the level of the anterior putamen and downstream portions of t he loop, including the ventral anterior thalamus, may ensure that corresponding semantic and lexical phonological representations are fine tuned during word selection. Tractography studies examined above are just a handful of the evidence demonstrating uti lity of diffusion tractography in the study of structural connectivity in vivo. However, this technique is not without its limitations. Incomplete signal modeling and inherent noise associated with imaging biological systems introduce uncertainty in both a cquisition and reconstruction of diffusion weighted data and, consequently, in tractography as well (Behrens et al., 2003, 2007). In addition, as Jbabdi and Johansen Berg point out, none of the current tractography methods allow users to establish formal s tatistical testing procedures to determine whether resulting tracts reflect underlying anatomy given the data (Jbabdi & Johansen Berg, 2011). This drawback is a result of the fact that in order to test against the null hypothesis (i.e. there are no tracts connecting A and B), we need a model quantifying the null diffusion distribution (Morris et al., 2008). For one possible model of a null distribution, each voxel within the inferred tract might be represented by a random diffusion orientation field, implyi ng that diffusion

PAGE 40

40 along any direction is equally likely. Although this may seem plausible for the null diffusion distribution, when we concatenate null distributions of all of the voxels along an inferred tract of any appreciable length, then probability o f reaching B from A will be vanishingly small. Thus, almost any pathway a few voxels long will be able to pass this significance criterion, thus defeating the purpose of statistical testing (Jbabdi & Jonansen Berg, 2011). It is not news, in other words, th at diffusion within brain tissue differs from diffusion within bulk water. It is clear that further theoretical development is necessary to enable representative statistical testing of tractography results. Meanwhile, in order to ensure selectivity in path way reconstruction users should use a prior anatomical knowledge to guide tracking algorithms. This procedure can be accomplished by imposing additional constraints on tracking results. The user may specify an inclusion mask if prior anatomical evidence ba sed on animal tracer studies, human dissection, or lesion addition, exclusion masks can be added to ensure that only the relevant pathways are being traced. Skillfu l application of these selectivity criteria in addition to careful examination of the reconstructed pathways and their plausibility given prior anatomical and functional data offers a unique opportunity to investigate structural connectivity in vivo The Present Study The study described in this manuscript investigates structural connectivity understanding of this connectivity will provide valuable insights into thalamic involveme nt in language processing. As discussed above, hemorrhage of the dominant

PAGE 41

41 thalamus frequently results in widespread language processing deficits collectively known as thalamic aphasia. Core symptoms of thalamic aphasia include difficulties in object naming fluent but paraphasic speech, and relatively well preserved repetition and comprehension. These symptoms are observed in patients who suffered thalamic infarcts affecting anterior thalamus (specifically the ventral anterior thalamus), as well as in patie nts with lesions to posterior thalamus (pulvinar). The degree of similarity of clinical characteristics between thalamic aphasia resulting from damage to anterior thalamus and to the pulvinar suggests that these thalamic nuclei may be part of a single thal amocortical network involved in language processing. Based on this evidence we hypothesize that both ventral anterior nucleus and the pulvinar share direct structural origin individual subcortical region of interest. In addition, in order to trace pathways connect strength for each of these pathways by obtaining quantitative measures of the inferred tract v olumes and edge weights, as explained below. The present study also investigates effects of previously reported thalamic thalamic lesions affecting either anterior or post erior portions of the thalamus will in both cases interrupt this network. Specifically, we suppose that anterior lesions will interrupt

PAGE 42

42 d the pulvinar. Conversely, posterior thalamic nucleus. The best procedure to examine t hese effects ideally would be to analyze corticothalamic connectivity in a group of patients with thalamic aphasia. However, patients with lesions segregated to the dominant thalamus resulting in language deficits are very rare. Future studies designed to examine the correlation between clinical manifestations due to dominant thalamic hemorrhage and damage to the pathways extension of the present work. The current project instead aims to evaluate the extent of interruption of these pathways using published examples of lesions following a dominant thalamic hemorrhage. To do so, we introduce lesion models derived from the two well documented clinical cases of thalamic aphasia mentioned earlier in this document (Crosson, et al., 1997; Raymer, 1997). The first case is the patient presented by Raymer and colleagues (referred to here as patient D) who suffered an infarction in the tuberothalamic artery territory resulting in a le sion affecting ventral anterior and ventral lateral nuclei of the thalamus. Examined acutely, the patient D presented with intact lexical processing, but had difficulties with retrieving items from the phonological output lexicon based on semantic informat ion and tended to produce semantic errors (i.e. the patient had difficulty naming pictures and often substituted items she could not name with semantically related items). The second case is the patient presented by Crosson and colleagues (referred to here as patient C), examined two months post onset, who had suffered a lesion affecting portions of the pulvinar, posterior limb of the

PAGE 43

43 internal capsule, and centromedial nucleus of the thalamus. Patient C presented with naming difficulties related to his medi cal condition or items around the hospital. Unlike patient D, patient C did not produce semantic errors, but rather was unable to produce a response within medical items category. This observation may indicate that patient C in addition to having deficits in semantic processing also suffered deficits to lexical processing or at the interface between these processing levels. indicate a more profound language system impairment than that of patient C, affecting only the medical items category chronically, it is important to note that language deficits in both patients were overlapping (i.e. fluent speech with difficulties accessing semantic information) while the lesion locations were not. Patient D presented with a lesion affecting the anterior extent of the thalamus (ventral anterior and ventral lateral nuclei), while patient C had a lesion affecting the posterior thalamus (the pulvinar). Given the similarities in language deficits a nd differences in lesion location within the thalamus in these two patients, modeling these lesions would provide a good measure to evaluate understanding of thalamic connectivity with cortical language substrates would provide valuable insights into function and organization of networks supporting language processing in the brain. In particular, an integrative approach that would regard thalamic nuclei implicated in language proce ssing as being part if an overlapping thalamocortical network could promote efforts to devise new rehabilitation treatments that are not tied strongly to lesion location.

PAGE 44

44 CHAPTER 2 METHODS Participants s area and the thalamus we acquired neuroimaging data in ten healthy volunteers (5M, average age=26.6 years, st. dev. =4.79 years). Table 2 1 lists demographics of our participants. All of the participants were right handed, native English speakers, with n o known neurological disorder were recruited. Written informed consent was obtained from all participants in compliance with Institutional Review Board guidelines of the University of Florida and airs Medical Center. Image Acquisition and Processing Acquisition Parameters We collected anatomical T1 weighted and DWI using the following parameters. Structural MP RAGE T1 weighted scans were collected with 130 1.0 mm sagittal slices, FOV=240 mm (AP) 180 mm (FH), matrix= 256X192, TR= 9.90 ms, TE= 4.60 ms, FA= 8, voxel size= 1.0 mmX0.94 mmX0.94 mm. Diffusion weighted images were acquired using single shot spin echo echo planar imaging (EPI) with 602.0 mm axial slices (no gap), FOV= 224 mm (AP)224 mm (RL), matrix= 112112, TR= 9509 ms, TE= 55 ms, FA=90, voxel size= 2.02.02.0 mm, and time of acquisition= 5 min 42 s. The diffusion weighting gradients were isotropically distributed over a sphere using a 64 direction acquisition scheme with b=1000 s/mm2. Six low b value (b=100 s/mm2) volumes were also collected. Two volumes with no diffusion weighting (b=0) were also acquired using these parameters.

PAGE 45

45 Image Processing To investigate structural connectivity of propose to use a newly developed high angular resolution diffusion weighted imaging (HARDI) tractography (Jian et al., 2007). Our tracking technique infers local fiber orientation by estimating probability associated w ith each direction based on the diffusion properties of the tissue. This method allows us to track crossing and branching fibers, which makes the current algorithm superior to traditional streamline tracking techniques based on the diffusion tensor models (Jian et al., 2007; Basser et al., 1994). Specifically, our algorithm estimates the displacement probability function in each brain voxel using the Method of Weishart (Jian et al., 2007) implemented using an in house software package written in IDL (Exelis Visual Information Systems, Boulder, CO). Then, the local maxima of the probability function are located using a gradient ascent method with multiple restarts. The found maxima are saved as input to the tracking algorithm. Tractography is performed by se eding every voxel in the brain with a 4x4x4 sub voxel grid of evenly spaced 64 seed points. For each seed point, one streamline is launched, bidirectionally, for each estimated maximum contained in that voxel. Each streamline front is propagated by steppin g 0.25 voxel width in the direction of the maximum that is most inline with the streamline's present direction of travel. In order to prevent streamlines from looping back we employ a limiting angular deviation of the track of 50 degrees. If the estimated track exceeds this threshold the streamline is stopped and that specific course is removed from inclusion. Many tracking algorithms explicitly exclude tracks coursing through grey matter by use of fractional anisotropy (FA) thresholding. Because our region s of interest include grey matter structures we did

PAGE 46

46 not apply an FA threshold in order to allow us to track pathways through grey matter regions. Regions of Interest Cortical regions of interest To trace white matter pathways of interest we identified two cortical regions corresponding to pars opercularis and pars triangularis and two subcoritical regions corresponding to the ventral anterior nucleus of the thalamus and the pulvinar. Cortical masks of pars triangularis and pars opercularis were drawn on the T1 weighted images and registered to diffusion space using FSL FLIRT module (Jenkinson and Smith 2001). The lateral most sagittal slice of the frontal cortex of the skull stripped T1 weighted scan was used as the lateral border of the cortical masks. The medial border was defined by the first sagittal slice traversing the insular cortex. The dorsal border of the masks was defined by the inferior frontal sulcus, while by the ventral border was be the Sylvian fissure. The anterior border of the pars triangul aris mask was defined by a coronal plane through the anterior margin of the anterior horizontal ramus of the Sylvian fissure, and its posterior border was defined by the anterior ascending ramus of the Sylvian fissure. The anterior border of the pars operc ularis mask was drawn by following the posterior border of the pars triangularis mask leaving one voxel distance between the two masks to ensure that they are non overlapping. The posterior border of pars opercularis mask was defined by the inferior precen tral sulcus. Subcortical Regions of Interest To create masks of ventral anterior nucleus and pulvinar we used a previously developed three dimensional deformable brain atlas for direct brain stimulation (DBS)

PAGE 47

47 targeting (Sudhyadhom et al., 2012). In additio n, we used the Schaltenbrand and Wahren atlas (1959) to adjust the borders of masks generated by the atlas. Exclusion masks For each dataset we created an exclusion mask to ensure that we do not include extraneous pathways into our calculations. The exclus ion mask consisted of four planes placed as follows. First we used a mid sagittal exclusion plane to ensure that only pathways within the left hemisphere are being traced. Next a coronal plane was placed two slices posteriorly to the pulvinar mask to exclu de any pathways coursing through to posterior perisylvian, posterior parietal, and occipital lobes. The third plane was an axial plane placed two slices below the pulvinar mask to exclude pathways within inferior temporal lobe and brainstem. The last exclu sion plane was an axial plane placed two slices above the corpus callosum. Thalamic lesion models To create our lesion models corresponding to the two clinical cases discussed above, we used lesion reconstructions from MRI to axial planes of the Schaltenbr and and Wahren atlas (1959) rendered in the original articles (Crosson, et al., 1997; Raymer, 1997). We created masks of the lesions in the MNI atlas space (Figure 2 1a, 2 1b) and then transformed them to the native acquisition space of each participant us ing the FNIRT non linear registration tool implemented in FSL (Andersson et al., 2007a, 2007b; Rueckert et al., 1999). Tractography analysis In order to trace pathways between cortical and subcortical regions of interest we intersected whole brain tractog space with their individually created pars opercularis and pars triangularis masks to infer

PAGE 48

48 fiber bundles passing through each of these regions. Next, we intersected the resulting tracts with ventral a nterior nucleus and pulvinar masks to compute pathways connecting pars opercularis and pars triagularis with each subcortical region. When tracing pathways between pars opercularis and pars triangularis and ventral anterior nucleus we applied two exclusion masks in order to ensure that our tracing approach infers only the pathways of interest. The first exclusion mask consisted of four exclusion planes as described above. In addition, we also used the pulvinar mask for exclusion to ensure that pathways betw than passing through it on the way to pulvinar. An important consideration of the application of these exclusion masks is that they eliminate all potential pathways pulvinar may give off collaterals to the ventral anterior nucleus as they pass through this nucleus on the way to the pulvinar. These pathways, however, would be represented by anterior nucleus and pulvinar (also referred to as fibers en passage below). Pathways between pars opercularis and pars triangularis and the pulvinar were traced in a similar fashion. We intersected pathways from pars opercularis and pars triangularis with the pulvinar mask and applied exclusion masks. The first mask was the four plane exclusion mask as before and then, in this case, the second exclusion mask was the ventral anterior nucleus mask. Ventral anterior nucleus mask was used as an and did not enter via ventral anterior nucleus.

PAGE 49

49 anterior nucleus, and the pulvinar. These fibers are believed to be penetrating the ventral anterior nucleus on their way to the pulvinar. In order to trace these pathways we first intersected whole brain tractography results with the ventral anterior nucleus mask. Next, the resulting pathways were intersected with pars opercularis (or pars triangularis) and pulvinar masks. This was done to ensure that the pathways course throu gh all three regions of interest (pars opercularis (or pars triangularis), ventral anterior nucleus, and the pulvinar). In addition, a four plane exclusion mask described above was applied to exclude extraneous pathways. Quantitative tractography measures After applying the above stated tractography analysis we generated quantitative measures describing the resulting pathways. These measures were tract volume and edge weight. Tract volume represents the number of voxels occupied by each of the pathways mult iplied by the volume of each voxel (measured in mm 3 ). Edge weight represents strength of connectivity of each pathway and is derived as an application of graph theory (Colon Perez et al., 2012) Edge weight is a unitless measure, or scalar, calculated i ndependent of the acquisition resolution and seeding density. This measure is computed for fiber tracts inferred between two regions of interest (A i and A j in the Equation 1 below). Each region of interest is represented as a node and each pathway connecting a pair of nodes represents an edge. In Equation 1 summation of the right represents the sum of all of the voxels (p=1 to p=P) that belong to a given ed ge (m). Summation on the left adds all of these voxels for all edges (m=1 to m=M) connecting regions A i and A j :

PAGE 50

50 (1) The resulting double sum represents the total number of voxels comprising pathways between two regions of interest, scaled by the total len gth of these pathways (l( f p,m ). Terms to the left of the double sum represent additional scaling factors. A i and A j are surface areas of the two regions of interest between which we are tracing the pathways. Scaling by the average of these surface areas a llows us to directly compare edge weight values for pathways generated using different pairs of regions of interest. This ensures that pathways delineated between larger regions of interest will not have larger edge weight values. V voxel in Equation 1 rep resents voxel acquisition resolution and P voxel represents the number of seeds per voxel used during the tractography analysis. In this manuscript we used 64 seeds per voxel. Application of the scaling factor (V voxel /P voxel ) ensures that the edge weight va lue is independent of the acquisition resolution and seed density. This approach allows us to compare edge weight values for data acquired using different acquisition resolutions and analyzed using different number of seeds per voxel.

PAGE 51

51 Table 2 1. Participant Demographics. Participant Demographics Participant ID Age Gender 1 28 M 2 27 M 3 32 M 4 36 F 5 29 F 6 23 F 7 23 M 8 21 F 9 24 M 10 22 F Average 26.50 St. Dev. 4.84 Figure 2 1. Lesion models rendered from two clinical cases (Raymer, et al.,1997; Crosson, et al., 1997). Left: (a1 and a2) lesion masks affecting anterior thalamus taken from the original article (a1, Raymer et al.1997) and reconstructed in the MNI space (a2). Right: (b1 and b2) lesion masks affecting posterior t halamus taken from the original article (b1, Crosson et al., 1997) and reconstructed in the MNI space (b2).

PAGE 52

52 CHAPTER 3 Following tractography analysis described in Chapter 2 of this manuscript we traced pathways conne pulvinar. A detailed description of the results of these analyses is provided below. Ventral Anterior Nucleus Pathways ventral anterior nucleus of the thalamus we intersected pathways originating in pars opercularis and pars triangularis with the ventral anterior nucleus, and applied pulvinar mask and the four plane mask as exclusion masks. Table 3 2 represents the resulti ng tract volumes of these fiber bundles for each of our ten participants. The average tract volume for pathways connecting pars opercularis and ventral anterior nucleus is 2879. 5 mm 3 (st. dev.=2078 mm 3 ). The average tract volume for pathways connecting pars triangularis and ventral anterior nucleus is 3606.5 mm 3 (st. dev.=2353.13 mm 3 ). We note that for most of our participants pathways between pars opercularis and ventral anterior nucleus are smaller in volume than those between pars triangularis and ventral anterior nucleus (with the exception of participants 1,4,6), although this difference was not statistically significant (p=0.23, tstat= 1.30, df=9). Table 3 3 represents edge weights for pathways con anterior nucleus in our ten participants. The average edge weight value for pathways connecting pars opercularis and ventral anterior nucleus is 5.43x10 4 (st. dev.=1.17x10 3 ) and the average edge weight for pathways connec ting pars triangularis and ventral anterior nucleus is 1.84x10 3 (st. dev.= 2.24x10 3 ). Although for most participants edge weights for pathways connecting pars triangularis with ventral anterior nucleus were

PAGE 53

53 larger than for pathways connecting pars opercu laris with ventral anterior nucleus this trend did not reach statistical significance (p=0.12, tstat= 1.72, df=9). We believe that smaller tract volumes and edge weights may reflect additional complexity of tracing pathways originating in pars opercularis. In particular, this region is located posterior to pars triangularis. Pathways originating in pars opercularis must therefore travel a more circuitous route around the circular sulcus to reach the thalamus. This added complexity in pathway trajectory requ ires larger turning angles in order to infer the underlying fiber architecture. Pathways with large turning angles are typically harder to trace using diffusion tractography due to the loop check restrictions used to ensure that pathways do not loop back o nto themselves. Thus, smaller tract volumes for pathways connecting pars opercularis and ventral anterior nucleus may represent underlying structural complexity and resulting tracking difficulty. Another potential explanation for smaller tract volumes for pathways connecting pars opercularis and ventral anterior nucleus could be smaller surface area of the pars opercularis region of interest as compared with the pars triangularis mask. Table 3 1 represents surface area measures for the two cortical as well as the two subcortical masks used in this analysis. We note that the average surface area for the pars opercularis mask is 1726.23 mm2 (st. dev.=310.5 mm2), while the average surface area for the pars triangularis mask is 1629.11 mm2 (st. dev.=491.25 mm2) When we performed a paired t test to determine whether there is a statistically significant difference between surface areas of the two cortical regions of interest for all of our participants we did not find significance (paired two tailed t test: p=0.5 7, tstat=0.59, df=9). This finding indicates that size of the cortical region of interest mask is not related

PAGE 54

54 to smaller tract volumes and edge weight values for pathways connecting pars opercularis and ventral anterior nucleus. Given these results, we bel ieve that the anatomical location of pars opercularis in reference to pars triangularis presents additional complexity for diffusion tractography resulting is smaller pathways. Figure 3 1 depicts projections between pars opercularis (a) and pars triangular is (b) and ventral anterior nucleus in a representative dataset (participant 1). Figure 3 2 depicts these pathways in all ten participants. Color gradient of the fibers represents local fiber orientation: red left right, green anterior posterior, and blue superior inferior. Projections from both pars triangularis and pars opercularis course medi ally, passing over and around the anterior superior portion of circular sulcus, and after passing through the insula, follow a nearly ninety degree turn to then trav el posterior ly towards the thalamus Projections from pars triangularis and pars opercularis enter ventral anterior nucleus in all participants. Results from most of our participants show that the pathways project to the anterior medial portion of the v entral anterior nucleus (participants 1, 2, 3, 4, 7, 8, 9, 10). A few participants deviate from this trend. In particular, results for participant 5 show the anterior l ateral portion (Figure 3 2 (5a, 5b)). Another pattern of connectivity can be seen in Figure 3 2 (2b) depicting pathways connecting pars triangularis and ventral anterior nucleus. In this figure we see that most of the pathways enter the anterior medial po rtion of the nucleus, while a second smaller portion of the pathway descends down into the nucleus at a more lateral and superior location. Lateral superior portion of the nucleus is also the target of projections for

PAGE 55

55 participant 6. In particular, we obser ve that in Figure 3 2 (6a, 6b) the pathways from pars opercularis project to the superior lateral extent of the nucleus, similar to that in Figure 3 2 (2b). In addition, we note that pathways connecting pars opercularis (tract volume=445 mm 3 edge weight=7 .35x10 7 ) and pars triangularis (tract volume=471 mm 3 edge weight=2.29x10 6 ) with ventral anterior nucleus for participant 10 are much smaller than those of the other nine participants. After re examining this dataset for artifacts that could explain poor tracking results we did not find any peculiarities in the quality of the data. One potential explanation of the tracking results for this participant could be differences in individual anatomy. In particular, the turning angles that the pathways travel ma y be generally too large to fit our loop check restrictions for this participant and as a result only a few streams within the pathways can be traced. A closer examination of pathways connecting pars opercularis/triangularis and ventral anterior nucleus al so reveals that in some participants a small portion these cortico thalamic or thalamocortical pathways travel around ventral anterior nucleus and enter more posterior extents of this region of interest. In particular, for participants 1, 3, 5, and 7 we no te that pathways connecting pars opercularis and ventral anterior nucleus exhibit this trend. A small portion of the pathways travels medial and posterior to the nucleus and enters the posterior border ventral anterior thalamus. Similarly, for participants 1, 2,3, 4, 5, 7, 8, 9 this trends is also present for pathways connecting pars triangularis and ventral anterior nucleus. Pulvinar Pathways originating in pars opercularis and pars triangularis with the pulvinar mask and applied

PAGE 56

56 both ventral anterior nucleus mask and the four plane mask as exclusions. Table 3 4 represents tract volumes of the resulting fiber bundles. The average tract volume for pathways connect ing pars opercularis and pulvinar is 3725.5 mm 3 (st. dev.=2620.33 mm 3 ). The average tract volume for fibers connecting pars triangularis and pulvinar is 4964.10 mm 3 (st.dev.=2254.78 mm 3 ). In contrast to nucleus of the thalamus, pathways connecting pars opercularis and pulvinar are statistically significantly smaller in volume than those connecting pars triangularis and pulvinar (though participants 1,4,6 are exceptions to this significant difference) (pa ired two tailed t test: p=0.04, tstat= 2.44, df=9). In addition, we note that tract volumes of the pathways connecting nucleus pathways (pars opercularis pulvinar tract volu me=3725.5 mm 3 pars triangularis pulvinar tract volume=4964.10 mm 3 pars opercularis ventral anterior nucleus tract volume=2879. 5 mm 3 pars triangularis ventral anterior nucleus tract volume=3606.5 mm 3 ). This trend however, did not reach statistical signi ficance when we apply a two tailed paired t test to tract volume distributions (pars opercularis p=0.23, tstat= 1.28, df=9; pars triangularis p=0.07, tstat= 2.05, df=9). Table 3 and pulvi nar. The average edge weight for pathways connecting pars opercularis and pulvinar is 6.46x10 4 (st. dev.=1.68x10 3 ). The average edge weight for pathways connecting pars triangularis and pulvinar is 1.33x10 3 (st. dev =1.54x10 3 ). The edge weight values f or pathways connecting pars triangularis with pulvinar are larger than those for pathways connecting pars opercularis and pulvinar in most of our participants,

PAGE 57

57 but this trend does not reach significance (paired two tailed t test: p=0.35, tstat= 0.98, df=9) anterior nucleus (paired two tailed t test: p=0.62, tstat= 0.52, df=9 for pathways connecting p ars opercularis with ventral anterior nucleus versus pathways connecting pars opercularis with pulvinar; p=0.49, tstat=0.71, df=9 for pathways connecting pars triangularis with ventral anterior nucleus versus pathways connecting pars triangularis with pulv inar). Figure 3 3 depicts projections between pars opercularis (a) and pars triangularis (b) and pulvinar in a representative dataset (participant 1). Figure 3 4 depicts these pathways in all ten participants. Color gradient of the fibers represent local f iber orientation as described above. particular, pathways originating in pars opercularis or pars triangularis travel medially around the circular sulcus and through the insular cortex. From there the pathways bend at a nearly 90 degree angle and course posteriorly towards the pulvinar. We note ea terminate in the anterior superior portion of the pulvinar similar in location to the area that has been previously implicated to be involved in object naming processing ( Johnson and Ojemann, 2000; Crosson et al., 1986). s area Ventral Anterior Nucleus Pulvinar Projections passing through the ventral anterior nucleus, we first traced the pathways originating in

PAGE 58

58 ventral anterior nucleus and then intersected resulting trac opercularis and pars triangularis separately) and pulvinar masks. Resulting tracts contained inferred fiber bundles connecting all three regions of interest. Four plane exclusion mask was also applied to exclude any extraneous pa thways. Table 3 6 represents tract volumes for these pathways for each of our participants. The average tract volume for pathways connecting pars opercularis with ventral anterior nucleus and pulvinar is 971.60 mm3 (st. dev.=1101.15 mm3). The average trac t volume for tracts connecting pars triangularis with ventral anterior nucleus and pulvinar is 1992.70 mm3 (st. dev.=882.38 mm3). In addition, pathways connecting pars opercularis with ventral anterior nucleus and pulvinar are significantly smaller in volu me than those connecting pars triangularis with these subcortical nuclei though participants 1 and 6 are exceptions to this significant difference) (paired two tailed t test: p=0.02, tstat= 2.72, df=9). nucleus or pulvinar exclusively. Specifically, the distribution of tract volumes for pathways connecting pars opercularis, ventral anterior nucleus, and pulvinar has significantly smaller volumes than the tract volume distributions for pathways connecting pars opercularis and ventral anterior nucleus and for pathways connecting pars opercularis and pulvinar (paired two tailed t test: p=0.0074, tstat=3.44, df=9, and p=0.0016, tstat=4.46, df=9 respectively). Tract volumes for pathways connecting pars triangularis, ventral anterior nucleus, and pulvinar are significantly smaller than tract volumes of c onnections between pars triangularis and ventral anterior nucleus and pars

PAGE 59

59 triangularis and pulvinar (paired two tailed t test: p=0.03, tstat=2.58, df=9; and p=0.0002, tstat=6.10, df=9 respectively). Table 3 7 represents edge weight values for pathways con necting pars opercularis/triangularis with ventral anterior nucleus and pulvinar. The average edge weight for pathways connecting pars opercularis with ventral anterior nucleus and pulvinar is 1.49x10 5 (st. dev.=3.33x10 5 ). The average edge weight for pat hways connecting pars triangularis with ventral anterior nucleus and pulvinar is 5.61x10 5 (st. dev.=7.42x10 5 ). There is a trend showing that for most participants edge weight values for pathways connecting pars triangularis with thalamic nuclei are larger than those for pathways connecting pars opercularis with ventral anterior nucleus and pulvinar (paired two t ailed t test: p=0.13, tstat= 1.67, df=9). Pathways connecting pars opercularis with ventral anterior nucleus and pulvinar did not have statistically significantly different edge weight values than pars opercularis ventral anterior nucleus pathways or pars opercularis pulvinar pathways (paired two tailed t test: p=0.18, tstat= 1.47, df=9; and p=0.26, tstat= 1.21, df=9 respectively). In contrast, edge weights for pathways connecting pars triangularis with ventral anterior nucleus and pulvinar are smaller tha n those for pars triangularis ventral anterior nucleus and for pars triangularis pulvinar pathways (paired two tailed t test: p=0.03, tstat= 2.55, df=9; and p=0.02, tstat= 2.71, df=9 respectively). Figure 3 5 depicts projections between pars opercularis (a ) and pars triangularis (b), ventral anterior nucleus, and pulvinar in a representative dataset (participant 1). Figure 3 6 depicts these pathways in all ten participants. Color gradient of the fibers represent local fiber orientation as described above. P athways connecting pars

PAGE 60

60 opercularis and pars triangularis with ventral anterior nucleus and pulvinar course medially and then posterior inferiorly to first reach the ventral anterior nucleus. From there pathways track posteriorly through the thalamus towar d the anterior superior portion of pulvinar. In addition, we note that in majority of our participants (particularly, participant 1, 3, 5, 6, 7, 9) we observe a second branch of the pathway tracking between the pulvinar and the cortex without passing throu gh the ventral anterior nucleus. One explanation for the presence of two branches is that the first branch (between pars opercularis/triangularis, ventral anterior nucleus, and pulvinar) represents the cortico ortical network, while the second branch (connecting pulvinar with pars opercularis/triangularis) represents thalamo cortical projections. Unfortunately, there is no method to deduce directionality of white matter bundles using diffusion tractography and f urther studies are necessary to explore this hypothesis. Our results show that our tracking method successfully allows us to trace r, we were able to track pathways connecting pars opercularis and pars triangularis with ventral anterior nucleus and pulvinar. Descriptive measures that we employed to characterize the strength of connectivity of these pathways indicate that connections b etween pars opercularis and thalamus may be less robust than those between pars triangularis and thalamus. This trend was evident for tract volumes of the three pathways considered here and reached statistical significance for pathways connecting pars opec ularis with pulvinar, and for pathways connecting this cortical region with both ventral anterior nucleus and pulvinar. We note that the edge weight values for these pathways also support this trend,

PAGE 61

61 although it did not reach statistical significance. A po tential explanation for this finding may be the fact that the underlying structural complexity of this region (i.e. tortuous trajectory around the circular sulcus) creates additional tracking difficulty. Anatomical location of pars opercularis results in l arger projection angle that the pathways must take in order to traverse the circular sulcus. Increasing the loop check angle of our tracking algorithm could potentially result in more tracts connecting pars opercularis with the thalamus, however, it would also increase the likelihood of generating fibers that would loop back onto themselves. Higher resolution may allow better visualization of these fiber tracts, however, increasing the resolution adds more noise and therefor more uncertainty to the data, in creasing tracking difficulty. One potential approach to further investigate circuitry connection pars opercularis with thalamic nuclei would be to be to increase the number of seeds per voxel and in addition, decrease the streamline propagation step size. Increasing the number of seeds per voxel may allow us to generate more streamlines from each voxel within pars opercularis and smaller step size could allow us to trace pathways with larger turning angles. In addition, pathways connecting pars opercularis/ triangularis with ventral anterior nucleus and pulvinar had smaller tract volumes and edge weights than pathways connecting these cortical regions with each of the thalamic nuclei separately. Tract volumes for pathways connecting pars opercularis with vent ral anterior nucleus and pulvinar were significantly smaller than pars opercularis ventral anterior nucleus and pars opercularis pulvinar pathways. Similarly, tract volumes and edge weight values were significantly smaller for pathways connecting pars tria ngularis with ventral

PAGE 62

62 anterior nucleus and pulvinar than pars triangularis ventral anterior nucleus and pars triangularis pulvinar pathways.

PAGE 63

63 Table 3 1. Region of Interest Surface Area Measures. Region Of Interest Surface Area Measures (mm 2 ) Particip ant Pars Opercularis Mask Surface Area (mm 2 ) Pars Triangularis Mask Surface Area (mm 2 ) Pulvinar Mask Surface Area (mm 2 ) Ventral Anterior Nucleus Surface Area (mm 2 ) 1 1498.89 1209.29 950.42 379.63 2 1838.15 1920.33 732.54 373.12 3 1449.92 1016.58 891.22 328.39 4 1634.49 2175.80 861.37 322.73 5 2315.29 1760.30 874.47 407.75 6 1367.00 960 728.11 319.29 7 2200.31 1336.80 886.76 317.56 8 1672.46 1904.77 756.22 394.74 9 1618.70 2409.78 702.13 347.87 10 1667.00 1597.45 828.42 378.92 Average 1726.23 1629.11 821.17 357.00 St. Dev. 310.50 491.25 85.20 33.84 Table 3 2 Tract Volumes Nucleus of the Thalamus. Participant Pars Opercular i s Ventral Anterior Nucleus Tract Volume (mm 3 ) Pars Triangularis Ventral Anterior Nucleus Tract Volume (mm 3 ) 1 6119 3900 2 1973 3925 3 1532 2041 4 910 4598 5 2153 721 6 3164 2923 7 4810 5924 8 1639 3255 9 6050 8307 10 445 471 Average 2879.5 3606.5 St. Dev. 2078 2353.13

PAGE 64

64 Table 3 3 Anterior Nucleus of the Thalamus Edge Weights) Participant Pars Opercular i s Ventral Anterior Nucleus Tract Edge Weight Pars Triangularis Ventral Anterior Nucleus Tract Edge Weight 1 3.77x10^ 3 1.57x10^ 3 2 2.22x10^ 5 2.57x10^ 3 3 5.31x10^ 5 1.60x10^ 4 4 2.58x10^ 6 4.39x10^ 3 5 5.87x10^ 5 4.25x10^ 5 6 8.89x10^ 4 6.85x10^ 3 7 2.51x10^ 4 7.85x10^ 4 8 3.91x10^ 5 3.03x10^ 4 9 3.42x10^ 4 1.77x10^ 3 10 7.35x10^ 7 2.29x10^ 6 Average 5.43x10^ 4 1.84x10^ 3 St. Dev. 1.17x10^ 3 2.24x10^ 3 Table 3 4 Tract Volumes Participant Pars Opercular i s Pulvinar Tract Volume (mm 3 ) Pars Triangularis Pulvinar Tract Volume (mm 3 ) 1 9826 8319 2 2181 5193 3 2166 3316 4 5877 7908 5 3075 1740 6 1748 2354 7 4596 6533 8 2741 5637 9 4280 5349 10 765 3292 Average 3725.5 4964.1 St. Dev. 2620.33 2254.78

PAGE 65

65 Table 3 5 Tract Weights) Participant Pars Opercular i s Pulvinar Tract Edge Weight Pars Triangularis Pulvinar Tract Edge Weight 1 5.41x10^ 3 1.72x10^ 3 2 7.39x10^ 6 1.00x10^ 3 3 2.91x10^ 5 7.29x10^ 4 4 5.19x10^ 5 4.69x10^ 3 5 3.16x10^ 4 1.31x10^ 4 6 4.97x10^ 5 1.83x10^ 4 7 5.47x10^ 4 8.98x10^ 4 8 3.32x10^ 5 2.08x10^ 4 9 1.26x10^ 5 3.39x10^ 3 10 1.51x10^ 6 3.24x10^ 4 Average 6.46x10^ 4 1.33x10^ 3 St. Dev. 1.68x10^ 3 1.54x10^ 3 Table 3 6 Tract Volumes Nucleus and Pulvinar. Ventral Anterior Nucleus Pulvinar Pathways Participant Pars Opercular i s Ventral Anterior Nucleus Pulvinar Tract Volume (mm 3 ) Pars Triangularis Ventral Anterior Nucleus Pulvinar Tract Volume (mm 3 ) 1 3599 2564 2 268 2425 3 763 1239 4 178 3027 5 1261 675 6 0 845 7 1978 3325 8 941 1859 9 189 2053 10 539 1915 Average 971.6 1992.7 St. Dev. 1101.15 882.38

PAGE 66

66 Table 3 7 Tract Edge Weights for Anterior Nucleus and Pulvinar. Ventral Anterior Nucleus Pulvinar Pathways (Edge Weights) Participant Pars Opercular i s Ventral Anterior Nucleus Pulvinar Tract Edge Weight Pars Triangularis Ventral Anterior Nucleus Pulvinar Tract Edge Weight 1 1.06x10^ 4 8.37x10^ 5 2 2.92x10^ 7 3.41x10^ 5 3 2.62x10^ 6 3.30x10^ 5 4 9.89x10^ 8 2.57x10^ 4 5 6.00x10^ 6 4.77x10^ 6 6 0 3.13x10^ 6 7 3.05x10^ 5 5.29x10^ 5 8 2.74x10^ 6 3.52x10^ 5 9 2.29x10^ 7 3.08x10^ 5 10 5.18x10^ 7 2.62x10^ 5 Average 1.49x10^ 5 5.61x10^ 5 St. Dev. 3.33x10^ 5 7.42x10^ 5

PAGE 67

67 Figure 3 1 Projections connecting pars opercularis (a) and pars triangularis, (b) with ventral anterior nucleus (pink) in participant 1. The color gradient of the pathways represents local fiber orientation: red left/right; green anterior/posterior; blue superior/inferior.

PAGE 68

68 Figure 3 2. Pathways connecting pars opercularis (a) and pars triangularis, (b) w ith ventral anterior nucleus of the thalamus (pink). The color gradient of the pathways represents local fiber orientation: red left/right; green anterior/posterior; blue superior/inferior.

PAGE 69

69 Figure 3 3. Projections connecting pars opercularis (a) an d pars triangularis, (b) with pulvinar (purple) in participant 1. The color gradient of the pathways represents local fiber orientation: red left/right; green anterior/posterior; blue superior/inferior.

PAGE 70

70 Figure 3 4. Pathways connecting pars opercularis (a) and pars triangularis, (b) with pulvinar (purple). The color gradient of the pathways represents local fiber orientation: red left/right; green anterior/posterior; blue superior/inferior.

PAGE 71

71 Figure 3 5. Projec tions among opercularis (a), pars triangularis (b), ventral anterior nucleus (pink), and pulvinar (purple) in participant 1. The color gradient of the pathways represents local fiber orientation: red left/right; green anterior/posterior; blue superior/i nferior.

PAGE 72

72 Figure 3 6. Pathways connecting pars opercularis (a) and pars triangularis (b) with ventral anterior nucleus of the thalamus (pink) and pulvinar (purple). The color gradient of the pathways represents local fiber orientation: red left/right; gr een anterior/posterior; blue superior/inferior. Sub figure 6a for participant 6 is blank because our tractography analysis did not reveal any tracts connecting pars opercularis with ventral anterior nucleus and pulvinar in this participant.

PAGE 73

73 CHAPTER 4 EFFECTS OF ANTERIOR AND POSTERIOR THALAMIC LESIONS ON PATHWAYS Anterior Thalamic Lesion Effects area and the thalamus we created models of two exemplar thalamic lesions based on previously described clinical cases (Raymer et al., 1996; Crosson et al., 1997) (Figure 2 1). The first model represents an anterior thalamic lesion affecting ventral anterior nucleus as well as portions of the vent ral lateral nucleus of the thalamus (Figure 2 1 (a)). To investigate effects of this type of anterior thalamic lesion we applied the lesion model as an exclusion mask to each of the six pathways described in the previous section. This approach eliminates a ny tracts that travel through the lesion allowing us to thalamocortical circuitry. Anterior Thalamic Lesion Effects rea and Ventral Anterior Nu cleus Table 4 1 represents tract volumes for pathways connecting pars opercularis and pars triangularis with ventral anterior nucleus before and after an anterior thalamic lesion. The last column of the table represents percent change in tract volumes as a resul t of the lesion. We note that percent change in tract volume is negative (or equal to zero) for all participants and all pathways indicating a decrease in tract volume (or no change in tract volume). The average tract volume following the lesion for pars opercularis pathways is 1257.70mm 3 (st. dev.=1484.60mm 3 ) and the average percent decrease in tract volume is 56.98% (st. dev.=30.65%). For pathways connecting pars opercularis and ventral

PAGE 74

74 anterior nucleus percent volume change ranges from 14.48% (participa nt 6) to 100% (participants 1 and 2). The average trace volume following the lesion for pars triangularis pathways is 1802.50mm 3 (st. dev.=1669.10mm 3 ) and the average percent volume change is 45.80% (st. dev.=26.17%). Percent volume change for pathways con necting pars triangularis and ventral anterior nucleus ranges from 15.71% (participant 10) to 100% (participant 1). Tract volumes following an anterior thalamic lesion for pathways connecting pars opercularis and ventral anterior nucleus and pars triangul aris and ventral anterior nucleus were not significantly different (p=0.09). Overall, an anterior thalamic lesion anterior nucleus severely damaging this circuitry. Table 4 2 represents edge weight values for pathways connecting pars opercularis/triangularis and ventral anterior nucleus before and after an anterior thalamic lesion. We note that the average edge weight for pathways connecting pars opercularis and ventral ant erior nucleus following the lesion is 1.40x10 4 (st. dev.=2.77x10 4 ) while the average percent decrease in edge weight value is 46.56% (st. dev.=38.01%). Percent decrease in edge weight value ranges from 0 (participant 6), to 100% (participants 1, 2). The average edge weight for pathways connecting pars triangularis with ventral anterior nucleus following the lesion is 2.39x10 4 (st. dev.=2.44x10 4 ) and the average percent decrease in edge weight value is 54.92% (st. dev.=39.91%). Percent decrease in edge weight value ranges from 2.5% (participant 3) to 100% (participants 1, 2). As a general trend we note that an anterior thalamic lesion

PAGE 75

75 decreases the edge weight values for pathways connecting pars opercularis/triangularis with ventral anterior nucleus by o ver 50% in most of our participants. Figure 4 1 (a1 area ventral anterior nucleus circuitry in participant 1 and Figure 4 2 depicts effects of this type of lesion in all ten participants. Examin ing the figure in detail, we note that the lesion affects anterior, medial, and lateral portions of the ventral anterior nucleus. Therefore pathways projecting to these regions of the nucleus are eliminated by the lesion. The only pathways that are preserv ed and reach the nucleus are the ones that project to the superior portions of the ventral anterior thalamus. As we noted in Chapter terminate in within the anterior medial portion of the nucleus. This pattern of connectivity makes this portion of the pathways more vulnerable to anterior thalamus lesions that affect anterior, medial, and lateral aspects of the ventral anterior nucleus. Anterior Thalamic Lesion Effects on Pat Pulvinar Table 4 before and after a virtual anterior thalamic lesion. The last column of the table lists percent change in tract volume. As for pathw volume change is negative (or equal to zero) for all participants indicating a reduction in tract volume (or no effect on tract volume). The average tract volume for pathways connecting par s opercularis and pulvinar following the lesion is 3205.70mm 3 (st. dev.=2214.88mm 3 ) and the average percent decrease in volume 13.39% (st.dev. = 12.37%). The percent volume change ranges from 0 (participants 4, 10) to 36.67%

PAGE 76

76 (participant 6). For pathways c onnecting pars triangularis with pulvinar the average tract volume following the lesion is 4121.80mm 3 (st. dev.=2198.16mm 3 ) and the average percent volume decrease is 19.62% (st.dev. = 20.85%). The percent volume change ranges from 0 (participant 5) to 29. 53% (participant 10). Tract volumes for pars opercularis pulvinar pathways were significantly smaller than those for pars triagularis pulvinar pathways following an anterior thalamic lesion (paired two tailed t test: p=0.05, tstat= 2.24, df=9). This findi ng is not surprising given the fact that the pathways connecting pars opercularis with pulvinar were significantly smaller in volume prior to the lesion as compared with pars triangularis pulvinar pathways (paired two tailed t test: p=0.04, tstat= 2.44, df =9). Overall, for most participants the lesion eliminates less than 20% of tracts connecting pars opercularis/triangularis with pulvinar. Table 4 4 represents edge weight values for pathways connecting pars opercularis/triangularis with pulvinar before and after an anterior thalamic lesion. The last column represents percent change in edge weight values following the lesion. The average edge weight value for pathways connecting pars opercularis with pulvinar following an anterior thalamic lesion is 6.04x10 4 (st. dev.=1.56x10 3 ) and the average percent decrease in edge weight is 2.51% (st. dev.=3.49%). Percent decrease ranges from 0 (participants 4, 5) to 9.28% (participant 3). The average edge weight value for pathways connecting pars triangularis and pulvi nar following an anterior thalamic lesion is 1.29x10 3 (st. dev.=1.56x10 3 ) and the average percent decrease in edge weight is 13.50% (st. dev.=29.24%). Percent decrease in edge weight value ranges from 0 (participants 8, 9) to 89.13% (participant 6). Overall we note that an anterior thalamic

PAGE 77

77 lesion results in no more than 10% d ecrease in edge weight value for pathways connecting pars opercularis/triangularis with pulvinar (with exception of participant 6). Figure 4 1 (b1 b4) represents effects of an anterior thalamic lesion on pathways ingle participant (participant 1) and Figure 4 3 depicts these results in all ten of our participants. Overall, we note that the lesion is located anterior and ventral to the anterior superior pulvinar. Due to its location, the lesion would impact only the pulvinar project to the anterior superior portion of the nucleus. This trajectory results in only a small portion of the tracts to be traveling ventrally and being affected by the lesion. Ventral Anterior Nucleus and Pulvinar Table 4 with ventral anterior nucleus and pulvinar before and after an anterior thalamic lesion. The average tract volume for pathways connecting pars opercularis with ventral anterior nucleus and pulvinar following an anterior thalamic lesion is 315.70mm 3 (st. de v.=375.95mm 3 ) and the average percent decrease in tract volume is 58% (st. dev. = 39.71%). Reduction in tract volume ranges from 0 (participants 4, 6) to 79.49% (participant 1). Similarly, the average tract volume for pathways connecting pars triangularis and pulvinar following an anterior thalamic lesion is 821.40mm 3 (st. dev.=514.43mm 3 ) and the average percent decrease in tract volume is 53.37% (st. dev. = 26.59%). Decrease in percent tract volume ranges from 15.11% (participant 5) to 80.51% (participant 7). These findings indicate that an anterior thalamic lesion creates a

PAGE 78

78 nucleus and pulvinar. This type of lesion, on average, eliminates more than half of the fibers thus s everely impairing this connectivity. In addition, the effects of an anterior thalamic lesion on pathways connecting pars opercularis with ventral anterior nucleus and pulvinar are more severe than for pathways connecting pars triangularis with these thalam ic nuclei considered exclusively. This result is supported by significantly smaller tract volumes for pars opercularis ventral anterior nucleus pulvinar pathways as compared with those for pars triangularis pathways (paired two tailed t test: p=0.05, tstat = 2.27, df=9). As we recall from Chapter 3, pathways connecting pars opercularis with the two thalamic nuclei exclusively were significantly smaller in volume than those connecting pars triangularis and the two thalamic nuclei inclusively (paired two taile d t test: p=0.02, tstat= 2.72, df=9). Therefore, it is not surprising that these pathways undergo larger absolute reduction in tract volumes following the lesion. Table 4 6 represents edge weight values for pathways connecting pars opercularis/triangularis with ventral anterior nucleus and pulvinar before and after an anterior thalamic lesion. The last column of the table represents percent change in edge weight values following the lesion. The average edge weight for pathways connecting pars opercularis wi th ventral anterior nucleus and pulvinar following the lesion is 8.25x10 6 (st. dev.=1.80x10 5 ) and the average percent decrease in edge weight is 48.14% (st. dev.=39.67%). Percent decease in edge weight values ranges from 0 (participant 4) to 100% (partic ipants 2, 9). The average edge weight value for pathways connecting pars triangularis with ventral anterior nucleus and pulvinar following the lesion is 2.33x10 5 (st. dev.=4.13x10 5 ) and the average percent decrease in edge

PAGE 79

79 weight value is 57.44% (st. dev .=35.20%). Percent decrease in edge weight ranges from 11.74% (participant 5) to 99.20% (participant 1). Overall we note that an anterior thalamic lesion results on average in 50% reduction in edge weight values for pathways connecting pars opercularis/tri angularis with ventral anterior nucleus and pulvinar. Figure 4 1 (c1 c4) depicts results of an anterior thalamic lesion on pathways (participant 1) and Figure 4 4 de picts the corresponding results in all ten of our participants. We note that in most of our participants an anterior thalamic lesion results anterior nucleus. The second pulvinar is affected less due to the fact that these pathways travel superiorly to the lesion. In addition, in some participants (2, 4, 6, and 9) the lesion eliminates all pathways connecting pars opercul aris and/or pars triangularis with ventral anterior nucleus and pulvinar. For most of the remaining participants an anterior thalamic lesion nuclei considered inclusive ly. Area and Thalamus In conclusion, our results demonstrate that an anterior thalamic lesion affecting ventral anterior nucleus and parts of ventral lateral nucleus produces varyin g degrees of greatest amount of damage to pathways connecting pars opercularis/triangularis with ventral anterior nucleus and pulvinar as the tract volumes for these pathway s following the lesion are almost always significantly smaller than those for the other pathways

PAGE 80

80 (p<0.01 for pars opercularis pulvinar versus pars opercularis ventral anterior nucleus pulvinar tract volumes; p=0.05 for pars triangularis ventral anterior nu cleus versus pars triangularis ventral anterior pulvinar pathways tract volumes; p<0.01 for pars triangularis pulvinar vs. pars triangularis ventral anterior pulvinar pathways). Following an anterior thalamic lesion more than half of pathways connecting pa rs opercularis/triangularis with ventral anterior nucleus and pulvinar are eliminated. In addition, edge weights for pathways connecting pars opercularis/triangularis with ventral anterior nucleus and pulvinar are smaller than those for the other two pathw ays. This trend reaches statistical significance for pars triangularis ventral anterior pulvinar pathways (p=0.02 for pars triangularis ventral anterior pulvinar versus pars triangularis ventral anterior nucleus pathways; p=0.03 for pars triangularis ventr al anterior pulvinar versus pars triangularis pulvinar pathways). We believe this occurs due to the fact that these pathways are tightly packed and highly contiguous and therefore are likely to be greatly affected by a focal lesion. In addition, the pathwa ys must travel through the ventral anterior nucleus most of which is engulfed by the lesion. Pathways connecting pars opercularis/triangularis with ventral anterior nucleus are also largely affected by anterior thalamic lesions. In particular, following th e lesion we see a larger reduction in tract volumes for these pathways than for pathways connecting pars opercularis/triangularis with pulvinar (this trend reaches significance for pars triangularis ventral anterior nucleus pathways, p=0.01 and the trend i s close to significance for pars opercularis ventral anterior nucleus pathways, p=0.06). Overall, we observed on average a 50% reduction in tract volumes following an anterior thalamic lesion. Comparisons of edge weight distributions following an anterior thalamic lesion

PAGE 81

81 did not reveal any statistically significant differences among the pathways connecting pars opercularis/triangularis with ventral anterior nucleus. We note that on average the lesion results in over 50% reduction in edge weights values in m ost of our participants. amount of damage as about one fifth of the pathways are eliminated following the lesion. In addition, most of our participants show a less than 10% reduction in edge weight values within these pathways following the lesion. These results are not surprising if we consider the underlying anatomy of the pathways and the lesion. The anterior lesion model largely overlaps with the ventral anterior nucleus and there fore affects a great number of pathways coursing towards the nucleus. Pulvinar on the other hand is located posteriorly and does not overlap with the lesion. Thus, only pathways traveling within the vicinity of the ventral anterior nucleus would be affecte d. Posterior Thalamic Lesion Effects Our second thalamic lesion model is based on a case presented by Crosson and colleagues. The patient in this case suffered a lesion affecting portions of the pulvinar, posterior limb of the internal capsule, and centro medial nucleus of the thalamus (see Figure 2 1 (b)). As a result of this lesion the patient was unable to produce a response within medical items category and did not improve chronically. This observation may indicate that patient in addition to having def icits in semantic processing also suffered deficits to lexical processing or at the interface between these processing levels. In order to investigate effects of a posterior thalamic lesion that produced chronic anomia for medical items in the patient exa mined by Crosson and colleagues, we used

PAGE 82

82 area and the thalamus. Below we describe in detail effects of a virtual posterior thalamic lesion on each of the pathways within Posterior Thalamic Lesion Effects rea and Ventral Anterior Nucleus Table 4 ventral anterior nucleus before and after a posterior thalamic lesion. The last column represents percent change in tract volume following the lesion. We note that as with an anterior thalamic lesion (described above), all of the percent change values are negative (or zero) indicating that tract volume is reduced (or remains constant). The average percent volume change for pathways connecting pars opercularis and ventral anterior nucleus is 13.64% (st.dev. = 21.11%). Tract volumes reduction ranges from 0 (participants 2, 4, 6, 8, 9, 10) to 61.62% (participant 7). We note that for six out of ten participants (participants 2, 4, 6, 8, 9, 10) the posterior lesion did not have any effect on pathways connecting pars opercularis and ventral anterior nucleus as tract volumes remained unchanged after appl ication of this exclusion mask. For the datasets in which the lesion did result in volume reduction, percent volume change ranges between 13.16% and 61.62% (participants 1 and 7 respectively). These reductions in tract volume are indicative of the fact tha t the posterior thalamic lesion extends anterior and affects a portion of the pathways connecting pars opercularis and ventral anterior nucleus that travels around the nucleus and enters its posterior extent. In particular, if we examine Figure 3 2 closely we note that for participants 1, 3, 5, and 7 some of the pathways connecting pars opercularis and ventral anterior nucleus appear to project to more posterior medial portions of the nucleus and travel far enough posteriorly to be affected by the lesion.

PAGE 83

83 The average percent volume change with posterior lesion for pathways connecting pars triangularis and ventral anterior nucleus is 11.27% (st. dev. = 12.46%) with range from 0 (participants 6, 10) to 40.97% (participant 7). Most of the datasets show below 10% tract volume decrease (participants 1, 2, 3, 4, 8), two datasets are not affected by the lesion (participants 6, 10), and three datasets show more than 10% volume reduction (participants 5, 7, and 9). As in the case of pars opercularis ventral anterior nucleus pathways, we note that a portion of pathways connecting pars triangularis and ventral anterior nucleus travels around the nucleus and projects to its posterior extent. In participants showing volumetric reduction in tract volumes following the les ion, this portion of the pathway extends posteriorly and is affected by the lesion. Table 4 8 represent edge weight values for pathways connecting pars opercularis/triangularis with ventral anterior thalamus before and after a posterior thalamic lesion. We note that following the lesion the percent decrease in edge weight value ranges from 0 (participants 2, 4, 6, 8, 9, 10) to 6.30% (participant 5) for pathways connecting pars opercularis with ventral anterior nucleus. The average percent edge weight decrea se for these pathways is 1.37% (st. dev.=2.41%). The average percent edge weight decease for pathways connecting pars triangularis and ventral anterior nucleus following the lesion is 0.56% (st. dev.=0.92%) with range from 0 (participants 1, 6, and 10) to 3.06% (participant 5). These results indicate that a posterior thalamic lesion should produce mild effects on circuitry connecting pars opercularis/triangularis with ventral anterior nucleus. In particular, in most of our participants, tract volumes are re duced by no more than 15% (with exception of participants 3, 5, and 7 for pars opercularis pathways; and with

PAGE 84

84 exception of participants 5 and 7 for pars triangularis pathways). Edge weight values are reduced by no more than 2% following the lesion (with ex ception of participants 1 and 5). Figure 4 5 (a1 anterior nucleus before and after a posterior thalamic lesion in participant 1. Figure 4 6 represents the corresponding lesion effects in all ten participants. We note that for all of our participants the lesion is located posterior and lateral to the ventral anterior nucleus. The lesion affects only the pathways that project to the lateral portions of the nucleus. Since most of the pathways betwee to the anterior and medial aspects of the nucleus, only a small fraction of the pathways is vulnerable to this posterior thalamic lesion. Posterior Thalamic Lesion Effects A rea and Pulvinar Table 4 9 represents tract volume changes before and after a virtual posterior percent change for pathways connecting pars opercularis and pulvinar is 57.17% (st. dev. = 18.63%) and the overall volume reduction ranges from 36% to 91.25% (participants 4 and 8 respectively). The average tract volume percent change for pathways connecting pars triangularis and pulvinar following a posterior thalamic lesion is 48.02% (st. dev. = 17.36%) and range from 18.1% to 72.26% (participants 3 and 9 respectively). Table 4 10 represents edge weight values for pathways connecting pars opercularis/triangularis with pulvinar following a posterior thalamic lesion. The aver age percent decrease in edge weight value for pathways connecting pars opercularis and

PAGE 85

85 pulvinar is 81.41% (st. dev.=19.78%) with range from 40.10% (participant 8) to 99.42% (participant 4). We note that for most of our participants (with exception of part icipant 8) percent reduction in edge weight is equal to more than 50%. Similarly, the average percent decrease in edge weight values for pathways connecting pars triangularis with pulvinar is 68.49% (st. dev.=29.72%) with range from 2.33% (participant 3) t o 99.35% (participant 1). Most of our participants demonstrate a profound reduction in edge weight values following the lesion (with exception of participant 3). These results show that a posterior thalamic lesion results in severe damage to the pathways participants the lesion eliminates more than half of pathways connecting pars opercularis/triangularis with pulvinar (with exception of participants 3, 5, 7, 8 for pars opercularis pathw ays; with exception of participants 2, 3, 5, 8 for pars triangularis pathways). Posterior thalamic lesion results in even greater reductions in edge weight values for these pathways. Specifically, we observe a 70% or more reduction in edge weight values fo r pathways connecting pars opercularis with pulvinar (with exception of participants 3, and 8) and more than 60% reduction for pars triangularis pulvinar pathways (with exception of participants 3, and 10). Figure 4 5 (b1 b4) represents effects of a poster ior thalamic lesion on pathways 7 depicts these effects in all ten participants. We note that the lesion is located anterior to the pulvinar affecting the anterior superior portions of the nucleus. Since most of the pathways pathways are very vulnerable to this kind of posterior thalamic lesion.

PAGE 86

86 Area with Ventral Anterior Nucleus and Pulvinar To conclude our analysis of posterior thalamic lesion effects on circuitry ior nucleus and pulvinar. Table 4 11 lists tract volumes before and a posterior thalamic lesion for these pathways. These results show that the average percent volume change for pathways connecting pars opercularis with ventral anterior nucleus and pulvina r is 72.29% (st. dev. = 38.62%). Tract volume reductions range from 0 (participant 6) to 100% (participants 2, 4, 9, 10) with most of participants showing over 50% tract volume decrease. The average tract volume reduction for pathways connecting pars tria ngularis with ventral anterior nucleus and pulvinar is 81.70% (st. dev. = 14.51%) and tract volume decrease ranges from 57.93% (participant 5) to 100% (participants 2, 4, 9, 10). Most of our participants show a significant tract volume reduction of more th an 50%. Table 4 12 represents edge weight values for pathways connecting pars opercularis/triangularis with ventral anterior nucleus and pulvinar before and after a posterior thalamic lesion. The average percent decrease in edge weight for pathways connect ing pars opercularis with ventral anterior nucleus and pulvinar is 78.70% (st. dev.=33.36%) with range from 0 (participant 6) to 100% (participants 2, 4, 9, 10). We note that in most of our participants edge weight values are reduced by more than 50% (with exception of participant 6). The average percent decrease in edge weight values for pathways connecting pars triangularis with ventral anterior nucleus and pulvinar is 84.65% (st. dev.=22.15%) with range from 37.11% (participant 5) to 100% (participants 6, 9). We observe that in

PAGE 87

87 most of our participants the edge weight values are reduced by 60% or more (with exception of participant 5). Based on these findings, we conclude that this posterior thalamic lesion results in profound damage to circuitry connect ing pars opercularis/triangularis with ventral anterior nucleus and pulvinar. Specifically we observe that tract volumes are reduced by more than 60% following the lesion (with exception of participant 3 for pars opercularis pathways; with exception of par ticipant 5 for pars triangularis pathways). Likewise, edge weight values are reduced by more than 80% (with exception of participants 3, and 5 for pars opercularis pathways; with exception of participants 1, 5, and 10 for pars triangularis pathways). Figur e 4 5 (c1 c4) depicts effects of a posterior thalamic lesion on pathways Figure 4 8 represents the corresponding results in all ten participants. We note that a posteri or thalamic lesion eliminated all pathways connecting pars opercularis and/or pars triangularis with the two thalamic nuclei in participants 2, 4, 6, 9, and 10. In the remaining participants the lesion mostly eliminated the branch of the pathways connectin anterior nucleus sustained less damage. This result is not surprising if we consider lesion location in reference to the two thalamic nuclei. The posterior lesion is located poste rior to the ventral anterior nucleus and anterior to the pulvinar. Thus, pathways traveling from the ventral anterior nucleus to the pulvinar and those between pulvinar be affected by the lesion.

PAGE 88

88 Area and Thalamus In conclusion we note that a posterior thalamic lesion produces varying degree of us. Pathways connecting pars opercularis/triangularis with ventral anterior nucleus appear to sustain the least amount of damage from posterior lesion. For most of our participants we observe no more than 15% reduction in tract volumes following the lesion (with exception of participants 3, 5, 7 for pars opercularis pathways; with exception of participants 5, 7 for pars triangularis pathways). The edge weight values for these pathways decrease by no more than 2% following the posterior lesion (with exceptio n of participants 1 and 5 for pars opercularis pathways; with exception of participant 5 for pars triangularis pathways). Posterior thalamic lesions result in profound damage to pathways connecting pars opercularis/triangularis and pulvinar. In our study, we observed a 50% or more reduction in tract volumes following the posterior lesion (with exception of participants 5, 7, 8 for pars opercularis pathways; with exception of participants 2, 3, 5, 7, 8 for pars triangularis pathways). The edge weight values for these pathways were reduced by more than 80% for pars opercularis pathways to pulvinar (with exception of participants 3, 7, 8) and by more than 60% for pars triangularis pathways to pulvinar (with exception of participants 3, 10). Pathways connecting pars opercularis/triangularis with ventral anterior nucleus and pulvinar also sustain a large degree of damage following a posterior thalamic lesion. In our participants, we observed a 60% or more reduction in tract volumes as a result of the posterior le sion (with exception of participant 3 for pars opercularis pathways; with exception of participant 5 for pars triangularis pathways). Edge weight

PAGE 89

89 values were reduced by more than 80% for pars opercularis pathways (with exception of participant 3) and by 60 % or more for pars triangularis pathways (with exception of participant 5). We believe that lesion location plays a crucial role in these results. The posterior thalamic lesion model is located posterior to the ventral anterior nucleus and anterior to the pulvinar affecting most of its anterior superior portion. Most of the pathways connecting pars opercularis/triangularis with ventral anterior nucleus travel anteriorly to this lesion. Thus this lesion only affects the most posterior portion of ventral ante rior nucleus pathways and results in only small amounts of damage to this circuitry. On the other hand, in most participants the posterior lesion leaves only the most superior aspect of the anterior superior of pulvinar unaffected. In this case pathways th at travel directly to this portion of the pulvinar are preserved. Thus, it is not surprising pulvinar. In addition, our results demonstrate that posterior thalamic lesion pr oduces and pulvinar. This fiber bundle is tightly packed in most of our participants and projects to the anterior superior portion of the pulvinar. The posterior lesion affecting this portion of the nucleus results in severe damage to this circuitry as supported by our results. Conclusions: Differential Effects of Anterior and Posterior Thalamic Lesions on Table 4 13 provides a summa ry of differential effects of anterior and posterior percent tract volume decrease and average percent edge weight value decrease following anterior/posterior thalamic lesions for the six pathways examined in this

PAGE 90

90 manuscript (values in parenthesis are standard deviations associated with each corresponding average value). Overall we note that an anterior thalamic lesion results in large deficits for pathways connecting pa rs opercularis/triangularis with the ventral anterior nucleus. On average, these pathways undergo a 50% reduction in tract volume and edge weight values following the anterior lesion. The greatest amount of damage imposed by an anterior thalamic lesion can be observed in the pathways connecting pars opercularis/triangularis with ventral anterior nucleus and pulvinar. These pathways are reduced by more than 50% in volume and in connectivity strength as depicted by the edge weight values. The least amount of deficit following an anterior thalamic lesion is observed in the pathways connecting pars opercularis/triangularis and pulvinar. These pathways undergo a less than 20% reduction in tract volume and less than 15% reduction in edge weight values following th e lesion. Posterior lesion results in the least amount of damage to the pathways connecting pars opercularis/triangularis and ventral anterior nucleus. Following this type of thalamic lesion the pathways are reduced in volume by no more than 15% and no mor e than 2% in edge weight. Pathways connecting pars opercularis/triangularis with pulvinar are affected to a much larger extent. Specifically, these pathways undergo a 50% reduction in volume and 70% reduction in edge weight values. The largest amount of da maged imposed by a posterior thalamic lesion is observed in pathways connecting pars opercularis/triangularis with ventral anterior nucleus and pulvinar. These pathways are reduced in volume by 70% or more and in edge weight values by 80% or more.

PAGE 91

91 Table 4 1 Ventral Anterior Nucleus of the Thalamus (Tract Volumes) Anterior Nucleus of the Thalamus (Tract Volumes) Participant Pars Opercularis Ventral Anterior Nucleus Tract Volume (mm 3 ) Pars Opercularis Ventral Anterior Nucleus Track Volume Post Anterior Thalamic Lesion (mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 6119 0 100 2 1973 0 100 3 1532 383 75 4 910 255 71.98 5 2153 1404 34.79 6 3164 2706 14.48 7 4810 1879 60.94 8 1639 1163 29 9 6050 4608 23.84 10 445 179 59.78 Average 2879.5 1257.7 56.98 St. Dev. 2078 1484.6 30.65 Participant Pars Triangularis Ventral Anterior Nucleus Tract Volume (mm 3 ) Pars Triangularis Ventral Anterior Nucleus Tract Volume Post Anterior Thalamic Lesion (mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 3900 0 100 2 3925 838 78.65 3 2041 1144 43.95 4 4598 2495 45.74 5 721 532 26.2 6 2923 2036 30.35 7 5924 2545 57 8 3255 2293 29.55 9 8307 5745 30.84 10 471 397 15.71 Average 3606.5 1802.5 45.80 St. Dev. 2353.13 1669.1 26.17

PAGE 92

92 Table 4 2 Anterior Thalamic Lesion Effects on Pathways Ventral Anterior Nucleus of the Thalamus (Tract Edge Weights) Anterior Nucleus of the Thalamus (Tract Edge Weights) Participant Pars Opercularis Ventral Anterior Nucleus Tract Edge Weight Pars Opercularis Ventral Anterior Nucleus Track Edge Weight Post Anterior Thalamic Lesion % Change in Tract Edge Weight Post Anterior Thalamic Lesion (%) 1 3.77x10^ 3 0 100 2 2.22x10^ 5 0 100 3 5.31x10^ 5 9.60x10^ 6 81.92 4 2.58x10^ 6 7.46x10^ 7 71.09 5 5.87x10^ 5 4.82x10^ 5 17.89 6 8.89x10^ 4 8.89x10^ 4 0 7 2.51x10^ 4 1.56x10^ 4 37.85 8 3.91x10^ 5 3.02x10^ 5 22.76 9 3.42x10^ 4 2.64x10^ 4 22.81 10 7.35x10^ 7 6.52x10^ 7 11.29 Average 5.43x10^ 4 1.40x10^ 4 46.56 St. Dev. 1.17x10^ 3 2.77x10^ 4 38.01 Participant Pars Triangularis Ventral Anterior Nucleus Tract Edge Weight Pars Triangularis Ventral Anterior Nucleus Tract Edge Weight Post Anterior Thalamic Lesion % Change in Tract Edge Weight Post Anterior Thalamic Lesion (%) 1 1.57x10^ 3 0 100 2 2.57x10^ 3 4.87x10^ 6 99.81 3 1.60x10^ 4 1.56x10^ 4 2.5 4 4.39x10^ 3 5.14x10^ 4 88.29 5 4.25x10^ 5 3.98x10^ 5 6.36 6 6.85x10^ 3 4.30x10^ 4 93.72 7 7.85x10^ 4 3.61x10^ 4 54.01 8 3.03x10^ 4 2.02x10^ 4 33.33 9 1.77x10^ 3 6.80x10^ 4 61.58 10 2.29x10^ 6 2.07x10^ 6 9.61 Average 1.84x10^ 3 2.39x10^ 4 54.92 St. Dev. 2.24x10^ 3 2.44x10^ 4 39.91

PAGE 93

93 Table 4 3 Pulvinar (Tract Volumes) (Tract Volumes) Participant Pars Opercularis Pulvinar Tract Volume (mm 3 ) Pars Opercularis Pulvinar Track Volume Post Anterior Thalamic Lesion ( mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 9826 7866 19.95 2 2181 1818 16.64 3 2166 1880 13.2 4 5877 5877 0 5 3075 2762 10.18 6 1748 1107 36.67 7 4596 3266 28.94 8 2741 2649 3.36 9 4280 4067 4.98 10 765 765 0 Average 3725.5 3205.7 13.39 St. Dev. 2620.33 2214.88 12.37 Participant Pars Triangularis Pulvinar Tract Volume (mm 3 ) Pars Triangularis Pulvinar Tract Volume Post Anterior Thalamic Lesion (mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 8319 7016 15.66 2 5193 4204 19.1 3 3316 2789 15.89 4 7908 7181 9.19 5 1740 1740 0 6 2354 664 71.79 7 6533 4733 27.55 8 5637 5360 4.91 9 5349 5211 2.58 10 3292 2320 29.53 Average 4964.1 4121.8 19.62 St. Dev. 2254.78 2198.16 20.85

PAGE 94

94 Table 4 4 Pulvinar (Tract Edge Weights) (Tract Edge Weights) Participant Pars Opercularis Pulvinar Tract Edge Weight Pars Opercularis Pulvinar Track Edge Weight Post Anterior Thalamic Lesion % Change in Tract Edge Weight Post Anterior Thalamic Lesion (%) 1 5.41x10^ 3 5.00x10^ 3 7.58 2 7.39x10^ 6 7.02x10^ 6 5.01 3 2.91x10^ 5 2.64x10^ 5 9.28 4 5.19x10^ 5 5.19x10^ 5 0 5 3.16x10^ 4 3.16x10^ 4 0 6 4.97x10^ 5 4.95x10^ 5 0.40 7 5.47x10^ 4 5.42x10^ 4 0.92 8 3.32x10^ 5 3.31x10^ 5 0.30 9 1.26x10^ 5 1.24x10^ 5 1.59 10 1.51x10^ 6 1.51x10^6 0 Average 6.46x10^ 4 6.04x10^ 4 2.51 St. Dev. 1.68x10^ 3 1.56x10^ 3 3.49 Participant Pars Triangularis Pulvinar Tract Edge Weight Pars Triangularis Pulvinar Tract Edge Weight Post Anterior Thalamic Lesion % Change in Tract Edge Weight Post Anterior Thalamic Lesion (%) 1 1.72x10^ 3 1.68x10^ 3 2.33 2 1.00x10^ 3 9.84x10^ 4 1.60 3 7.29x10^ 4 7.24x10^ 4 0.69 4 4.69x10^ 3 4.68x10^ 3 0.21 5 1.31x10^ 4 1.31x10^ 4 0 6 1.83x10^ 4 1.99x10^ 5 89.13 7 8.98x10^ 4 8.84x10^ 4 1.56 8 2.08x10^ 4 2.08x10^ 4 0 9 3.39x10^ 3 3.39x10^ 3 0 10 3.24x10^ 4 1.96x10^ 4 39.51 Average 1.33x10^ 3 1.29x10^ 3 13.50 St. Dev. 1.54x10^ 3 1.56x10^ 3 29.24

PAGE 95

95 Table 4 5 Ventral Anterior Nucleus of the Thalamus, and Pulvinar (Tract Volumes) Anterior Nucleus of the Thalamus, and Pulvinar (Tract Volumes) Participant Pars Opercularis Ventral Anterior Nucleus Pulvinar Tract Volume (mm 3 ) Pa rs Opercularis Ventral Anterior Nucleus Pulvinar Track Volume Post Anterior Thalamic Lesion (mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 3599 738 79.5 2 268 0 100 3 763 247 67.63 4 178 178 0 5 1261 1159 8.1 6 0 0 0 7 1978 469 76.29 8 941 207 78 9 189 0 100 10 539 159 70.5 Average 971.6 315.7 58.00 St. Dev. 1101.15 375.95 39.71 Participant Pars Triangularis Ventral Anterior Nucleus Pulvinar Tract Volume (mm 3 ) Pars Triangularis Ventral Anterior Nucleus Pulvinar Tract Volume Post Anterior Thalamic Lesion (mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 2564 563 78 2 2425 505 79.18 3 1239 986 20.42 4 3027 1905 37.1 5 675 573 15.11 6 845 336 60.24 7 3325 648 80.51 8 1859 804 56.75 9 2053 1519 26 10 1915 375 80.42 Average 1992.7 821.4 53.37 St. Dev. 882.37 514.43 26.59

PAGE 96

96 Table 4 6 Ventral Anterior Nucleus of the Thalamus, and Pulvinar (Tract Edge Weights) Anterior Nucleus of the Thalamus, and Pulvinar (Tract Edge Weights) Participant Pars Opercularis Ventral Anterior Nucleus Pulvinar Tract Edge Weight P ars Opercularis Ventral Anterior Nucleus Pulvinar Track Edge Weight Post Anterior Thalamic Lesion % Change in Tract Edge Weight Post Anterior Thalamic Lesion (%) 1 1.06x10^ 4 5.68x10^ 5 46.42 2 2.92x10^ 7 0 100 3 2.62x10^ 6 1.69x10^ 6 35.50 4 9.89x10^ 8 9.89x10^ 8 0 5 6.00x10^ 6 5.47x10^ 6 8.83 6 0 0 6 7 3.05x10^ 5 1.81x10^ 5 40.66 8 2.74x10^ 6 9.01x10^ 8 96.71 9 2.29x10^ 7 0 100 10 5.18x10^ 7 2.42x10^ 7 53.28 Average 1.49x10^ 5 8.25x10^ 6 48.14 St. Dev. 3.33x10^ 5 1.80x10^ 5 39.67 Participant Pars Triangularis Ventral Anterior Nucleus Pulvinar Tract Edge Weight Pars Triangularis Ventral Anterior Nucleus Pulvinar Tract Edge Weight Post Anterior Thalamic Lesion % Change in Tract Edge Weight Post Anterior Thalamic Lesion (%) 1 8.37x10^ 5 6.68x10^ 7 99.20 2 3.41x10^ 5 5.07x10^ 7 98.51 3 3.30x10^ 5 2.89x10^ 5 12.42 4 2.57x10^ 4 1.36x10^ 4 47.08 5 4.77x10^ 6 4.21x10^ 6 11.74 6 3.13x10^ 6 7.74x10^ 7 75.27 7 5.29x10^ 5 2.71x10^ 5 48.77 8 3.52x10^ 5 1.74x10^ 6 95.06 9 3.08x10^ 5 2.53x10^ 5 17.86 10 2.62x10^ 5 8.26x10^ 6 68.47 Average 5.61x10^ 5 2.33x10^ 5 57.44 St. Dev. 7.42x10^ 5 4.13x10^ 5 35.20

PAGE 97

97 Table 4 7 Ventral Anterior Nucleus of the Thalamus (Tract Volumes) Anterior Nucleus of the Thalamus (Tract Volumes) Participant Pars Opercularis Ventral Anterior Nucleus Tract Volume (mm 3 ) Pars Ope rcularis Ventral Anterior Nucleus Track Volume Post Posterior Thalamic Lesion (mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 6119 5314 13.16 2 1973 1973 0 3 1532 1050 31.46 4 910 910 0 5 2153 1504 30.14 6 3164 3164 0 7 4810 1846 61.62 8 1639 1639 0 9 6050 6050 0 10 445 445 0 Average 2879.5 2389.5 13.64 St. Dev. 2078 1889.5 21.11 Participant Pars Triangularis Ventral Anterior Nucleus Tract Volume (mm 3 ) Pars Triangularis Ventral Anterior Nucleus Tract Volume Post Posterior Thalamic Lesion (mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 3900 3617 7.26 2 3925 3632 7.47 3 2041 1857 9 4 4598 4438 3.48 5 721 551 23.58 6 2923 2923 0 7 5924 3497 40.96 8 3255 2994 8 9 8307 7233 12.93 10 471 471 0 Average 3606.5 3121.3 11.27 St. Dev. 2353.13 1959.93 12.46

PAGE 98

98 Table 4 8 Ventral Anterior Nucleus of the Thalamus (Tract Edge Weights) Posterior Thalamic Lesion Anterior Nucleus (Tract Edge Weights) Participant Pars Opercularis Ventral Anterior Nucleus Tract Edge Weight Pars Opercularis Ventral Anterior Nucleus Track Edge Weight Post Posterior Thalamic Le sion % Change in Tract Edge Weight Post Posterior Thalamic Lesion (%) 1 3.77x10^ 3 3.57x10^ 3 5.31 2 2.22x10^ 5 2.22x10^ 5 0 3 5.31x10^ 5 5.22x10^ 5 1.70 4 2.58x10^ 6 2.58x10^ 6 0 5 5.87x10^ 5 5.50x10^ 5 6.30 6 8.89x10^ 4 8.89x10^ 4 0 7 2.51x10^ 4 2.50x10^ 4 0.40 8 3.91x10^ 5 3.91x10^ 5 0 9 3.42x10^ 4 3.42x10^ 4 0 10 7.35x10^ 7 7.35x10^ 7 0 Average 5.43x10^ 4 5.22x10^ 4 1.37 St. Dev. 1.17x10^ 3 1.1x10^ 3 2.41 Participant Pars Triangularis Ventral Anterior Nucleus Tract Edge Weight Pars Triangularis Ventral Anterior Nucleus Tract Edge Weight Post Posterior Thalamic Lesion % Change in Tract Edge Weight Post Posterior Thalamic Lesion (%) 1 1.57x10^ 3 1.57x10^ 3 0 2 2.57x10^ 3 2.56x10^ 3 0.40 3 1.60x10^ 4 1.59x10^ 4 0.63 4 4.39x10^ 3 4.38x10^ 3 0.23 5 4.25x10^ 5 4.12x10^ 5 3.06 6 6.85x10^ 3 6.85x10^ 3 0 7 7.85x10^ 4 7.82x10^ 4 0.38 8 3.03x10^ 4 3.02x10^ 4 0.33 9 1.77x10^ 3 1.76x10^03 0.57 10 2.29x10^ 6 2.29x10^ 6 0 Average 1.84x10^ 3 1.84x10^ 3 0.56 St. Dev. 2.24x10^ 3 2.24x10^ 3 0.91

PAGE 99

99 Table 4 9 Pulvinar (Tract Volumes) (Tract Volumes) Participant Pars Opercularis Pulvinar Tract Volume (mm 3 ) Pars Opercularis Pulvinar Track Volume Post Posterior Thalamic Lesion (mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 9826 4412 55.1 2 2181 561 74.28 3 2166 1375 36.52 4 5877 514 91.25 5 3075 1659 46.1 6 1748 702 59.84 7 4596 2910 36.68 8 2741 1754 36 9 4280 1307 69.46 10 765 256 66.54 Average 3725.5 1545 57.17 St. Dev. 2620.33 1274.9 18.63 Participant Pars Triangularis Pulvinar Tract Volume (mm 3 ) Pars Triangularis Pulvinar Tract Volume Post Posterior Thalamic Lesion (mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 8319 2726 67.23 2 5193 2956 43.1 3 3316 2716 18.1 4 7908 3937 50.22 5 1740 1254 27.93 6 2354 840 64.32 7 6533 3584 45.14 8 5637 3527 37.43 9 5349 1484 72.26 10 3292 1499 54.47 Average 4964.1 2452.3 48.02 St. Dev. 2254.78 1101.52 17.36

PAGE 100

100 Table 4 10 and Pulvinar (Tract Edge Weights) (Tract Edge Weights) Participant Pars Opercularis Pulvinar Tract Edge Weight Pars Opercularis Pulvinar Track Edge Weight Post Posterior Thalamic Lesion % Change in Tract Edge Weight Post Anterior Thalamic Lesion (%) 1 5.41x10^ 3 1.68x10^ 4 96.90 2 7.39x10^ 6 6.35x10^ 7 91.41 3 2.91x10^ 5 1.29x10^ 5 55.67 4 5.19x10^ 5 3.02x10^ 7 99.42 5 3.16x10^ 4 6.52x10^ 5 79.37 6 4.97x10^ 5 1.38x10^ 6 97.22 7 5.47x10^ 4 1.46x10^ 4 73.31 8 3.32x10^ 5 1.99x10^ 5 40.10 9 1.26x10^ 5 1.05x10^ 6 91.67 10 1.51x10^ 6 1.65x10^ 7 89.10 Average 6.46x10^ 4 4.16x10^ 5 81.41 St. Dev. 1.68x10^ 3 6.42x10^ 5 19.78 Participant Pars Triangularis Pulvinar Tract Edge Weight Pars Triangularis Pulvinar Tract Edge Weight Post Posterior Thalamic Lesion % Change in Tract Edge Weight Post Anterior Thalamic Lesion (%) 1 1.72x10^ 3 1.12x10^ 5 99.35 2 1.00x10^ 3 9.10x10^ 5 90.90 3 7.29x10^ 4 7.12x10^ 4 2.33 4 4.69x10^ 3 3.44x10^ 4 92.67 5 1.31x10^ 4 5.21x10^ 5 60.23 6 1.83x10^ 4 7.38x10^ 5 59.67 7 8.98x10^ 4 2.50x10^ 4 72.16 8 2.08x10^ 4 7.96x10^ 5 61.73 9 3.39x10^ 3 5.64x10^ 5 98.34 10 3.24x10^ 4 1.70x10^ 4 47.53 Average 1.33x10^ 3 1.84x10^ 4 68.49 St. Dev. 1.54x10^ 3 2.12x10^ 4 29.72

PAGE 101

101 Table 4 1 1 Ventral Anterior Nucleus, and Pulvinar (Tract Volumes). Area, Ventral Anterior Nucleus, and Pulvinar (Tract Volumes) Participant Pars Opercularis Ventral Anterior Nucleus Pulvinar Tract Volume (mm 3 ) Pars Opercularis Ventral Anterior Nucleus Pulvinar Track Volume Post Posterior Thalamic Lesion (mm 3 ) % Ch ange in Tract Volume Post Anterior Thalamic Lesion (%) 1 3599 1348 62.55 2 268 0 100 3 763 724 5.11 4 178 0 100 5 1261 232 81.6 6 0 0 0 7 1978 213 89.23 8 941 147 84.38 9 189 0 100 10 539 0 100 Average 971.6 266.4 72.29 St. Dev. 1101.15 441.26 38.62 Participant Pars Triangularis Ventral Anterior Nucleus Pulvinar Tract Volume (mm 3 ) Pars Triangularis Ventral Anterior Nucleus Pulvinar Tract Volume Post Posterior Thalamic Lesion (mm 3 ) % Change in Tract Volume Post Anterior Thalamic Lesion (%) 1 2564 807 68.53 2 2425 291 88 3 1239 390 68.52 4 3027 655 78.36 5 675 284 57.93 6 845 0 100 7 3325 209 93.71 8 1859 495 73.37 9 2053 0 100 10 1915 219 88.56 Average 1992.7 335 81.70 St. Dev. 882.37 260.8 14.51

PAGE 102

102 Table 4 1 2 Ventral Anterior Nucleus, and Pulvinar (Tract Edge Weights). Anterior Nucleus, and Pulvinar (Edge Weights) Participant Pars Opercularis Ventral Anterior Nucleus Pulvinar Tract Edge Weight Pars Opercularis Ventral Anterior Nucleus Pulvinar Track Edge Weight Post Posterior Thalamic Lesion % Change in Tract Edge Weight Post Anterior Thalamic Lesion (%) 1 1.06x10^ 4 1.68x10^ 5 84.15 2 2.92x10^ 7 0 100 3 2.62x10^ 6 1.16x10^ 6 55.73 4 9.89x10^ 8 0 100 5 6.00x10^ 6 2.89x10^ 6 51.83 6 0 0 0 7 3.05x10^ 5 7.09x10^ 8 99.77 8 2.74x10^ 6 1.22x10^ 7 95.55 9 2.29x10^ 7 0 100 10 5.18x10^ 7 0 100 Average 1.49x10^ 5 2.10x10^ 6 78.70 St. Dev. 3.33x10^ 5 5.25x10^ 6 33.36 Participant Pars Triangularis Ventral Anterior Nucleus Pulvinar Tract Edge Weight Pars Triangularis Ventral Anterior Nucleus Pulvinar Tract Edge Weight Post Posterior Thalamic Lesion % Change in Tract Edge Weight Post Anterior Thalamic Lesion (%) 1 8.37x10^ 5 3.41x10^ 5 59.26 2 3.41x10^ 5 2.36x10^ 7 99.31 3 3.30x10^ 5 4.34x10^ 6 86.85 4 2.57x10^ 4 1.59x10^ 6 99.38 5 4.77x10^ 6 3.00x10^ 6 37.11 6 3.13x10^ 6 0 100 7 5.29x10^ 5 9.84x10^ 8 99.81 8 3.52x10^ 5 1.47x10^ 6 95.82 9 3.08x10^ 5 0 100 10 2.62x10^ 5 8.14x10^ 6 68.93 Average 5.61x10^ 5 5.30x10^ 6 84.65 St. Dev. 7.42x10^ 5 1.04x10^ 5 22.15

PAGE 103

103 Table 4 13. Effects of Anterior and Thalamocortical Circuitry. Circuitry Anterior Thalamic Lesion Posterior Thalamic Lesion % Change in Tract Volumes % Change in Edge Weight Values % Change in Tract Volumes % Change in Edge Weight Values Pars Opercularis Ventral Anterior Nucleus Pathways 56.98 (30.65) 45.56 (38.01) 13.64 (21.11) 1.37 (2.41) Pars Triangularis Ventral Anterior Nucleus Pathways 45.80 (26.17) 54.92 (39.91) 11.27 (12.46) 0.57 (0.91) Pars Opercularis Pulvinar Pathways 13.39 (12.37) 2.51 (3.49) 57.17 (18.63) 81.41 (19.78) Pars Triangularis Pulvinar Pathways 19.62 (20.85) 13.50 (29.24) 48.02 (17.36) 68.49 (29.72) Pars Opercularis Vent ral Anterior Nucleus Pulvinar Pathways 58.00 (39.71) 48.14 (39.67) 72.29 (38.62) 78.70 (22.26) Pars Triangularis Ventral Anterior Nucleus Pulvinar Pathways 53.37 (26.59) 57.44 (35.20) 81.70 (14.51) 84.65 (22.15)

PAGE 104

104 Figure 4 1. Effects of an and the thalamus. Row (a) represents pathways between pars opercularis and pars triangularis and ventral anterior nucleus before an anterior lesion (a1 and a2) and after the lesion (a3 and a4) f or participant 1. Images a3 and a4 are blank because the lesion eliminated all of the pathways connecting represents pathways between pars opercularis and pars triangularis and pulvina r before (b1 and b2) and after the lesion (b3 and b4). Image b3 is blank because the lesion eliminated all of the pathways connecting pars opercularis and ventral anterior nucleus in this participant. Row (c) represents pathways connecting pars opercularis and pars triangularis with ventral anterior nucleus and pulvinar before (c1 and c2) and after an anterior lesion (c3 and c4).

PAGE 105

105 Figure 4 2. Anterior thalamic lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) and ventral anterior nucleus (pink). The lesion is depicted in black and in most cases it surrounds the ventral anterior nucleus. Images 1a, 1b, and 2a are blank because the lesion eliminated all of the pathways connecting pars opercularis and/or pars triangul aris and ventral anterior nucleus is these participants.

PAGE 106

106 Figure 4 3. Anterior thalamic lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) and pulvinar (purple). The lesion is depicted in black and is located anterior to the pulvinar. Image 1a is blank because the lesion eliminated all of the pathways connecting pars opercularis and pulvinar in this participant.

PAGE 107

107 Figure 4 4. Anterior thalamic lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) with ventral anterior nucleus (pink) and pulvinar (purple). The lesion is depicted in black and in most cases surrounds the ventral anterior nucleus. Image 2a, 2b, 4a, 6a, and 9b are blank because the lesion eliminated all of the pathways connecting pars opercularis and/or pars triangularis with ventral anterior nucleus and pulvinar in these participants.

PAGE 108

108 Figure 4 5. and the thalamus. Row (a) represents pathways be tween pars opercularis and pars triangularis and ventral anterior nucleus before a posterior lesion (a1 and a2) and after the lesion (a3 and a4) for participant 1. Row (b) represents pathways between pars opercularis and pars triangularis and pulvinar befo re (b1 and b2) and after the lesion (b3 and b4). Row (c) represents pathways connecting pars opercularis and pars triangularis with ventral anterior nucleus and pulvinar before (c1 and c2) and after a posterior lesion (c3 and c4).

PAGE 109

109 Figure 4 6. Posterior thalamic lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) and ventral anterior nucleus (pink). The lesion is depicted in black and is located posterior to the ventral anterior nucleus.

PAGE 110

110 Figure 4 7. Posterior lesion ef fects on pathways connecting pars opercularis (a) and pars triangularis (b) with pulvinar (purple). The lesion is depicted in black and in most cases surrounds the anterior superior portion of the pulvinar.

PAGE 111

111 Figure 4 8. Posterior thalamic lesion effects on pathways connecting pars opercularis (a) and pars triangularis (b) with ventral anterior nucleus (pink) and pulvinar (purple). The lesion is depicted in black and in most cases surrounds the anterior superior portion of the pulvinar. Image 2a, 2b, 4a, 6 a, 6b, 9a, 9b, 10a are blank because the lesion eliminated all of the pathways connecting pars opercularis and/or pars triangularis with ventral anterior nucleus and pulvinar in these participants.

PAGE 112

112 CHAPTER 5 DISCUSSION try: Functional Implications thalamus. In particular, using diffusion weighted tractography we inferred pathways between pars opercularis/triangularis, ventral anterior nucleus of the thalamus, and pulvinar. Our results show that both pars opercularis and pars triangularis share direct connectivity with these thalamic nuclei. In addition, we present pathways en passage that appear to connect each of these cortical zones with the pul vinar passing through area with the thalamus: pars opercularis/triangularis ventral anterior nucleus pathway, pars opercularis/triangularis pulvinar pathway, and pars opercularis/triangularis ventral anterior nucleus pulvinar pathway. Invasive animal tracer studies previously established the presence of some of these pathways in macaque. In particular, Romanski and colleagues demonstrated projections between macaque BA 45 and 46 and central/lateral medial pulvinar complex (Romanski et al., 1997). Akert and Hartmann Von Manakov demonstrated dense reciprocal connections between macaque areas 8 and 45, ventral anterior nucleus, and medial pulvinar (Akert and Hartmann Von Ma nakov, 1980), while Goldman Rakic and Porrino described pathways between macaque prefrontal cortex, medial dorsal nucleus, ventral anterior nucleus, and pulvinar (Godlman Rakic and Porrino, 1985). In addition, our prior work demonstrated direct connectivit y between ventral anterior nucleus and pars opercularis/triangularis in humans using diffusion weighted tractography (Ford et al., in review). The present study

PAGE 113

113 area wi th both ventral anterior nucleus and pulvinar. Our findings demonstrating direct structural connectivity between a language for thalamic involvement in language processin g. This involvement has also been implied by electorophysiological, neuroimaging, and lesion studies as reviewed in Chapter 1. Specifically, ventral anterior nucleus has been implicated in supporting speech production as determined by electric stimulation of this region during neurosurgery (Schaltenbrand 1965, 1975). In addition, ventral anterior thalamus is important for verbal short term memory as stimulation of this nucleus during encoding reduces subsequent recall errors (Johnson and Ojemann, 2000). Les ions to this region often result in deficits in speech production, aphasia, and category specific naming deficits (Graff Radford et al., 1985; Alexander and LoVerme, 1980; Nadeau & Crosson, 1997; Raymer et al., 1997). Results from our study show that both pars opercularis and pars triangularis share direct structural connectivity with the ventral anterior nucleus. A closer examination of these pathways indicates that both cortical regions project to a topographically similar and largely overlapping location within the nucleus (Figure 3 1). Pathways from pars opercularis/triangularis travel medially, bend around the circular sulcus, and enter the ventral anterior nucleus within its anterior superior portion. Converging pathways connecting distinct cortical zo nes, supporting similar functional processes, with the same thalamic nucleus have been previously identified and studied in detail (Theyel et al., 2010). In particular, Theyel and colleagues applied electrical stimulation to a slice of a mouse brain contai ning primary somatosensory (S1),

PAGE 114

114 secondary somatosensory (S2) cortical areas, and posteromedial nucleus of the thalamus. Importantly, direct connections between S1 and S2 were cut to eliminate cortico cortical interaction. The results showed that stimulati on applied to S1 activated posteromedial thalamus and S2. When posteromedial thalamus was inactivated by a chemical agent, stimulation in S1 did not result in S2 activation. This finding may indicate that corticothalamo cortical relays are able to transfer information from one cortical zone to the other (Theyl et al., 2010; Sherman and Guillery 2006 ; Llano and Sherman 2008). In the present study influence whether thalamic relay neurons are in a high or low fidelity transfer mode or may transfer information between them via thalamic nuclei. These types of cortico thalamo cortical mechanisms within the language network may be important for both production and comprehension. In discourse, pars triangularis may formulat e the appropriate semantic content and syntax, then enhance its phonological processing and articulatory representation through activity modulation within pars opercularis. Pars opercularis, on the other hand, may modulate activity within pars triangularis during speech comprehension to ensure that phonological representations of lexical items are matched to appropriate semantic stores. The pulvinar is another thalamic nucleus that has been implicated to be involved in language processing. Electrical stim ulation of the anterior superior pulvinar results in object naming difficulties, while retaining ability to speak (Johnson and Ojeman 2000). Lesions to the dominant pulvinar result in object naming difficulties, aphasia, and category specific naming diffic ulties, while speech remains relatively fluent (Van Buren 1975; Alexander and LoVerme, 1980; Crosson et al., 1986, 1997). In addition, growing

PAGE 115

115 evidence suggests that the dominant thalamus, and in particular pulvinar, may be involved in recruitment of langu age eloquent cortical regions during lexical semantic processing (Crosson, 2012). In his recent paper reviewing thalamic involvement in language processing and models of thalamic aphasia, Crosson proposed that pulvinar may be important for selectively recr uiting cortical regions that store multimodal features of a given lexical item and binding them together during language processing (Crosson, 2012). Indeed, if we consider what constitutes our knowledge of a given concept, for example, a dog, we can recall its defining visual characteristics (four legs, tail, fur, and typical motion patterns), characteristic auditory signature (barking), and tactile and odor associations. All of these features are stored across many different cortical areas. For example, co ncepts of living objects are stored in the lateral fusiform gyrus, whereas non living objects are represented within the medial fusiform gyrus within the dominant hemisphere (Wierenga et al., 2009). Visual representation of movement is likely to be stored within the middle temporal (MT) area (Beauchamp and Martin, 2007; Born and Bradley, 2005). Thus, during language comprehension or production all of the features residing in multiple cortical areas must bind together successfully into a semantic representat ion that will allow us to retrieve the corresponding lexical item. Anatomically, the most plausible candidate to accomplish this feature binding process is the thalamus as it shares extensive connectivity with frontal, temporal, parietal, and occipital cor tex (Behrens et al., 2003b). In fact, work by Kraut and colleagues revealed thalamic activation for word pairs that bind together to produce a third item from its features (Kraut et al., 2002). This hypothesis is further supported by electrophysiological e vidence where feature binding processing has been characterized by drop in low

PAGE 116

116 frequency oscillations within the thalamus and cortex, followed by fast rhythm (20 60 Hz) oscillations in the thalamus and select cortical zones (Slotnick et al., 2002). Nadeau increase activity within other cortical areas pertinent to a given task through its c onnections with thalamic nucleus reticularis. Nucleus reticularis, in turn, influences activity state of a given thalamic nucleus and cortical processors with which it is connected (Nadeau and Crosson 1997). Although originally the model proposed that the centromedian nucleus was the key thalamic nucleus influencing activity of the cortex, recent anatomical evidence suggests that this nucleus lacks widespread cortical connectivity necessary to accomplish this task (Sadikot and Rymar, 2009). In light of this evidence, and the evidence demonstrating its connectivity with frontal, posterior temporal, and inferior parietal cortical zones, Crosson proposed that the pulvinar may be the nucleus involved in selective engagement (Goldman Rakic and Porrino, 1985; Cros son, 2012). The present study further supports this hypothesis by demonstrating connectivity between pulvinar and language findings suggest that projections between pars opercularis/triangularis travel medially, be nd around the circular sulcus and thence course posteriorly towards the pulvinar (Figure 3 3). Pathways from pars opercularis and pars triangularis enter the anterior superior portion of the nucleus. Interestingly, this portion of the nucleus is topographi cally similar to a location previously identified by Ojemann and Johnson to be involved in object naming (Ojemann and Johnson, 2000). Direct connectivity between

PAGE 117

117 language proc essing involved in this task. In object naming an individual must correctly match features of an object (either visual characteristics if matching to a picture, or area may alter activity within pulvinar (directly and via nucleus reticularis) to modulate activity within this nucleus and the many cortical zones to which it projects. Semantic c ortical recruitment to enhance activation only in those cortical areas that store semantically appropriate features. Once all of the features are activated they bind together to enhance activation of lexical items that match the semantic feature content. In this to enable feature binding and lexical search processing crucial for language th of these nuclei may allow for intrathalamic information transfer to ensure coherence between linguistic processes supported by ventral anterior nucleus and pulvinar. Circuit ry And Corresponding Language Deficits Anterior Thalamic Lesion Effects thalamus and investigated effects of prototypical anterior and posterior thalamic lesions on this connectivity. Results reported in Chapter 4 of this manuscript describe differential thalamus. Our main findings indicate that anterior thalamic lesions create the largest pulvinar. T hese pathways sustain more than 50% reduction in tract volume and edge

PAGE 118

118 with ventral anterior nucleus sustained the second largest amount of damage as indicated by about 5 0% reduction in tract volume and edge weight measures. In contrast, anterior thalamic lesions result in least amount of damage to pathways connecting pars opercularis/triangularis with pulvinar, where we observed no more than 20% reductions in tract volume s and edge weights following the lesion. We believe that the differential deficits produced by lesions in each of these pathways results from the relative proximity of the lesion to each thalamic nucleus. In particular, this anterior lesion model was locat ed anteriorly to the ventral anterior nucleus and extended posteriorly beyond the posterior border of the nucleus. In most cases this anterior lesion engulfed most of the surface area of the ventral anterior nucleus leaving only the most superior portion u naffected. As described in Chapter 3 projections from pars opercularis/triangularis enter the anterior superior portion of the ventral anterior nucleus. Given this trajectory this anterior thalamic lesion interrupts most of these pathways by affecting a la rge portion of the projection site. Pathways that are preserved following this anterior lesion project to the most posterior aspect of the anterior superior portion of the ventral anterior nucleus. Similarly, pathways connecting pars opercularis/triangular is with ventral anterior nucleus and pulvinar also target the anterior superior portion of the ventral anterior nucleus. This pathway is much smaller in volume than the two other rable to thalamic lesions located in close proximity to either nucleus. Pathways connecting pars opercularis/triangularis and pulvinar sustained the least amount of damage following this anterior thalamic lesion. In all of our ten

PAGE 119

119 participants the lesion mask was located anterior and inferior to pulvinar and it did not reach the anterior border of the pulvinar. The pathways connecting pars opercularis/triangularis and pulvinar target the anterior superior portion of the pulvinar and most of the tracts with in this bundle are located superior to the anterior lesion mask. Thus, in all ten cases the anterior lesion affects only the most inferior tracts within this pathway resulting in less than one fifth reduction in tract volumes and edge weights. What could be some of the potential functional deficits of anterior thalamic and thalamus? To answer this question we must recall the case of patient presented by Raymer and colle agues (Raymer at el., 1997), which was used here as the guide to our anterior thalamic lesion model. This patient suffered a tuberothalamic artery infarction and presented with fluent speech full of semantic substitutions, intact repetition and comprehensi on. However, the main deficit exhibited by the patient was poor oral and written picture naming, as well as oral naming to auditory definition. The authors concluded that following an anterior thalamic lesion the patient suffered impairment in retrieval of lexical items based on semantic information (Raymer et al., 1997; Crosson 1997). Based on results found in the present study we can hypothesize that this patient sustained considerable amount of damage to pars opercularis/triangularis ventral anterior nuc leus pulvinar pathways, as well as pathways connecting pars opercularis/triangularis with ventral anterior nucleus. Also based on our findings, pathways connecting pars opercularis/triangularis with pulvinar were likely to sustain the least amount of damag thalamocortical circuitry could potentially explain language impairments observed in this

PAGE 120

120 patient. In particular, earlier in our discussion of functional significance of these pathways we proposed t nucleus may be influencing lexical search mechanisms supported by the pre SMA basal ganglia ventral anterior nucleus loop (Kraut et al., 2003). An anterior thalamic lesion affecting ventral anteri ventral anterior nucleus circuitry (as observed in the present study), as well as within the pre SMA ventral anterior nucleus pathways. These structural deficits are likely to impair lexical se arch processing supported by these networks resulting in difficulties in object naming both from pictures and from auditory definitions. In addition, given our pulvinar connectivity results in minimal deficits following anterior thalamic lesions, we can hypothesize that feature binding processing remains relatively well preserved. This observation is supported by finding of Raymer and colleagues that their patient was able to match auditory and written forms of objects t o correct pictures (Raymer et al., 1997). It is likely that for this patient, their anterior thalamic lesion resulted in small impairment of the pars opercularis/triangularis pulvinar pathways leaving semantic feature binding per se relatively intact. Matc hing the auditory or a written form of an object to its picture would plausibly involve activation of a results in activity modulation within the pulvinar and cortical zones storing semantically related features of the object. Since anterior thalamic lesions result in least amount of damage to these pathways, it is plausible that this feature binding proce ss remains intact in this case. However, once semantic feature binding is complete and lexical items that match the

PAGE 121

121 features are activated within lexical stores impairments within the lexical search/selection system leave the patient unable to select appro priate lexical item. It is plausible that anterior thalamic lesions result in inability to suppress competing lexical items and enhance the best fitting item due to the structural damage within the pre SMA basal ganglia ventral anterior thalamus loop. This damage results in decreased signal to noise ratio within the language system resulting in inability to access correct paraphasias during discourse. It could be the case th at once a particular semantic category is activated through the selective engagement mechanism the patient is unable to select the most appropriate lexical item and instead selects one of the competing items within the same category. Posterior Thalamic Les ion Effects Posterior thalamic lesions also resulted in differential amount of deficits within in the case with anterior thalamic lesions, pathways connecting pars operc ularis ventral anterior nucleus pulvinar sustain the largest amount of damage following a posterior thalamic lesion. In our study, this lesion resulted in more than 70% reduction in tract volumes and edge weight values within this pathway. In addition, we observed more than 50% reduction in tract volumes and edge weight values within pathways area with the ventral anterior nucleus showed the least amount of damage amounting to no more than 14% reduction in tract volumes and edge weight values. Our posterior thalamic lesion model was located anterior to the pulvinar affecting its anterior superior extent. As noted in Chapter 3, projections between pars

PAGE 122

122 opercularis/triangulari s target this particular portion of the nucleus as their projection site. It is thus not surprising that we observed large reductions in tract volume and edge weight values within this circuitry as most of the pathways within this bundle are affected by th e lesion. Pathways connecting pars opercularis/triangularis with ventral anterior nucleus and pulvinar also project to the anterior superior portion of the pulvinar. This pathway likely sustained the greatest amount of damage due to its small tract volume and highly convergent projection pattern within the pulvinar. The posterior thalamic lesion model was located posterior to the ventral anterior nucleus in all of our participants. The lesion only affected the most posterior aspects of the nucleus leaving the main projection to the ventral anterior nucleus projection site relatively unaffected. Thus, most of the pathways connecting pars opercularis/triangularis with ventral anterior nucleus remained preserved following a posterior thalamic lesion. Potential functional implications of differential effects of posterior thalamic lesions deficits of thalamic aphasia cases following damage within posterior thalamus (Crosson et al. 1986, 1997). One of the first such cases was presented by Crosson and colleagues in 1986 where a patient presented with a number of language deficits following lesion to the dorsal pulvinar (including anterior superior portion of the nucleus). Acutely, the patient had fluent speech with frequent semantic paraphasias and word finding difficulties. In addition, the patient exhibited poor comprehension and significant amount of semantic substitutions during reading. Chronically, the most evident language de ficits consisted of word finding difficulties and semantic substitutions during

PAGE 123

123 suffered substantial amount of damage to pathways connecting pars opercularis/tria ngularis, ventral anterior nucleus, and pulvinar as well as pars opercularis/triangularis pulvinar pathways. It is plausible that partial interruption of pars opercularis/triangularis pathways results in functional deficits within the selective engagement network. In discourse or during language comprehension semantic, influence selection of cortical zones storing lexical items based on the most relevant content. Interruption of activate only some of the semantically relevant lexical items. Under these conditions, the best matchin g lexical item may not be activated limiting lexical search and selection to semantically competing items. This may explain the presence of semantic paraphasias and word access to the best m atching lexical item may be restricted or fully hindered forcing the individual to select between semantically related items. In addition, if connections between pulvinar and cortical zones storing lexical items (for example connections between pulvinar an d fusiform gyrus, or middle temporal area) are completely severed by the lesion, an individual may not be able to access an entire semantic category of lexical items. Such an impairment was observed by Crosson and colleagues in 1997 when they presented the case of a patient suffering circumscribed anomia for man made medical items and medical conditions (Crosson et al., 1997). The patient suffered

PAGE 124

124 damage to his dominant pulvinar and chronically was unable to produce items within this semantic category even when given phonemic cues to ease recall. Taken together, language deficits reported in these two cases and results from the present study indicate that posterior thalamic lesions may interrupt the selective ea pulvinar circuitry. Specifically, it appears that posterior thalamic lesions partially interrupt these connections and may result in decreased efficiency of the selective engagement mechanism. This decreased efficiency may result in only partial activat ion of semantically relevant categories and could leave the best matching lexical items inactivated. Lexical selection mechanism ventral anterior nucleus circuitry and by the pre SMA basal ganglia ventral anterior nucleus loop remains relatively spared by these lesions. However, since the pulvinar based selective engagement mechanism is unable to activate all of the semantically relevant items, lexical output selection is made from a limited sample. This may result in selection of a semantically related but not the best matching lexical item (i.e. semantic paraphasia). In addition, if a patient suffers damage to pathways between pulvinar and cortical zones storing lexical items this may conceivably result in category specific na ming deficits. In this case the selective engagement mechanism can be postulated to lose access to an entire semantic category of lexical items that would result in circumscribed anomia similar to that observed by Crosson and colleagues in their 1997 thala mic aphasia case. Study Limitations and Future Directions thalamus is not without limitations. In particular, we used anatomical landmarks to identify sub ea and thalamic nuclei and used these regions of

PAGE 125

125 interest to infer corticothalamo cortical connectivity. Cortical gross anatomy based on the sulcal and gyral patterns often does not maintain a one to one relationship with the underlying cytoarchitecture an d functional activations (Amunts et al., 1999). Future studies employing functional tasks that would identify sub based on functional activations (i.e. semantic task versus phonological task) would provide additional insights in to functional significance of networks identified in the present study. In addition, applications of dynamic causal modeling (DCM) may help to thalamic versus thalamo cortical, ver sus cortico thalamo cortical) as current tractography approaches do no allow us to make any inferences regarding directionality of a pathway. Our results from Chapter 3 showed a considerable amount of individual variability in track volumes and edge weigh nuclei. We believe that this variability may be stemming from a number of different factors. First, this variability may be due to individual differences in white matter organization. This hypothesis is supported by findings that individual patterns of gyral and sulcal distributions as well as the underlying cytorachitectonic divisions vary a lot between different individuals (Amunts et al., 1999). Given that our tractography approach was guided by corti cal and subcortical grey matter regions of interest variability in size and location of these regions introduces variability in trajectories and size of the resulting white matter pathways. Another potential explanation for variability in tract volumes a nd edge weights reported in this study is susceptibility of diffusion weighted data to motion related

PAGE 126

126 artifacts. The inherent resolution of our diffusion weighted data imposes considerable constraints on the amount of motion that could be present in a data set without affecting tractography results. Our acquisition resolution in this study was 2x2x2 mm 3 which suggests that any movement larger than 2 mm would result signal averaging between neighboring voxels. Spat ial signal averaging is further enhanced by motion correction algorithms that attempt to correct spatial misalignments by registering diffusion weighted volumes to each other. If a participant moves by 2 mm or more during acquisition of one or more diffusion weighted data volumes the final motion co rrected dataset could contain substantial amount of spatial averaging from the affected volumes. Unfortunately, since all of our participants were scanned without sedation and tion in the range of 2 5 mm is unusual. Data acquisition of excised tissue would eliminate motion related artifacts and we are currently working on acquiring and analyzing ex vivo diffusion weighted data from another set of participants. Another limitatio n inherent to this and other diffusion tractography studies is lack of explicitly defined null hypothesis (Jbabdi & Johansen Berg, 2011). Ideally, we would like to be able to perform formal statistical testing on results generated by a tracking algorithm c ontrolling for type I and type II errors. However, to accomplish this task we must first establish an appropriate null distribution within each data voxel that would support the null hypothesis (i.e. no tracts between region A and region B exist). Presentl y, no such distribution has been formulated and efforts are now being made to accomplish this task (Morris et al., 2008). In an effort to overcome this limitation in the present study, we performed an additional analysis examining whether our tracking

PAGE 127

127 meth od would generate a false positive result by attempting to trace pathways that are not believed to be present based on the animal tracer literature. We attempted to trace pathways between globus pallidus and pars triangularis/opercularis. Globus pallidus is not believed to share direct connectivity with the cortex, but rather these connections are mediated by the striatum on the input side and by the thalamus on the output side (Alexander et al., 1986; Middleton & Strick, 2000). In order to attempt to visualize these pathways we first generated all of the pathways that course through globus pallidus and intersected these pathways with pars triangularis and pars opercularis masks. As a result of this analysis, we found that our tracking method does not generate pathways directly connecting cortical regions of interest with globus pallidus. Although formal null hypothesis testing not presently available f or tractography studies, our results show that our tracking method is able to accurately infer known anatomical patterns of connectivity. In addition, the reader must keep in mind that although diffusion tractography has been shown to have good corresponde nce with our prior knowledge of structural organization of white matter within the brain based on either human post mortem dissection studies or invasive animal neuronal tracer studies, pathways delineated by a tracking algorithm do not represent actual ne ural fibers but rather our best estimate of fiber trajectories inferred based on local diffusion properties of the tissues. Animal tracer studies provide at least partial support for presence of the pathways presented in the current manuscript (Akert and H artmann Von Manakov, 1980; Romanski et al., 1997; Godlman Rakic and Porrino 1985). In particular, these tracer studies show direct connectivity between macaque areas 8, 45, and 46 with ventral anterior nucleus and

PAGE 128

128 pulvinar. Additional studies investigating with ventral anterior nucleus and pulvinar would provide further support for existence of this connectivity in humans. Another approach that proved to provide valuable insights into neural organization as a functi on of evolution are comparative studies investigating structural connectivity in different animal species such as macaques, chimpanzees, and bonobos (Rilling et al., 2008). It is plausible that uniqueness of fully developed language faculty may result in s pecies specific structural changes in humans as compared with our animal counterparts. Further investigation of connectivity of language eloquent cortex and its cytoarchitectural homologue in other specifies may enrich our knowledge of the evolution of lan guage within the neural system. The present study employed lesion models derived from previously published cases of thalamic aphasia to examine effects of thalamic lesions of circuitry connecting to quantify potential structural deficits resulting from each lesion it does not provide direct information about potential effects on language comprehension or production. We propose to further investigate effects of thalamic lesions on pathways presente d in the current study in actual cases of thalamic aphasia. This approach will allow us to correlate structural damage within these networks with specific symptoms experienced by the patient. In addition, we propose to access both structural (as measured b y tractography) and functional (as measured by clinical and neuropsychological testing) deficits acutely and chronically. Assessment and correlation of structural and functional deficits at two time points will allow us to measure changes within the networ k as a function of recovery.

PAGE 129

129 LIST OF REFERENCES Akert, K., Hartmann von Monakow, K., 1980. Relationships of precentral premotor and prefrontal cortex to the mediodorsal and intralaminar nuclei of the monkey thalamus. Acta Neurobiol Exp (Wars) 40, 7 25. Alexander, M.P., LoVerme, S.R., Jr., 1980. Aphasia after left hemispheric intracerebral hemorrhage. Neurology 30, 1193 1202. Amunts, K., Schleicher, A., Brgel, U., Mohlberg, H., Uylings, H.B., Zilles, K., 1999. Broca's region revisited: cytoarchitec ture and intersubject variability. J Comp Neurol 412, 319 341. Amunts, K., Weiss, P.H., Mohlberg, H., Pieperhoff, P., Eickhoff, S., Gurd, J.M., Marshall, J.C., Shah, N.J., Fink, G.R., Zilles, K., 2004. Analysis of neural mechanisms underlying verbal fluenc y in cytoarchitectonically defined stereotaxic space -the roles of Brodmann areas 44 and 45. Neuroimage 22, 42 56. Andersson, J.L.R., Jenkinson, M., Smith, S., 2007a. Non linear optimization. FMRIB technical report TR07JA2, www.fmrib.ox.ac.uk/analysis/techrep Andersson, J.L.R., Jenkinson, M., Smith, S., 2007b. Non linear registration, aka Spatial normalizaiton. FMRIB technical report TR07JA2, www.fmrib.ox.ac.uk/analysis/techrep Assaf, M., Calhoun, V.D., Kuzu, C.H., Kraut, M.A., Rivkin, P.R., Hart, J., Pearlson, G.D., 2006. Neural correlates of the object recall process in semantic memory. Psychiatry Res 147, 11 5 126. Basser, P.J., Mattiello, J., LeBihan, D., 1994. Estimation of the effective self diffusion tensor from the NMR spin echo. J Magn Reson B 103, 247 254. Beauchamp, M.S., Martin, A., 2007. Grounding object concepts in perception and action: evidence fr om fMRI studies of tools. Cortex 43, 461 468. Behrens, T.E., Berg, H.J., Jbabdi, S., Rushworth, M.F., Woolrich, M.W., 2007. Probabilistic diffusion tractography with multiple fibre orientations: What can we gain? Neuroimage 34, 144 155. Behrens, T.E., Wool rich, M.W., Jenkinson, M., Johansen Berg, H., Nunes, R.G., Clare, S., Matthews, P.M., Brady, J.M., Smith, S.M., 2003a. Characterization and propagation of uncertainty in diffusion weighted MR imaging. Magn Reson Med 50, 1077 1088. Behrens, T.E., Johansen B erg, H., Woolrich, M.W., Smith, S.M., Wheeler Kingshott, C.A., Boulby, P.A., Barker, G.J., Sillery, E.L., Sheehan, K., Ciccarelli, O., Thompson, A.J., Brady, J.M., Matthews, P.M., 2003 b Non invasive mapping of connections between human thalamus and cortex using diffusion imaging. Nat Neurosci 6, 750 757.

PAGE 130

130 Binder, J.R., Frost, J.A., Hammeke, T.A., Rao, S.M., Cox, R.W., 1996. Function of the left planum temporale in auditory and linguistic processing. Brain 119 ( Pt 4), 1239 1247. Binkofski, F., Amunts, K., S tephan, K.M., Posse, S., Schormann, T., Freund, H.J., Zilles, K., Seitz, R.J., 2000. Broca's region subserves imagery of motion: a combined cytoarchitectonic and fMRI study. Hum Brain Mapp 11, 273 285. Bornkessel, I., Zysset, S., Friederici, A.D., von Cram on, D.Y., Schlesewsky, M., 2005. Who did what to whom? The neural basis of argument hierarchies during language comprehension. Neuroimage 26, 221 233. Cabeza, R., Nyberg, L., 2000a. Neural bases of learning and memory: functional neuroimaging evidence. Cur r Opin Neurol 13, 415 421. Cabeza, R., Nyberg, L., 2000b. Neural bases of learning and memory: functional neuroimaging evidence. Curr Opin Neurol 13, 415 421. Cantalupo, C., Hopkins, W.D., 2001. Asymmetric Broca's area in great apes. Nature 414, 505. Catan i, M., 2006. Diffusion tensor magnetic resonance imaging tractography in cognitive disorders. Curr Opin Neurol 19, 599 606. Catani, M., Jones, D.K., ffytche, D.H., 2005. Perisylvian language networks of the human brain. Ann Neurol 57, 8 16. Catani, M., Thi ebaut de Schotten, M., 2008. A diffusion tensor imaging tractography atlas for virtual in vivo dissections. Cortex 44, 1105 1132. Chein, J.M., Fissell, K., Jacobs, S., Fiez, J.A., 2002. Functional heterogeneity within Broca's area during verbal working mem ory. Physiol Behav 77, 635 639. Colon Perez, L.M., Spindler, C., Goicohea, S., Triplett, W., Parekh, M., Montie, E., Carney, P.R., Mareci, T., 2012. Brain network metric derived from DWI, application to the limbic system. Proc Intl Soc Mag Reson Med 6, 686 Corballis, M.C., 2003. From mouth to hand: gesture, speech, and the evolution of right handedness. Behav Brain Sci 26, 199 208; discussion 208 160. Costafreda, S.G., Fu, C.H., Lee, L., Everitt, B., Brammer, M.J., David, A.S., 2006. A systematic review an d quantitative appraisal of fMRI studies of verbal fluency: role of the left inferior frontal gyrus. Hum Brain Mapp 27, 799 810. Crosson, B., 2012. Thalamic mechanisms in language: A reconsideration based on recent findings and concepts. Brain Lang.

PAGE 131

131 Crosso n, B., Benefield, H., Cato, M.A., Sadek, J.R., Moore, A.B., Wierenga, C.E., Gopinath, K., Soltysik, D., Bauer, R.M., Auerbach, E.J., Gokcay, D., Leonard, C.M., Briggs, R.W., 2003. Left and right basal ganglia and frontal activity during language generation : contributions to lexical, semantic, and phonological processes. J Int Neuropsychol Soc 9, 1061 1077. Crosson, B., McGregor, K., Gopinath, K.S., Conway, T.W., Benjamin, M., Chang, Y.L., Moore, A.B., Raymer, A.M., Briggs, R.W., Sherod, M.G., Wierenga, C.E. White, K.D., 2007. Functional MRI of language in aphasia: a review of the literature and the methodological challenges. Neuropsychol Rev 17, 157 177. Crosson, B., Moberg, P.J., Boone, J.R., Rothi, L.J., Raymer, A., 1997. Category specific naming deficit for medical terms after dominant thalamic/capsular hemorrhage. Brain Lang 60, 407 442. Crosson, B., Parker, J.C., Kim, A.K., Warren, R.L., Kepes, J.J., Tully, R., 1986. A case of thalamic aphasia with postmortem verification. Brain Lang 29, 301 314. Devlin J.T., Matthews, P.M., Rushworth, M.F.S., 2003. Semantic processing in the left inferior prefrontal cortex: A combined functional magnetic resonance imaging and transcranial magnetic stimulation study. J Cog Neurosci 15, 71 84. Draganski, B., Kherif, F., Klppel, S., Cook, P.A., Alexander, D.C., Parker, G.J., Deichmann, R., Ashburner, J., Frackowiak, R.S., 2008. Evidence for segregated and integrative connectivity patterns in the human Basal Ganglia. J Neurosci 28, 7143 7152. Fitch, W.T., Hauser, M.D., 200 4. Computational constraints on syntactic processing in a nonhuman primate. Science 303, 377 380. Ford, A., McGregor, K.M., Case, K., Crosson, B., White, K.D., 2010. Structural connectivity of Broca's area and medial frontal cortex. Neuroimage 52, 1230 1237. Frey, S., Campbell, J.S., Pike, G.B., Petrides, M., 2008. Dissociating the human language pathways with high angular resolution diffusion fiber tractography. J Neurosci 28, 11435 11444. Friederici, A.D., 2002. Towards a neural basis of auditory sentence processing. Trends Cogn Sci 6, 78 84. Friederici, A.D., 2004. Processing local transitions versus long distance syntactic hierarchies. Trends Cogn Sci 8, 245 247. Friederici, A.D., 2009. Pathways to language: fiber tracts in the human brain. Trend s Cogn Sci 13, 175 181. Friederici, A.D., Bahlmann, J., Heim, S., Schubotz, R.I., Anwander, A., 2006. The brain differentiates human and non human grammars: functional localization and structural connectivity. Proc Natl Acad Sci U S A 103, 2458 2463.

PAGE 132

132 Glass er, M.F., Rilling, J.K., 2008. DTI tractography of the human brain's language pathways. Cereb Cortex 18, 2471 2482. Goldman Rakic, P.S., 1995. Architecture of the prefrontal cortex and the central executive. Ann N Y Acad Sci 769, 71 83. Goldman Rakic, P.S. Porrino, L.J., 1985. The primate mediodorsal (MD) nucleus and its projection to the frontal lobe. J Comp Neurol 242, 535 560. Graff Radford, N.R., Damasio, H., Yamada, T., Eslinger, P.J., Damasio, A.R., 1985. Nonhaemorrhagic thalamic infarction. Clinical neuropsychological and electrophysiological findings in four anatomical groups defined by compu terized tomography. Brain 108 ( Pt 2), 485 516. Grafton, S.T., Arbib, M.A., Fadiga, L., Rizzolatti, G., 1996. Localization of grasp representations in humans by positron emission tomography. 2. Observation compared with imagination. Exp Brain Res 112, 103 111. Hagoort, P., Hald, L., Bastiaansen, M., Petersson, K.M., 2004. Integration of word meaning and world knowledge in language comprehension. Science 304, 438 441. Heim, S., Eickhoff, S.B., Amunts, K., 2008. Specialisation in Broca's region for semantic, phonological, and syntactic fluency? Neuroimage 40, 1362 1368. Heim, S., Friederici, A.D., Schiller, N.O., Rschemeyer, S.A., Amunts, K., 2009. The determiner c ongruency effect in language production investigated with functional MRI. Hum Brain Mapp 30, 928 940. Heiser, M., Iacoboni, M., Maeda, F., Marcus, J., Mazziotta, J.C., 2003. The essential role of Broca's area in imitation. Eur J Neurosci 17, 1123 1128. Hic kok, G., Poeppel, D., 2000. Towards a functional neuroanatomy of speech perception. Trends Cogn Sci 4, 131 138. Horwitz, B., Amunts, K., Bhattacharyya, R., Patkin, D., Jeffries, K., Zilles, K., Braun, A.R., 2003. Activation of Broca's area during the produ ction of spoken and signed language: a combined cytoarchitectonic mapping and PET analysis. Neuropsychologia 41, 1868 1876. Husband, E.M., Kelly, L.A., Zhu, D.C., 2011. Using complement coercion to understand the neural basis of semantic composition: evide nce from an fMRI study. J Cogn Neurosci 23, 3254 3266. Iacoboni, M., Woods, R.P., Mazziotta, J.C., 1996. Brain behavior relationships: evidence from practice effects in spatial stimulus response compatibility. J Neurophysiol 76, 321 331.

PAGE 133

133 Jbabdi, S., Johans en Berg, H., 2011. Tractography: where do we go from here? Brain Connect 1, 169 183. Jenkinson, M., Smith, S., 2001. A global optimisation method for robust affine registration of brain images. Med Image Anal 5, 143 156. Jian, B., Vemuri, B.C., Ozarslan, E ., Carney, P.R., Mareci, T.H., 2007. A novel tensor distribution model for the diffusion weighted MR signal. Neuroimage 37, 164 176. Johansen Berg, H., Behrens, T.E., Sillery, E., Ciccarelli, O., Thompson, A.J., Smith, S.M., Matthews, P.M., 2005. Functiona l anatomical validation and individual variation of diffusion tractography based segmentation of the human thalamus. Cereb Cortex 15, 31 39. Johnson, M.D., Ojemann, G.A., 2000. The role of the human thalamus in language and memory: evidence from electrophy siological studies. Brain Cogn 42, 218 230. Klein, J.C., Rushworth, M.F., Behrens, T.E., Mackay, C.E., de Crespigny, A.J., D'Arceuil, H., Johansen Berg, H., 2010. Topography of connections between human prefrontal cortex and mediodorsal thalamus studied wi th diffusion tractography. Neuroimage 51, 555 564. Kraut, M.A., Kremen, S., Segal, J.B., Calhoun, V., Moo, L.R., Hart, J., 2002. Object activation from features in the semantic system. J Cogn Neurosci 14, 24 36. Llano, D.A., Sherman, S.M., 2008. Evidence f or nonreciprocal organization of the mouse auditory thalamocortical corticothalamic projection systems. J Comp Neurol 507, 1209 1227. Makris, N., Pandya, D.N., 2009. The extreme capsule in humans and rethinking of the language circuitry. Brain Struct Funct 213, 343 358. McDermott, K.B., Petersen, S.E., Watson, J.M., Ojemann, J.G., 2003. A procedure for identifying regions preferentially activated by attention to semantic and phonological relations using functional magnetic resonance imaging. Neuropsychologi a 41, 293 303. Molnar Szakacs, I., Iacoboni, M., Koski, L., Mazziotta, J.C., 2005. Functional segregation within pars opercularis of the inferior frontal gyrus: evidence from fMRI studies of imitation and action observation. Cereb Cortex 15, 986 994. Morris, D.M., Embleton, K.V., Parker, G.J., 2008. Probabilistic fibre tracking: differentiation of connections from chance events. Neuroimage 42, 1329 1339. Nadeau, S.E., Crosson, B., 1997. Subcortical aphasia. Brain Lang 58, 355 402; discussion 418 323. N ishitani, N., Avikainen, S., Hari, R., 2004. Abnormal imitation related cortical activation sequences in Asperger's syndrome. Ann Neurol 55, 558 562.

PAGE 134

134 Nishitani, N., Hari, R., 2002. Viewing lip forms: cortical dynamics. Neuron 36, 1211 1220. Nyberg, L., Per sson, J., Habib, R., Tulving, E., McIntosh, A.R., Cabeza, R., Houle, S., 2000. Large scale neurocognitive networks underlying episodic memory. J Cogn Neurosci 12, 163 173. Ojemann, G., 1977. Language and verbal memory functions during and after human thala mic stimulation. Neurol Neurocir Psiquiatr 18, 35 45. Ojemann, G.A., 1983. Neurosurgical management of epilepsy: a personal perspective in 1983. Appl Neurophysiol 46, 11 18. Ojemann, G., Fedio, P., 1968. Effect of stimulation of the human thalamus and pari etal and temporal white matter on short term memory. J Neurosurg 29, 51 59. Ojemann, G.A., Van Buren, J.M., 1967. Respiratory, heart rate, and GSR responses from human diencephalon. Arch Neurol 16, 74 88. Parker, G.J., Luzzi, S., Alexander, D.C., Wheeler K ingshott, C.A., Ciccarelli, O., Lambon Ralph, M.A., 2005. Lateralization of ventral and dorsal auditory language pathways in the human brain. Neuroimage 24, 656 666. Rajkowska, G., Goldman Rakic, P.S., 1995. Cytoarchitectonic definition of prefrontal areas in the normal human cortex: I. Remapping of areas 9 and 46 using quantitative criteria. Cereb Cortex 5, 307 322. Rauschecker, J.P., Tian, B., 2000. Mechanisms and streams for processing of "what" and "where" in auditory cortex. Proc Natl Acad Sci U S A 97 11800 11806. Raymer, A.M., Moberg, P., Crosson, B., Nadeau, S., Rothi, L.J., 1997. Lexical semantic deficits in two patients with dominant thalamic infarction. Neuropsychologia 35, 211 219. Riegele, L., 1931. Die cytoarchitektonik der felder der brocasch en region. Journal fur Psychologie und Neurologie 42, 496 514. Rilling, J.K., Glasser, M.F., Preuss, T.M., Ma, X., Zhao, T., Hu, X., Behrens, T.E., 2008. The evolution of the arcuate fasciculus revealed with comparative DTI. Nat Neurosci 11, 426 428. Rizzo latti, G., Craighero, L., 2004. The mirror neuron system. Annu Rev Neurosci 27, 169 192. Rodd, J.M., Davis, M.H., Johnsrude, I.S., 2005. The neural mechanisms of speech comprehension: fMRI studies of semantic ambiguity. Cereb Cortex 15, 1261 1269.

PAGE 135

135 Romanski L.M., Giguere, M., Bates, J.F., Goldman Rakic, P.S., 1997. Topographic organization of medial pulvinar connections with the prefrontal cortex in the rhesus monkey. J Comp Neurol 379, 313 332. Rueckert, D., Sonoda, L.I., Hayes, C., Hill, D.L., Leach, M.O. Hawkes, D.J., 1999. Nonrigid registration using free form deformations: application to breast MR images. IEEE Trans Med Imaging 18, 712 721. Sadikot, A.F., Rymar, V.V., 2009. The primate centromedian parafascicular complex: anatomical organization with a note on neuromodulation. Brain Res Bull 78, 122 130. Schaltenbrand, G., 1965. The effects of stereotactic electrical stimulation in the depth of the brain. Brain 88, 835 840. Schaltenbrand, G., 1975. The effects on speech and language of stereotactical st imulation in thalamus and corpus callosum. Brain Lang 2, 70 77. Schaltenbrand, G., Spuler, H., Wahren, W., Rumler, B., 1971. Electroanatomy of the thalamic ventro oral nucleus based on stereotaxic stimulation in man. Z Neurol 199, 259 276. Schaltenbrand, G ., Wahren, W., 1977. Atlas for stereotaxy of the human brain. Georg Thieme Verlag, Rudigerstrasse 14, D 70469 Stuttgart. Sherman, S.M., Guillery, R.W., 2006. Exploring the thalamus. MIT Press, Cambridge, MA. Slotnick, S.D., Moo, L.R., Kraut, M.A., Lesser, R.P., Hart, J., 2002. Interactions between thalamic and cortical rhythms during semantic memory recall in human. Proc Natl Acad Sci U S A 99, 6440 6443. Sudhyadhom, A., Okun, M.S., Foote, K.D., Rahman, M., Bova, F.J., 2012. A Three dimensional Deformable B rain Atlas for DBS Targeting. I. Methodology for Atlas Creation and Artifact Reduction. Open Neuroimag J 6, 92 98. Theyel, B.B., Llano, D.A., Sherman, S.M., 2010. The corticothalamocortical circuit drives higher order cortex in the mouse. Nat Neurosci 13, 84 88. Ullman, M.T., 2004. Contributions of memory circuits to language: the declarative/procedural model. Cognition 92, 231 270. Van Buren, J.M., 1975. The question of thalamic participation in speech mechanisms. Brain Lang 2, 31 44. Wierenga, C.E., Perlstein, W.M., Benjamin, M., Leonard, C.M., Rothi, L.G., Conway, T., Cato, M.A., Gopinath, K., Briggs, R., Crosson, B., 2009. Neural substrates of object identification: Functional magnetic resonance imaging evidence that category and visual attribute co ntribute to semantic knowledge. J Int Neuropsychol Soc 15, 169 181.

PAGE 136

136 BIOGRAPHICAL SKETCH Anastasia Ford received her undergraduate degree in m athematics from University of Florida in 2008. She completed her undergraduate thesis on probability theory and its applications in medical imaging under guidance of Dr Murali Rao and received honors of Summa c um Laude upon graduation. Ms Ford completed her m d egree thesis in 2011 and received an Honorable Ment ion National Science Foundation Graduate Research Fellowship and became a Semi Finalist Recommended for an award under SMART from the Department of Defense for this work. Ms Ford completed three summer internships at the National Institute for Neurologica l Disease She received three Exceptional Summer Student Awards for her work at NINDS. In addition, Ms Ford selected twice as an outstanding graduate student in Cognit ive Psychology and received the E. Porter Horne Memorial Scholarship from the Psychology Department at the University of Florida. Ms Ford received her Ph.D. from the University of Florida in spring of 2013.