Anatomy of Aphasia Revisited (2018)

doi:10.1093/brain/awx363

BRAIN 2018: 141; 848–862

|

848

Anatomy of aphasia revisited Julius Fridrikss Fridriksson, on,1 Dirk-Bart den Ouden,1 Argye E. Hillis,2,3 Gregory Hickok,4 Chris Rorden,5 Alexandra Basilakos,1 Grigori Yourganov5 and Leonardo Bonilha 6

In most cases, aphasia is caused by strokes involving the left hemisphere, with more extensive damage typically being associated with more severe aphasia. The classical model of aphasia commonly adhered to in the Western world is the Wernicke-Lichtheim model. The model has been in existence for over a century, and classification of aphasic symptomatology continues to rely on it. However, far more detailed models of speech and language localization in the brain have been formulated. In this regard, the dual stream model of cortical brain organization proposed by Hickok and Poeppel is particularly influential. Their model describes two processi proc essing ng rout routes, es, a dors dorsal al stre stream am and a ven ventral tral stream, stream, that roug roughly hly supp support ort spee speech ch prod product uction ion and spe speech ech comp comprehe rehensio nsion, n, respectively, in normal subjects. Despite the strong influence of the dual stream model in current neuropsychological research, there has been relatively limited focus on explaining aphasic symptoms in the context of this model. Given that the dual stream modell repr mode represe esents nts a more nuanced nuanced pic picture ture of cor cortic tical al spee speech ch and lan langua guage ge orga organiz nizati ation, on, cort cortica icall dama damage ge that causes apha aphasic sic impairm impa irment ent shou should ld map cle clearl arlyy onto the dual processin processingg stre streams. ams. Here, we pre present sent a fol follow low-up -up stud study y to our prev previous ious work that used les lesion ion data to rev reveal eal the anat anatomi omical cal boundaries boundaries of the dors dorsal al and ven ventral tral streams streams supp supporti orting ng spe speech ech and lang languag uagee processing. processi ng. Specifically, Specifically, by emphasizing clinical measures, we examine the effect of cortic cortical al damage and disconne disconnection ction involving the dorsal and ventral streams on aphasic impairment. The results reveal that measures of motor speech impairment mostly involve damage dama ge to the dors dorsal al stre stream, am, whe whereas reas measures measures of impa impaire ired d spee speech ch com compreh prehens ension ion are more strongly strongly ass associ ociated ated with vent ventral ral stream involvement. Equally important, many clinical tests that target behaviours such as naming, speech repetition, or grammatical processing rely on interactions between the two streams. This latter finding explains why patients with seemingly disparate lesion locations often experience similar impairments impairments on given subtests. Namely, these individuals’ individuals’ cortic cortical al damage, although dissimilar, affects a broad cortical network that plays a role in carrying out a given speech or language task. The current data suggest this is a more accurate characterization than ascribing specific lesion locations as responsible for specific language deficits.

1 Department of Communication Sciences Sciences and Disorders, University of South Carolina, Carolina, Columbia, USA 2 Department of Neurology, Johns Hopkins Hopkins University School of Medicine; Department Department of Cognitive Science, Johns Hopkins University, Baltimore, USA 3 Department of Physical Medicine and Rehabilitation, Rehabilitation, Johns Hopkins School of Medicine; Department Department of Cognitive Science, Johns Hopkins University, Baltimore, MD, USA 4 Cogn Cognitive itive Sciences, Sciences, School of Socia Sociall Scien Sciences, ces, University University of Calif Californi ornia, a, Irvine, USA 5 Depar Departmen tmentt of Psychology, Psychology, University University of South Carolina, Carolina, Columbia, Columbia, USA 6 Department of Neurology, Medical University University of South Carolina, Charleston, South Carolina, Carolina, USA Correspondence to: Julius Fridriksson Professor and Endowed Chair, SmartState Department of Communication Sciences and Disorders University of South Carolina Columbia, USA E-mail: [email protected] Keywords: aphasia; imaging; rehabilitation; clinical practice; neuroanatomy Abbreviations: CLSM = connectome lesion-symptom mapping; PNT = Philadelphia Naming Test; RLSM = region-wise lesionsymptom mapping; SMG = supramarginal gyrus; STG = superior temporal gyrus Received January 28, 2017. Revised October 26, 2017. Accepted November 8, 2017. Advance Access publication January 17, 2018  The Author(s) (2018). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]

Downloaded from https://academic.oup.co https://academic.oup.com/brain/arti m/brain/article-abstract cle-abstract/141/3/848/4 /141/3/848/4816245 816245 by Rutgers University Libraries/Technical Libraries/Technical Services user

Anatomy of aphasia

Introduction Stroke is the most common cause of aphasia, with approximately mat ely 20– 20–40% 40% of all str stroke okess res result ulting ing in acu acute te aph aphasi asia a (Engelter et al., 2006). 2006). The general pattern of speech and language impairment that results from stroke is somewhat predictable as the type of aphasia is associated with specific lesion patterns (Yourganov ( Yourganov et al ., ., 2015). 2015). Despite the heterogeneity of lesion locations in people with the same aphasia type, there is enough similarity in lesion patterns within a given aphasia type to differentiate it from other aphasia types. The overall pattern of speech and language impairment is simil similar ar in patie patients nts classified classified as havin having g the same type of apha aphasia sia (e.g. Broc Broca’s a’s aphasia) compared compared to thos thosee who have different kinds of aphasia (e.g. Wernicke’s aphasia or conducti cond uction on apha aphasia). sia). Neve Neverthel rtheless, ess, even amon among g patie patients nts who are cla classi ssified fied as hav having ing the sam samee kin kind d of aph aphasi asia, a, there is considerable variability in impairment and task performance form ance.. Two prima primary ry fact factors ors cont contribut ributee to the pred predictictability of aphasic impairment: the first factor pertains to the fact that the anat anatomy omy of cere cerebrova brovascul scular ar perf perfusion usion territories tor ies is rel relati ativel vely y sim simila ilarr acr across oss ind indivi ividua duals ls and and,, as a result, a stroke that affects the territory of a given segment of a cer cerebr ebral al art artery ery res result ultss in som somewh ewhat at sim simila ilarr dam damage age across acr oss pat patien ients ts (Caviness et al ., ., 200 2002 2). Ap Apha hasi sia a is mo most st common com monly ly the res result ult of an occ occlus lusion ion wit within hin the mid middle dle cerebral cere bral artery (MC (MCA) A) terri territory tory.. Aft After er its origi origin n from the internal carotid artery, the MCA bifurcates into a superior and an inferior branch. Occlusions involving the superior division tend to lead to similar lesion patterns, which are different than the patterns yielded by strokes resulting from occlusion of the inferior division of the MCA. The second factor is that although there is some degree of variability between individuals in the cortical organization of speech and language processing processing (Ojem Ojemann ann and Whit Whitaker, aker, 1978; 1978; Amunts et al ., ., 1999; 1999; Fischl et al ., ., 2008; 2008; Fedorenko et al ., ., 2010), 2010 ), the overall regional distribution of language is fairly consistent across healthy individuals. That is, cortical activation vat ion stu studie diess com common monly ly rev reveal eal rem remark arkabl ablee sim simila ilarit rities ies across healthy individuals in cortical areas recruited to execute a given speech or language task. As a result, damage to a given brain region tends to result in somewhat similar patt pa tter erns ns of sp spee eech ch an and d la lang ngua uage ge im impa pair irme ment nt ac acro ross ss individuals. The classical typology of aphasia, which has its roots in the Wern rnic ick kee-L Lic icht hthe heim im mode dell ( We Wern rnic icke ke,, 18 1874 74,, Lichthei Lich theim, m, 1885 1885)) an and d wa wass la late terr re refin fined ed by Geschwind (1970),, ass (1970) associ ociate atess th thee maj major or aph aphasi asia a typ types es wit with h spe specifi cificc lesion les ion loc locati ations ons.. For exa exampl mple, e, We Werni rnicke cke’s ’s and Bro Broca’ ca’ss aphasia apha sia were associated associated with damag damagee to Wernicke’s Wernicke’s and Broca’s Broc a’s areas areas,, resp respecti ectively. vely. However, as has been amply discussed disc ussed elsew elsewhere here (Mohr et al ., ., 197 1978 8; Dronkers et al ., ., 2007;; Lazar 2007 Lazar and Mo Mohr, hr, 201 2011 1; Fridriksson et al ., ., 201 2014 4), localized damage to these regions rarely results in complete Wernicke Wern icke’s ’s or Broc Broca’s a’s apha aphasia. sia. The Wern Wernicke icke-Lic -Lichthe htheim im model has been highly influential and is still being taught


BRAIN 2018: 141; 848–862

|

849

in clinical curricula today. However, it is an oversimplification of how speech and language are rooted in the brain as extensive research has shown how many other cortical and sub subco corti rtical cal reg region ionss bes beside idess Bro Broca’ ca’ss and Wer Wernic nicke’ ke’ss area ar eass ar aree in invo volv lved ed in pr proc oces essi sing ng sp spee eech ch an and d la lang ngua uage ge (Hi Hick ckok ok an and d Po Poep eppe pel, l, 20 2007 07;; Tourvil Tourville le and Gue Guenth nther, er, 2011;; Ueno et al ., 2011 ., 2011). 2011). One of the most influential contemporary neuropsychological models of speech and language organization in the brain is the dual stream model (Hickok (Hickok and Poeppel, 2004, 2004 , 2007), 2007 ), which is associationist, like the Wernicke-Lichtheim model, mode l, but place placess the main emph emphasis asis on the connection connectionss between cortical regions. The dual stream model describes two lar largege-sca scale le pro proces cessin sing g str stream eams. s. A ven ventra trall str stream eam,, rooted in the bilateral temporal lobes, supports the processing of auditory-to-meaning information and is essential for successfu succ essfull audit auditory ory comp comprehe rehension nsion.. A dorsa dorsall stre stream am processess audi cesse auditorytory-to-a to-articu rticulatio lation n info informat rmation ion and is unila unilatterally era lly org organi anized zed acr across oss lef left-h t-hemi emisph sphere ere fro fronta ntall spe speech ech areas and a regio region n loca located ted at the temporal-pa temporal-parieta rietall junc junc-tion. The dorsal stream provides ad hoc auditory and proprioce pri ocepti ptive ve fee feedba dback ck tha thatt is cru crucia ciall for pro produc ducing ing flu fluent ent speech. In a recent paper, our group performed an extensivee eva siv evalua luatio tion n of lan langua guage ge defi deficit citss in rel relati ation on to pos posttstroke lesion location, mapping the grey matter localization of the ventral and dorsal streams (Fridriksson ( Fridriksson et al ., ., 2016). 2016). In lin linee wit with h the the theore oretic tical al pre predic dictio tions ns by Hickok Hickok and Poeppel (2007), (2007), the dorsal stream involves fronto-parietal regions, regio ns, incl includin uding g the pars oper opercular cularis, is, pars trian triangular gularis, is, pre-- and postcent pre postcentral ral reg region ions, s, as wel welll as por portio tions ns of the pariet par ietal al lob lobe. e. In con contra trast, st, the ven ventra trall str stream eam inv involv olves es much of the lateral temporal lobe, extending into the posterior-inferior frontal gyrus pars orbitalis via the uncinate fasciculus. This study provided strong support for the cortical boundaries boundaries of two anatomical anatomical strea streams ms that support speech spe ech and lan langua guage ge pro proce cessi ssing. ng. Not Notabl ably, y, our stu study dy revealed the anatomical boundaries of the dorsal and ventral stre st ream amss to be mo more re in li line ne wi with th Hi Hick ckok ok an and d Po Poep eppe pel’ l’ss framework than with other dual stream models of speech proces pro cessin sing g pro propos posed ed in the lit litera eratur turee (Rausc Rauscheck hecker er and Tian, 2000; 2000; Rauschecker and Scott, 2009). 2009 ). Although it is possible to infer aphasic symptomatology from the dual stream model, it is important to point out thatt the mo tha model del des descri cribes bes th thee ana anato tomic mical al fou founda ndatio tions ns of normal, and not disordered, speech and language processing. Given the outdated utility of the Wernicke-Lichtheim model and the contemporary emphasis on the dual stream model, the purpose of the current study was to investigate how damage to the ventral and dorsal streams identified by al.. (2016) relates Fridriksson et al relates to perf performan ormance ce on test testss common com monly ly use used d to ass assess ess aph aphasi asia. a. As suc such, h, the cur curren rentt study represents a follow-up study to our previous work (Fridriksson et al ., ., 2016). 2016). We expected that measures that addres add resss rel relati ative ve iso isolat lation ion of spe speech ech com compre prehen hensio sion n and motor speech processes would load strongly onto the ventral and dorsal streams, respectively. We further expected

850

| BRAIN 2018: 141; 848–862

J. Fridriksson et al .

Table 1 List of measures included in RLSM and CLSM analyses and the domain of communication assessed Test/scale

Domain assessed

Participants completed

WAB Aphasia Quotient WAB Speech Fluency WAB Speech Repetition WAB Word Recognition WAB Reading WAB Writing Pyramids and Palm Trees PNT Correct PNT Semantic Errors PNT Phonemic Errors Syllable Identification Sentence Comprehension Non-Canonical Comprehension ASRS AOS Severity Speech rate Articulator y rate

Overall deficit severity Speech production Speech perception/production Speech comprehension Reading comprehension Written language Semantic processing (non-verbal) Verbal naming Verbal naming Verbal naming Speech Perception Grammatical processing (all sentence types) Grammatical processing (non-canonical relative to canonical) Speech production/ar ticulation Speech production Speech production/ar ticulation

159 159 159 159 55 55 117 105 105 105 42 57 57 65 103 103

AOS = apraxia of speech; ASRS = Apraxia of Speech Rating Scale; WAB = Western Aphasia Battery.

that tests requ requiring iring the invo involveme lvement nt and coor coordinat dination ion of several different processes would involve both streams. The meth methodol odology ogy used here invo involved lved both regio region-wi n-wise se lesion symptom mapping (RLSM) and connectome lesionsymptom mapping (CLSM; Yourganov (CLSM; Yourganov et al ., ., 2016). 2016). CLSM usess the sam use samee sta statis tistic tical al app approa roach ch as tra tradit dition ional al les lesion ion-symptom mapping methods with one major exception: instead ste ad of rel relati ating ng a giv given en les lesion ion loc locati ation on to imp impair airmen ment, t, CLSM associates associates dama damage ge invol involving ving whit whitee matt matter er conn connecections between anatomical regions and behavioural impairment.. Thus ment Thus,, rathe ratherr than isolating isolating lesio lesion n locat locations, ions, CLSM makes it possible to reveal cortical networks that are crucial ci al fo forr pe perf rfor ormi ming ng a gi give ven n ta task sk.. Fo Forr bo both th RL RLSM SM an and d CLSM, we carried out univariate and multivariate analyses to provide slightly different perspectives based on different data analysis approaches of the same data.

Materials and methods Participants The data for this project were obtained from an archival databasee in the Aph bas Aphasi asia a Lab Lab,, Un Unive iversi rsity ty of Sou South th Car Caroli olina na and Medical Medi cal Unive University rsity of South Carolina. Carolina. Parti Participan cipants ts had sustained a single-event stroke to the left hemisphere at least 6 months prior to study inclusion and were tested either as part of an aphasia treatment study or strictly for the purpose of lesion-sym lesio n-symptom ptom mapping research. research. A lesio lesion n over overlay lay map is included inclu ded in Suppleme Supplementary ntary Fig. 1. Th Thee av aver erag agee ti time me po post st-stroke str oke was 36. 36.4 4 mon months ths [st [stand andard ard dev deviat iation ion (SD (SD)) = 43. 43.1]. 1]. The total sam sample ple size was 159 chronic chronic str strok okee sur surviv vivors ors and the numb number er of tested participants participants varied across the differ different ent assessment batteries used here to assess speech and language impairmen impai rment. t. All parti participan cipants ts were native spea speakers kers of Engli English sh with wi th a me mean an ag agee of 60 60.0 .0 ye year arss (S (SD D = 11 11.2 .2), ), an and d 68 we were re female fem ale.. Par Partic ticipa ipants nts wer weree exc exclud luded ed if the they y had a his histor tory y of


dementia or other neurological problems (as per self/caregiver or med medica icall rep repor ort). t). All par partic ticipa ipants nts wer weree rec recrui ruited ted thr throug ough h local loc al adv advert ertise isemen mentt and wer weree enr enroll olled ed at the Uni Univer versit sity y of Sout So uth h Ca Caro roli lina na or at th thee Me Medi dica call Un Univ iver ersi sity ty of So Sout uth h Carolina. Carol ina. They provi provided ded informed consent to parti participat cipatee in this study, which was approved by the Institutional Institutional Review Boar Bo ards ds at th thee Un Univ iver ersi sity ty of So Sout uth h Ca Caro roli lina na an and d at th thee Medical University of South Carolina.

Speech and language testing The behavioural battery included 16 tests and tasks selected to reflect speech and language impairments typically included as part pa rt of th thee cl clin inic ical al ma mana nage geme ment nt of ap apha hasi sia a in th thee US USA A (Table 1). 1). It was not meant to reflect in-depth assessment of lingui lin guisti sticc pro proces cessin sing g or neu neurop rophys hysiol iology ogy of spe speech ech,, as the intended audience here consists of clinicians and clinician-scientists who might be more interested in the effects of strokerelate rel ated d br brain ain dam damage age on com commu munic nicati ation on ski skills lls tha thatt mig might ht reflect real-life situations (e.g. speech fluency and overall severity of aphasia). However, we selected tests and tasks that primarily mar ily tax mot motor or spe speech ech pro proces cesses ses wit with h rel relati ativel vely y min minima imall speech speec h comp comprehe rehension nsion requirement requirementss as well as the oppo opposite; site; tests that focus specifically on speech recognition and comprehens he nsio ion n wi with th li litt ttle le or no in inpu putt fr from om mo moto torr sp spee eech ch.. Th This is allowed us to compare and contrast cortical network damage that prim primarily arily affects moto motorr speec speech h versu versuss audi auditory tory comp comprerehension, two aspects of processing that are commonly assessed as par partt of a com compre prehen hensiv sivee aph aphasi asia a wor work-u k-up. p. In add additi ition, on, other typical subtests included on aphasia test batteries were used to test functions such as reading, writing, speech repetition, verbal naming, and grammatical processing of sentences. Unfortunate Unfor tunately, ly, our dataset did not incl include ude a comp comprehe rehensive nsive measu mea sure re of pho phonol nologi ogical cal inp input, ut, so somet methin hing g tha thatt is cle clearl arly y a focus of Hickok and Poeppel’s dual stream model. The speech and language battery included six tests or rating scal sc ales es fr from om th thee We West ster ern n Ap Apha hasi sia a Ba Batte ttery ry (Kerte Kertesz, sz, 1982 1982): ): Apha Ap hasi sia a Qu Quot otie ient nt,, a 0– 0–10 100 0 po poin intt ap apha hasi sia a se seve veri rity ty sc scal ale; e; Speech Speec h Flue Fluency, ncy, a 10-p 10-point oint qualitative qualitative ratin rating g scale of speec speech h

Anatomy of aphasia

BRAIN 2018: 141; 848–862

production; Speech Repetition, a 60-point test of speech repetiti ti tion on of re real al wo word rdss an and d pr prog ogre ress ssiv ivel ely y lo long nger er se sent nten ence ces; s; Auditory Word Recognition, a 60-point test where the participantt is req pan requir uired ed to poi point nt to a pic pictur turee or obj object ect that cor corres res-ponds to spoken words presented by a clinician (picture and obje ob ject ct ta targ rgets ets ar aree pr pres esen ente ted d al alon ong g wi with th fiv fivee di dist stra racto ctors rs); ); Reading ability, a collection of nine subtests with a maximum possible score of 20 points; and Writing ability, a collection of seven subtests subtests with a maxi maximum mum possible possible score of 20 poin points. ts. Three scores were derived from the Philadelphia Naming Test (PNT), a 175-item test of picture naming (Roach (Roach et al ., ., 1996): 1996): correc cor rectt nam naming ing;; pho phonol nologi ogical cal err errors ors;; and sem semant antic ic err errors ors.. Assessme Asse ssment nt of audit auditory ory sent sentence ence comprehensio comprehension n relie relied d on a 45-item test where participants were required to match a clinician-spo ician -spoken ken sentence to a target picture that was presented along with a semantic fo foiil and an unrelate ted d foil (Magnusdottir et al ., ., 201 2013 3). Two scores scores fro from m thi thiss sen senten tence ce comprehe comp rehensio nsion n test were analy analysed: sed: overa overall ll sente sentence nce comp comprerehension hens ion accuracy and scor scores es for nonnon-canon canonical ical sente sentences nces (in compar com pariso ison n to can canon onica icall sen senten tences ces). ). To ass assess ess spe speech ech rat ratee and articulation rate, we relied on discourse measures of picture descriptions where speech rate was defined as the number of wor words ds spo spoken ken per min minute ute and art articu iculat lation ion rat ratee refl reflect ected ed speaking speak ing time minu minuss paus pauses. es. A 30-ite 30-item m in-ho in-house use speech perception cepti on task (syllable identification identification)) was inclu included ded wher wheree participants listened to one of three syllables (/pa/, /ta/, /ka/) and indicated which syllable they heard by pointing to its written representation (‘PA’, ‘TA’, ‘KA’) on a computer screen. Other tests included ratings of apraxia of speech on the Apraxia of Speech Spe ech Rating Rating Sca Scale le (Strand et al ., ., 2014), 2014), and the Pyramids and Palm Trees Test, which permits the assessment of amodal semantic processing based on 52-item matching of semantically related pictures (Howard ( Howard and Patterson, 1992). 1992 ). A correlation matrix that shows the relation between the behavioural tasks and tests is included in Supplementary Fig. 2. 2.

Brain imaging All participants underwent an extensive MRI work-up using a Siemens Trio 3 T scanner equipped equipped with a 12-element head coil either at the University of South Carolina or at the Medical Univers Uni versity ity of Sou South th Caro Carolina lina.. For the pur purpos posee of this study, three thr ee kin kinds ds of ima images ges were used: used: (i) T 1-wei -weighte ghted d MRI usin using g an MP MP-R -RAG AGE E se sequ quen ence ce wi with th 1 mm is isot otro ropi picc vo voxe xels ls,, a 256  256 matrix matrix size, size, an and d a 9 fli flip p an angle gle.. T1 images images use used d eithe eit herr a 160 sli slice ce seq sequen uence ce wit with h rep repeti etitio tion n tim timee = 22 2250 50 ms ms,, inver in versio sion n tim timee = 900ms, ech echo o tim timee = 4.5 4.52 2 ms or a 192 sli slice ce seque se quenc ncee wit with h re repet petiti ition on tim timee = 225 2250 0 ms ms,, inv inver ersio sion n tim timee = 925ms, echo time = 4.15 4.15 with para parallel llel imaging (GRA (GRAPPA PPA = 2, 80 reference lines); (ii) T2-weighted MRI using a sampling perfection fecti on with appl applicati ication on opti optimize mized d cont contrast rastss usin using g a diff differen erentt flip angle evolution (3D-SPACE) sequence. This 3D TSE scan uses a repe repetiti tition on time = 2800ms, an echo time of 402 ms, variable flip angle, 256  256 matrix matrix scan with 192 slices slices (1 mm thick thi ck), ), usi using ng par parall allel el ima imagin ging g (GR (GRAPP APPA A = 2, 12 120 0 re refer feren ence ce lines); and (iii) diffusion EPI scan that uses 30 directions with b = 1000 s/mm2 and b = 2000 s/mm2, repetition time = 6100 6100 ms, echo ech o tim timee = 101ms, 82  82 mat matrix rix,, 222  22 222 2 mm fie field ld of view,, with para view parallel llel imag imaging ing GRA GRAPPA PPA = 2, 80 45 cont contiguo iguous us 2.7mm axia axiall slic slices. es. This sequence sequence was acquired acquired twice, as well as a thi third rd ser series ies that was identic identical al in all respect respectss but only 


|

851

included nine B = 0s/mm2. Therefore, in total we acquired 60 included volumes with B = 1000, 60 with B = 2000 and 11 with B = 0.

Image preprocessing Lesions The chronic chronic str stroke oke les lesion ion was dem demarc arcate ated d on T 2-weighted images by a neurologist (L.B.), who was blinded to the participants’ language scores. The T 2 image was co-registered to the T1 image, image, and these parameters parameters were used to reslice the lesion into the native T1 space. The resliced lesion maps were smoothed smoo thed with a 3 mm full-width full-width at halfhalf-maxim maximum um Gaussian kernel to remove jagged edges associated with manual drawing. Enantiomorphic normalization (Nachev ( Nachev et al ., ., 2008) 2008) relied on SPM SPM12 12 and MA MATLA TLAB B scr script iptss dev develo eloped ped by two aut author horss (C.R (C .R., ., G. G.Y. Y.). ). A mi mirr rror ored ed im imag agee of th thee T1 image (refle (reflected cted around the midline) was coregistered to the native T1 image. We th then en cr crea eate ted d a ch chim imer eric ic im imag agee ba base sed d on th thee na nati tive ve T1 image with the lesioned tissue replaced by tissue from the mirrored image (using the smoothed lesion map to modulate this blending, feathering the lesion edge). SPM12’s unified segmentation-normalization (Ashburner (Ashburner and Friston, 2005) 2005 ) was used to warp this chimeric image to standard space, with the resulting spatial transform applied to the actual T 1 image as well as the lesion map. The normalized lesion map was then binarized, using a 50% probability threshold. Oncee the les Onc lesion ion dat data a had bee been n tra transf nsfor ormed med to sta standa ndard rd spac sp ace, e, ea each ch im imag agee wa wass di divi vide ded d in into to 11 118 8 an anat atom omic ical al gr grey ey matter mat ter bra brain in reg region ionss bas based ed on a sta stand ndard ardize ized d bra brain in atl atlas as (Faria et al ., ., 2012). 2012). This step was included because the connectome is constructed for each individual by determining the density of white matter fibres extending from one anatomical region reg ion of int intere erest st to ano anothe ther. r. To com comput putee les lesion ion load, we aligned the anatomical brain atlas containing the parcellation with each indiv individua idual’s l’s T 1-weig -weighted hted images. The T1-weighted images ima ges wer weree seg segmen mented ted int into o pro probab babili ilisti sticc gre grey y and whi white te matter mat ter maps, maps, and the gre grey y mat matter ter map was divided divided int into o regions according to the atlas. Then, lesion load was computed as the proportion of intact (i.e. not lesioned) voxels per each grey gre y mat matter ter reg region ion.. For bot both h les lesion ion and con connec nectom tome-b e-base ased d analyses, cerebellar regions were excluded. Connectome Connectom Conn ectome-bas e-based ed damag damagee was computed as the number of diffusion tensor imaging (DTI) tracts that connected each pair of the grey matter regions. The following steps were used to build each parti participan cipant’s t’s connectome: connectome: (i) segm segmentin enting g the T 1weighted images using SPM12’s unified segmentation-normalization process to determine the probabilistic grey and white matter maps; (ii) divid dividing ing the prob probabili abilistic stic grey matter map into in to an anat atom omic ical al re regi gion ons, s, us usin ing g th thee pa parc rcel ella lati tion on sc sche heme me descri des cribed bed in the pre previo vious us sec sectio tion; n; (ii (iii) i) reg regist isteri ering ng the whi white te matter mat ter and cor cortic tical al par parcel cellati lation on map mapss int into o the DTI spa space; ce; (iv)) com (iv comput puting ing gre grey y mat matter ter pai pairwi rwise se pr proba obabil bilist istic ic DTI fibr fibree tracking; (v) measuring the weight of each pairwise connectivity link as a function of the number of streamlines connecting the grey mat matter ter region region pai pair, r, and correct correcting ing it bas based ed on the distance travelled by each streamline and by the total volume of the connected connected regio regions; ns; and (vi) constructing constructing an adjace adjacency ncy matrix to summarize the individual connectome. Of note, we used use d an app appro roach ach des descri cribed bed by Bonilha et al . (2 (201 014 4a) to

852

| BRAIN 2018: 141; 848–862

attenuate attenu ate the dis distor tortin ting g eff effect ectss of str strok oke-r e-rela elated ted nec necrot rotic ic changes on the brain anatomy and fibre tracking. The preprocessing approach also excluded the lesion site from all tractography tracings (including cortical seeding, cortical waypoints or white matter track tracking ing regions) thus contr controllin olling g for lesion influen infl uence ce on the fina finall tra tract ct tra tracin cing. g. For this rea reason son,, ste steps ps to control for the overall lesion size were not included in subsequentt statis quen statistical tical analyses. analyses. Thes Thesee steps are descr described ibed in more detail below. To align the diffusion image with the lesion map, the T 2weighted image (co-registered to match the T 1-weighted image, and therefore in the same space as the native T 1 image) was normalize norm alized d to match the nonnon-diffus diffusion ion weighted image from the diffusion MRI sequence (the B0 image) and the resulting spatial transform was used to register the probabilistic maps of white and grey matter (the latter divided into regions of interest)) and the str est stroke oke lesion lesion int into o the diffusio diffusion n MR MRII spa space. ce. All subsequent calculations were performed in diffusion space. Probabilistic tractography was applied to evaluate pairwise grey matte matterr stru structura cturall conn connectivi ectivity. ty. Tract Tractograp ography hy was estimated through the FMRIB Diffusion Toolbox (FDT) probabilistic ist ic met method hod (Behrens et al ., 200 2007 7) wit with h FDT FDT’s ’s BED BEDPO POST ST being used to build default distributions of diffusion parameters ete rs at eac each h vox voxel, el, fol follow lowed ed by pro probab babili ilisti sticc tra tracto ctogra graph phy y using usin g FDT’ FDT’ss prob probtrack trackX X (para (parameter meters: s: 5000 indi individua viduall pathways drawn through the probability distributions on principle fibre direction, curvature threshold set at 0.2, 200 maximum steps,, step length 0.5 mm and dista steps distance nce correction). correction). The white matter probabilistic map excluding the stroke lesion was used as a way waypoi point nt mas mask. k. The con connec nectiv tivity ity bet betwee ween n reg region ionss was defined defi ned as the num number ber of str stream eamlin lines es arr arrivi iving ng in on onee reg region ion when whe n ano anothe therr reg region ion was see seeded ded and vice ver versa. sa. Thu Thus, s, the weighted weigh ted connectivity connectivity between regions A and B was defined as the number of probabilistic streamlines arriving at region B when region A was seeded, averaged with the number of probabilistic abili stic streamlines streamlines arriving arriving at regio region n A when region B was seeded. seede d. The calculation calculation of the prob probabili abilistic stic streamlines streamlines was correc cor rected ted bas based ed on the dis distan tance ce tra travel velled led by the str stream eamlin linee connecting conn ecting regions regions A and B (‘dis (‘distance tance correction’ correction’ built into probtr pro btrack ackX). X). To com compen pensat satee for the une unequa quall siz sizee of gre grey y matter mat ter reg region ions, s, the num number ber of str stream eamlin lines es con connec nectin ting g eac each h pair pa ir of re regi gion onss wa wass di divi vide ded d by th thee su sum m of th thee vo volu lume mess of these regions. When a given region was completely destroyed, the number of streamlines between that region and other regions was automatically set to zero.

Data analyses All univariate statistical analyses that related brain damage to speech and language impairment were implemented using the NiiStat toolbox for MATLAB (https://www.nitrc.org/projects/(https://www.nitrc.org/projects/niistat/ ). ). The lesion and connectome analyses focused on grey matter regions of interest within the dual streams defined in Fridriksson et al (2016). Reg Region ionss of int intere erest st in the dorsal dorsal al.. (2016). stream strea m inclu included ded pars oper opercular cularis, is, pars trian triangular gularis, is, anter anterior ior insula, insu la, poste posterior rior insu insula, la, prec precentra entrall gyru gyrus, s, post postcentr central al gyrus gyrus,, middle frontal gyrus, supramarginal gyrus (SMG), globus pallidus, lid us, and put putame amen. n. The ven ventra trall str stream eam inc includ luded ed pos poster terior ior middle midd le tempo temporal ral gyrus (pMTG), poste posterior rior superior temporal gyrus (pSTG), MTG, STG, superior temporal temporal pole pole,, angu angular lar gyrus, middle temporal pole, pars orbitalis, inferior temporal gyru gy rus, s, an and d mi midd ddle le oc occi cipi pita tall gy gyru rus. s. Th Thee un univ ivar aria iate te le lesi sion on



analyses analys es rel relied ied on con conven ventio tional nal les lesion ion sym sympto ptom m map mappin ping: g: General Gener al Linea Linearr Mode Modell (GLM) (pooled variance t -test) -test) with (one-tailed) and contr control ol for multi multiple ple comp compariso arisons ns P 5 0.05 (one-tailed) used us ed pe perm rmut utati ation on th thre resh shho hold ldin ing g (4 (400 000 0 pe perm rmut utati ation ons) s).. Similarly, the univariate CLSM analyses also relied on GLM with P 5 0.05 and 4000 permutations to control for multiple compariso comp arisons. ns. A hand handful ful of stud studies ies have show shown n that aphasia severity changes considerably among some patients, even in the chronic phase (Naeser ( Naeser et al ., ., 1998; 1998; Holland et al ., ., 2017; 2017; Hope ., 2017). 2017). To verify that time post-stroke would not influet al ., ence the results, we examined the correlation between each of the 16 tas tasks ks and tes tests ts and tim timee po postst-str stroke oke.. No sta statis tistic ticall ally y significant signi ficant corr correlatio elations ns were reveal revealed. ed. Ther Therefore efore,, the lesio lesionnsymptom analyses were not adjusted for time post-stroke. For mul multivar tivariate iate anal analyses yses,, we reli relied ed on step stepwise wise regression regression implemen impl emented ted in the ‘Aut ‘Automat omatic ic Line Linear ar Mod Modeling eling’’ modu module le in SPSS (Edition 24.0) were the dorsal and ventral stream regions of int inter erest est and lin links ks bet betwee ween n tho those se reg regio ions ns of int inter eres estt wer weree inclu inc luded ded as pr predi edicto ctors rs for RLS RLSM M and CLS CLSM, M, re respe specti ctivel vely. y. Criteria for entry and removal of factors in a step-wise model selection used P 5 0.05 for factor inclusion and P 4 0.10 for factor removal. Unlike the univariate analyses, the multivariate analyses did not rely on permutation thresholding. At this time, we are not aware of com compar parab able le met method hodss tha thatt en enabl ablee bot both h univariate and multivariate analyses of RLSM or CLSM data. Hence, a direct comparison between the univariate and multivariate results should be considered with this caveat in mind. Principal component analysis (PCA) was used in an attempt to isola isolate te behav behaviour ioural al comp componen onents ts that prim primarily arily reflect impaired paire d spee speech ch comp comprehe rehension nsion versus speec speech h prod productio uction. n. To demonstrate cortical damage associated with poor speech production or speech comprehension, univariate and multivariate analyses analys es were carried out with patients’ components components scores included as dependent factors. The PCA included Varimax rotation and only components with an eigenvalue of 1 and above were considered for further analyses. As was the case for the multivariate analyses, the PCA relied on SPSS.

Results Region-wise lesion-symptom mapping results All univa univariate riate RLSM anal analyses yses yield yielded ed stati statistic stically ally sign signifiificantt res can result ultss (Fig. 1 and Supplement Supplementary ary Fig. 3A–D 3A–D), ), wit with h thee ex th exce cept ptio ion n of ‘p ‘pho hono nolo logi gica call na nami ming ng er erro rors rs’’ on th thee PNT. PN T. No Nott su surp rpri risi sing ngly ly,, pe perf rfor orma manc ncee on so some me of th thee speech spe ech and lan langua guage ge tes tests ts inc includ luded ed her heree was rel relate ated d to damage to the same set of cortical regions. For example, a few of the Western Aphasia Battery scores such as ‘aphasia quotient,’ ‘speech fluency,’ ‘auditory-word recognition,’ and ‘sp ‘speec eech h rep repeti etitio tion’ n’ wer weree pre predic dicted ted by eac each h of the 20 analysed regions of interest even though the strongest predictors dicto rs varie varied d acros acrosss tests (Table (Table 2 presents 2 presents the three strongest predictors; for a complete list of damaged regions that predicted predi cted test perf performan ormance ce see Suppleme Supplementary ntary Table 1). Perfor Per forman mance ce on ot other her tests was pre predic dicted ted by a sub subset set of the regions of interest. The regions that most often emerged

Anatomy of aphasia

BRAIN 2018: 141; 848–862

|

853

(red-yellow) and CLSM (blue-green) results for each of the speech and language tests. For CLSM, Figure 1 Univariate RLSM (red-yellow) both colour intensity and link thickness denote how strongly damage to a given link predicts speech or language test performance. AQ = aphasia quotient; quotie nt; AOS = apraxi apraxiaa of speech speech;; N.S.= non-sig non-significa nificant. nt.

as pred predictor ictorss of spee speech ch and language performance performance in the univariate RLSM were: (i) pars opercularis; (ii) STG; and (iii) (ii i) SM SMG. G. Dam Damage age to eac each h of the these se thr three ee reg region ionss was a statistic stat istically ally signi significan ficantt pred predicto ictorr of perf performan ormance ce on 14 of the 16 speech and language tests. Because of the nature of multivariate analyses where the best predictor (region of interest damage) tends to have no or low cor correl relati ation on wit with h sub subseq sequen uentt mo model del fac facto tors rs tha thatt reduce redu ce pred predictio iction n erro error, r, cons considera iderably bly fewe fewerr regio regions ns were revealed in the multivariate analyses compared to the univariat var iatee ana analys lyses es (Fi Fig. g. 2 and Supplemen Supplementary tary Fig. 4A–D 4A–D). ). Neverthe Neve rtheless, less, all 16 mult multivari ivariate ate RLSM analyses yielded statistically significant results. The regions that most often predic pre dicted ted pe perfo rforma rmance nce on the 16 spe speech ech and lan langua guage ge measures in the multivariate analyses were: (i) pSTG (predicto dic torr on eig eight ht tes tests) ts);; (ii (ii)) pre precen centra trall gyr gyrus us (pr (predi edicto ctorr on seven sev en tes tests) ts);; and (ii (iii) i) po poste sterio riorr ins insula ula (pr (predi edicto ctorr on six tests). Table tests). Table 3 includes 3 includes all statistically significant predictors for eac each h mul multiv tivari ariate ate RL RLSM SM ana analys lysis. is. It is wor worth th not noting ing that the best predictor of each speech and language measure explained proportionally far greater proportion of the varian var iance ce tha than n sub subseq sequen uentt pre predic dictor torss inc includ luded ed in eac each h model. This fact is illustrated in Fig. 2 as well as listed in Table 3. 3.


Connectome lesion-symptom mapping results As was the case for the univariate RLSM analyses, not all univariate univa riate CLSM analy analyses ses yield yielded ed stat statistic istically ally signi significan ficantt result res ults, s, inc includ luding ing ‘ph ‘phono onolog logica icall nam naming ing err errors ors’’ on the PNT (Fig. (Fig. 1 and and Table 4). 4). In addition, impaired ‘comprehension of non-canonical sentences’ (in comparison to canonical noni cal sent sentence ences) s) and slow slower er ‘arti ‘articulat culation ion rate’ were not associated assoc iated with damag damagee to spec specific ific netw network ork conn connecti ections. ons. The ove overal ralll sev severi erity ty of aph aphasi asia a (‘a (‘apha phasia sia qu quoti otient ent’) ’) and ‘speech fluency’ were predicted by damage to an extensive network primarily involving the dorsal stream, with somewhat fewer conn connecti ections ons to the vent ventral ral stre stream. am. ‘Aud ‘Auditory itory word recognition’, ‘correct naming’, and ‘speech repetition’ were also predicted by damage to an extensive cortical network. Much of the damage associated with ‘auditory word recogn rec ogniti ition on’’ inv involv olved ed the ven ventra trall str stream eam wit with h rel relati ativel vely y fewerr links involving fewe involving the dorsal stre stream, am, where whereas as ‘spe ‘speech ech repetition’ and ‘correct naming’ were associated with both dors do rsal al an and d ve vent ntra rall st stre ream am da dama mage ge.. Ac Accu cura racy cy on th thee Pyramids and Palm Trees test was mostly associated with damage to a fronto-temporal network with the pars triangularis being the region most often included as part of a

854

| BRAIN 2018: 141; 848–862


Table 2 The top three brain regions predictive in the univariate univa riate analyses analyses of each test/task test/task and the z-score associated with brain damage and each given region Test/scale

Region

Aphasia Quotient

STG Posterior Ins Posterior STG

Speech Fluency

STG IFG opercularis Posterior Ins

Speech Repetition

Posterior STG STG SMG

Audi Au dito tory ry Wor ord d Re Reco cogn gnit itio ion n

STG STG Posterior STG Posterior MTG

Reading

Posterior STG STG AG

Writing

S MG Posterior Ins Posterior STG

Pyramids and Palm Trees

PrCG PoCG Ins

PNT Correct

Posterior STG AG SMG

PNT Semantic Errors

Syllable Identification

Sent Se nten ence ce Co Comp mpre rehe hens nsio ion n

Posterior MTG MOG AG Posterior STG Posterior Ins AG Poster Post erio iorr ST STG G STG Posterior MTG

Non-Canonical

STG-pole Posterior STG Posterior Ins

AOS

PrCG PoCG SMG

Speech Rate

PrCG IFG opercularis Posterior Ins

Articulation rate

PrCG Posterior Ins Putamen

Z-Score

7.84 7.65 7.61  

7.74 7.73 7.71 

7.78 7.69 7.41  

7.17 6.99 6.77 

5.72 5.71 5.65  

5.75 5.72 5.65  

4.34 4.12 3.75 

5.13 4.96 4.75 

damaged link damaged link.. Othe Otherr func function tionss such as ‘audi ‘auditory tory syllable discri dis crimin minati ation’ on’,, ‘sp ‘speec eech h rat rate’ e’ and ‘ap ‘aprax raxia ia of spe speech ech’’ involved relatively few links. Not surprisingly, the number of cortical regions implicated in the univariate RLSM analyses was correlated with the number of links revealed in the CLSM analyses, analyses, r = = 0.86, P 5 0.00 0.0001. 01. Howe However, ver, the number num ber of par partic ticipa ipants nts inc includ luded ed in the ana analys lyses es of eac each h beha be havi viou oura rall te test st or ta task sk wa wass no nott co corr rrel elat ated ed wi with th th thee number of regions revealed by RLSM ( r = = 0.38, P = 0.17) or CLSM (r = = 0.46, P = 0.07 0.075). 5). To appre appreciate ciate whic which h cortical regions were most often included as part of a damaged link lin k rel relate ated d to tes testt per perfor forman mance, ce, we cou counte nted d ho how w oft often en each ea ch re regi gion on of in inte tere rest st oc occu curr rred ed in th thee CL CLSM SM re resu sult lts. s. Three regions stood out: (i) pars triangularis ( n = 69) and pars oper opercula cularis ris (n = 68 68); ); (i (ii) i) SM SMG G (n = 74) and ang angula ularr gyrus (n = 76); and (iii) MTG ( n = 73). Similar to the RLSM analyses, far fewer links were identified in the multivariate CLSM analyses compared to the univariate CLSM analyses (Fig. ( Fig. 2). 2). Nonetheless, the multivariatee CLSM analyses revealed damage assoc variat associate iated d with extensive exte nsive cortical cortical netw networks orks for perfo performan rmance ce on test testss and tasks such as speech repetition, ratings of ‘speech fluency’, ‘spee ‘sp eech ch rat rate’, e’, ‘ar ‘artic ticula ulatio tion n rat rate’, e’, cor correc rectt nam naming ing on the PNT, and over overall all seve severity rity of apha aphasia sia (aph (aphasia asia quotient). quotient). Thee ar Th area eass th that at mo most st of ofte ten n we were re in incl clud uded ed as pa part rtss of damaged damag ed links in the mult multivari ivariate ate analyses were: (i) pars opercularis (n = 21); (ii) precentral gyrus ( n = 19); (iii) angular gyrus (n = 19); and (iv) posterior STG ( n = 12).



4.64 4.57 4.33 4.67 4.25 4.05  

7.06 6.58 6.40  

4.08 3.92 3.73 

4.56 4.01 3.90 6.72 6.15 6.05 

6.20 5.44 5.24  

No regions of interest survived the analysis that included phonemic errors on the PNT. AG = angular gyrus; AOS = apraxia of speech; IFG = inferior frontal gyrus; Ins = insula; MOG = middle occipital gyrus; MTG = middle temporal gyrus; PoCG = post-central gyrus; PrCG = precentral gyrus. A full list of regions significantly significantly predictive predictive of each test/scale test/scale is prese presented nted in Supplementary Table 1. 1.


Principal components analysis Consistentt with our prev Consisten previous ious resea research rch sugge suggesting sting phonologica log icall fo formrm-toto-art articu iculat lation ion and pho phonol nologi ogical cal for formm-totomeaning processing primarily rely on the dorsal and ventral streams, strea ms, resp respecti ectively, vely, we expl explored ored what regio regions ns and link linkss are primarily associated with speech production and speech comprehension. For this purpose, we carried out a principal components analysis (PCA) of four tests or tasks included in our battery that primarily reflect speech production ability (spe (speech ech rate rate,, spee speech ch artic articulat ulation, ion, speech fluen fluency, cy, and apraxia of speech) and three tests or tasks that primarily reflect speech perception or comprehension (auditory word recognitio recog nition, n, audit auditory ory sylla syllable ble iden identific tification ation,, and sent sentence ence comprehension). The PCA yielded two main components that largely reflect speech production and speech comprehension abilities (Supplementary Fig. 5). 5). The primary loadings for component 1 were the four tests we had identified as primarily reflecting speech production ability. Component 2 was not as clear-cut in that ‘speech fluency’ was the second strongestt infl ges influen uence ce aft after er ‘wo ‘word rd rec recogn ogniti ition’ on’.. Yet Yet,, thr three ee of the four highest loadings loadings for component component 2 invol involved ved the test testss or tasks that primarily reflect speech perception or speech comprehe comp rehension nsion ability. The univa univariate riate RLSM analy analyses ses of components 1 and 2 revealed a fairly clear division between regions included in the dorsal and ventral streams, respectively (Fig. (Fig. 3). 3). However, one region, the SMG, was included

Anatomy of aphasia

BRAIN 2018: 141; 848–862

|

855

(red-yellow) results for each of the speech and language tests. Note that Figure 2 Multivariate RLSM (blue-green) and CLSM (red-yellow) the colour scales represent amount of variance (R 2) explained by each region of interest or link. AQ = aphasia quotient; AOS = apraxia of speech.

in bot both h les lesion ion map maps. s. The mul multiv tivari ariate ate RL RLSM SM ana analys lysis is of compon com ponent entss 1 and 2 als also o rev reveal ealed ed reg region ionss th that at mos mostly tly loaded onto the dorsal and ventral stream regions of interest with only one region, middle frontal gyrus, included in both lesion maps. For both the univariate and multivariate RLSM analyses, the precentral gyrus was the strongest predict di ctor or of co comp mpon onen entt 1 an and d th thee po post ster erio iorr ST STG G wa wass th thee strongest predictor of component 2. Somewhat less convergence across the univariate and multivariate analyses was revealed for the CLSM results. A link between the precentral gyrus and the middle frontal gyrus was the best predictor dict or of comp componen onentt 1, wher whereas eas two diff differen erentt link linkss were identified as the strongest predictor of component 2 in the univariat univ ariatee (ang (angular ular gyrus $ SMG) and multivariate analyses (posterior (posterior STG $ MTG). MTG). Bo Both th the un univa ivaria riate te and multivariate CLSM analyses for component 1 mainly highlighted links between dorsal stream regions. Although the multivari mult ivariate ate CLSM analysis most mostly ly reve revealed aled links across ventral stream regions, this was not the case for the univariate varia te resu results, lts, whic which h inclu included ded links between both dorsa dorsall and ventral stream regions.

Discussion The current study demonstrates that overall aphasia severity, as reflected by the ‘aphasia quotient’ on the Western


Aphasia Batt Aphasia Battery, ery, is assoc associated iated with exte extensive nsive cortical cortical network damage, mostly involving the dorsal stream and, to a lesser extent, the ventral stream. This is perhaps not surprisin pri sing g as the aph aphasi asia a qu quoti otient ent is hea heavil vily y wei weight ghted ed for speech spe ech pro produc ductio tion n whe where re th three ree ou outt of the fou fourr fac factor torss thatt com tha compri prise se aph aphasi asia a quo quotie tient nt (sp (spee eech ch flue fluency ncy,, spe speech ech repetition, and naming) rely on speech production. Given that the ratin rating g of ‘spee ‘speech ch fluen fluency’ cy’ and ‘speech repetition’ repetition’ ability are two of the subtests that comprise aphasia quotient, it stands to reason that there would be considerable overla ove rlap p amo among ng the reg region ionss and con connec nectio tions ns tha thatt pre predic dictt each of these factors. Aphasia can involve different degrees of impai impairmen rmentt to mult multiple iple speech and langu language age proc processes esses that subse subserve rve comm communic unication ation func functioni tioning. ng. As such such,, even relatively relat ively smaller strok strokes es that affect the cort cortical ical language network, netw ork, especially especially if the affected affected areas involve links between the dorsal and ventral streams, seem likely to cause aphasia lasting beyond the subacute stage. Whereas traditional RLSM can reveal damage that predicts a given speech or language impairment, CLSM provide vi dess co comp mple leme ment ntar ary y in info form rmat atio ion n hi high ghli ligh ghti ting ng li link nkss between betw een regions that, when damaged, damaged, have a parti particular cularly ly detrimen detr imental tal effe effect ct on func functioni tioning. ng. CLSM and RLSM are inter-related since CLSM is also dependent on lesion location: tio n: whi white te mat matter ter pat pathwa hways ys rel relate ated d to are areas as of co corti rtical cal damage are more commonly lesioned. Nonetheless, the specificc whi cifi white te mat matter ter pro projec jectio tions ns fro from m the are areas as of co corti rtical cal

856

| BRAIN 2018: 141; 848–862


Table 3 All regions surviving multivariate lesion analyses for each test/task, along with corresponding R2 and R2 change values Test/scale

Region

R2

R2 change

Aphasia Quotient

STG PrCG Poster io ior STG Posterior Ins

0.43 0.48 0.50 0.51

0.43 0.05 0.02 0.01

Speech Fluency

IFG opercularis Poster io ior STG Putamen PrCG

0.40 0.53 0.56 0.57

0.40 0.13 0.02 0. 0.02

Speech Repetition

Posterior STG Posterior Ins ITG MTG

0.42 0.47 0.48 0.49

0.42 0.05 0.02 0.01

Auditor y Word Recognition

STG AG MTG pole

0.31 0.36 0.39

0.31 0.05 0.03

Reading

STG MOG Posterior Ins

0.42 0.51 0.54

0.42 0.08 0.03

Writing

Posterior Ins AG GP MOG

0.43 0.52 0.56 0.60

0.43 0.10 0.04 0.03

PPTT

PrCG ITG AG Post ster eriior MTG

0.18 0.22 0.24 0.2 .27 7

0.18 0.04 0.03 0.0 .03 3

PNT Correct

Posterior STG PrCG

0.22 0.29

0.22 0.06

PNT Semantic Errors

AG GP MOG MTG

0.21 0.31 0.37 0.41

0.21 0.10 0.07 0.04

PNT PN T Ph Pho ono nolo logi gica call Er Errrors

Post ster eriior STG

0.0 .06 6

0.0 .06 6

Syllable Identification

Poster io ior STG

0.34

0.34

Sen Se nte ten nce Com omp pre rehe hen nsi sio on

Post ster eriior STG IFG orbitalis

0.6 .60 0 0.63

0.6 .60 0 0.03

Non-Canonical

Posterior STG IFG IF G tri rian angu gula lari riss

0.22 0.2 .27 7

0.22 0.0 .05 5

AOS

PrCG

0.21

0.21

Speech Rate

PrCG Posterior Ins Putamen IFG orbitalis

0.31 0.38 0.40 0.42

0.31 0.06 0.02 0. 0.02

Articulation Rate

PrCG Posterior Ins IFG orbitalis Putamen

0.26 0.31 0.36 0.38

0.26 0.05 0.05 0.02 0.

AG = angular gyrus; AOS = apraxia of speech; GP = globus pallidus; IFG = inferior frontal gyrus; Ins = insula; ITG = inferior temporal gyrus; MOG = middle occipital gyrus; MTG = middle temporal gyrus; PrCG = precentral gyrus.


damage are not sys damage system temati atical cally ly map mapped ped by RL RLSM. SM. CSL CSLM M can thu thuss defi define ne the anatomy anatomy of sub subcor cortic tical al net network workss tha thatt are related to behaviour. Importantly, CLSM can also disclose the relationship between cortical disconnection and behaviour. Post-stroke cortical damage can be understood as the combination of direct vascular injury (necrosis or gliosis) and disconnection (Bonilha (Bonilha et al ., ., 2014a, b). However, disconn co nnec ecte ted d are areas as th that at ar aree se seem emin ingl gly y pr pres eserv erved ed are no nott mapped by RLSM. By disclosing crucial pathways extending beyond the stroke lesion, CLSM can identify which remotely disconnected areas are directly related to cognitive function. In this context, it is worth noting that cortical damage seen on clinical scans does not only show a given lesion location associated with speech and language processing but also reflects flec ts dis discon conne nect ctio ions ns to re regio gions ns th that at may be im impo porta rtant nt as parts of a cortical speech and language network. This point is well demonstrated by damage that predicts naming impairment, anomia. In the current study, ‘correct naming’ was predicted in the univariate analysis by a lesion location mostly involvin invo lving g pos posteri terior or stru structu ctures, res, inc includ luding ing the post posterio eriorr STG and angular gyrus. However, CLSM showed that anomia is associated with extensive network damage that includes various regions that make up the ventral and dorsal streams. The involvem invo lvement ent of mul multipl tiplee regi regions ons asso associat ciated ed with anomia is further demonstrated in the multivariate analyses, which highlighted both posterior (posterior STG) and anterior structures (precentral gyrus) and connections involving those regions as being bein g cru crucial cial for corr correct ect naming on the PNT. Anomia has often ofte n bee been n dee deemed med the hall hallmark mark impairment impairment of apha aphasia— sia— pati pa tien ents ts wh who o ha have ve no wo word rd-fi -find ndin ing g pr prob oble lems ms at al alll ar aree highl hig hly y un unlik likel ely y to ha have ve aph aphasi asia a (Goo Goodgla dglass ss and Win Wingfiel gfield, d, 1997). 1997 ). An Anom omic ic ap aphas hasia ia,, the le least ast sev severe ere fo form rm of aph aphasi asia, a, is characterized by fluent speech and good auditory comprehensio hen sion n bu butt po poor or lex lexic ical al ret retrie rieval val an and, d, in th thee mo more re sev severe ere cases cas es,, em empt pty y sp spee eech ch (Hel Helm-E m-Estab stabrook rookss and Alb Albert, ert, 1991 1991). ). et al. (2015), As de demo mons nstr trat ated ed in Yourganov et (2015), un unli like ke Wernick Wer nicke’s, e’s, Broc Broca’s, a’s, con conduc duction tion,, or glob global al apha aphasia, sia, ano anomic mic aphasia has no specific lesion location. Based on the current data,, it make data makess sen sense se that various various diff differe erent nt lesi lesion on loc locatio ations ns would result in anomia as naming relies on such an extensive cortic cor tical al net netwo work. rk. Fr From om a cli clinic nical al sta standp ndpoi oint nt,, a pa patie tient nt’s ’s difficulty to name could be caused by impairment at several different levels of processing—phonology, lexical, semantic, or motor speech—that are impaired to different degrees depending on wh what at pa parts rts of th thee co corti rtical cal ne netwo twork rk th that at sup suppo ports rts naming were affected (DeLeon (DeLeon et al ., ., 2007). 2007). The findi findings ngs pres presente ented d here demo demonstra nstrate te an impo importan rtantt clinical point: speech and language processing relies on an extensive cortical network and damage to different subcomponent pon entss of thi thiss ne netwo twork rk can res result ult in dif difficu ficulty lty wit with h the same communication task. This does not mean that different parts of the network are equipotential. Quite the contrary; our data suggest that the dorsal stream is very much motor speech-driven, whereas speech comprehension relies much muc h mor moree on the ven ventra trall str stream eam.. It is the harmony harmony of stream strea m inte interact raction ion that make makess comm communic unicatio ation n possi possible, ble, and an d th thee lo loca cattio ion n of da dama mag ge, both st stru ruct ctur ura al an and d

Anatomy of aphasia

Table 4 The top 10 connections surviving in the univariate analyses for each test/task and the corresponding z-score Test/scale

Connections

Aphasi Aph asiaa Qu Quoti otient ent

IFG ope opercu rcular laris is $ IF IFG G tr tria iang ngul ular aris is IFG opercularis $ PrCG SMG $ AG PoCG $ AG AG $ posterior STG MTG $ ITG IFG orbitalis $ STG MTG $ putamen PoCG $ posterior MTG PoCG $ posterior STG IFG IF G op oper ercu cula lari riss $ PrCG IFG opercularis $ IF IFG G tr tria iang ngul ular aris is SMG $ AG PoCG $ AG IFG orbitalis $ IFG tr ia iangularis MFG $ PrCG PoCG $ pMTG PrCG $ SMG IFG opercularis $ PoCG IFG orbitalis $ STG MTG MT G $ ITG IFG opercularis $ PrCG SMG $ AG AG $ posterior STG IFG opercularis $ IF IFG G tr tria iang ngul ular aris is PoCG $ AG AG $ posterior MTG IFG orbitalis $ ITG PrCG $ posterior STG IFG orbitalis $ STG STG $ posterior STG

5.65 5.65 5.39 5.35 5.34 5.06 5.06 4.98 4.89 4.77 4.73 6.47 6.23 6. 23 5.71 5.61 5.30 5.26 5.14 5.09 5.05 5.05 5.30 5.29 5.26 5.16 4.80 4. 80 4.77 4.61 4.53 4.45 4.44 4.75

AG $ posterior STG MTG $ posterior MTG Posterior STG $ po post ster eriior MTG MTG $ ITG IFG opercularis $ IF IFG G tr tria iang ngul ular aris is ITG $ MOG ITG $ posterior MTG MOG $ posterior MTG MTG $ posterior STG SMG $ AG AG $ posterior MTG MTG $ ITG IFG opercularis $ PrCG AG $ MOG IFG opercularis $ IF IFG G tr tria iang ngul ular aris is IFG triangularis $ PrCG PrCG $ SMG SMG $ pMTG PrCG $ AG IFG opercularis $ PrCG SMG $ AG MTG $ ITG PrCG $ SMG AG $ pMTG PoCG $ AG

4.29 4.18 4.1 .17 7 4.07 3.97 3. 97 3.97 3.91 3.80 3.80 4.21 4.18 4.18 4.07 4.07 3.69 3. 69 3.69 3.67 3.58 3.56 4.97 4.80 4.52 4.50 4.36 4.34

Spee Sp eech ch Fl Flue uenc ncyy

Spee Sp eech ch Re Repe peti titi tion on

Aud Word Recognition

Reading

Writing

|

BRAIN 2018: 141; 848–862

Table 4 Continued Test/scale

Z-score

(continued) Downloaded from https://academic.oup.co https://academic.oup.com/brain/arti m/brain/article-abstract cle-abstract/141/3/848/4 /141/3/848/4816245 816245 by Rutgers University Libraries/Technical Libraries/Technical Services user

857

Pyramids and Pyramids Palm Trees

PNT Correct

PNT Semantic Errorsa

Syllable Identificationa

Sentence Comprehensiona

AOSa

Speech Ratea

a

Connections

IFG opercularis $ IF IFG G tr tria iang ngul ular aris is SMG $ posterior MTG IFG triangularis $ PrCG PrCG $ AG IFG opercularis $ IF IFG G tr tria iang ngul ular aris is IFG orbitalis $ IFG triangularis STG $ Ins IFG orbitalis $ Ins MFG $ PrCG MTG pole $ Ins Putamen $ posterior Ins PoCG $ AG SMG $ posterior Ins MTG $ Ins PrCG $ posterior STG SMG $ AG PoCG $ AG IFG opercularis $ post ster erio iorr ST STG G IFG opercularis $ PrCG MFG $ posterior STG MOG $ GP SMG $ posterior MTG IFG opercularis $ STG AG $ MOG SOG $ Thal MTG $ SOG STG $ SOG SMG $ AG AG $ posterior STG MFG $ posterior STG STG $ Ins PoCG $ AG MTG $ Ins Ins $ posterior Ins SMG $ MOG PrCG $ MOG MTG $ ITG MOG $ Ins IFG opercularis $ IF IFG G tria triang ngul ular aris is SMG $ AG PoCG $ SMG MFG $ PrCG PoCG $ AG PrCG $ SMG PoCG $ SMG PoCG $ pSTG PoCG $ pMTG PoCG $ AG IFG opercularis $ IF IFG G tr tria iang ngul ular aris is

Z-score

4.29 4. 29 4.24 4.15 4.07 3.36 3. 36 3.28 3.02 3.00 2.93 2.82 2.77 2.76 2.74 2.73 4.21 3.91 3.84 3.7 .74 4 3.64 3.51 3.50 3.45 3.32 3.27 3.47 3.24 3.05 3.64 

3.26 3.20 3.09 3.01 2.98 2.98 2.91 2.86 3.92 3.11 3.06 3. 06 3.03 2.92 2.75 2.73 2.42 4.02 3.69 3.59 3.51 3.44 3. 44

Behaviours that were predicted by 10 or fewer connections; in these cases, all predictive connections are listed. All connections are between interhemispheric, left hemisphere grey matter regions. No connections survived analyses that included ‘phonemic errors’ on the PNT, ‘grammatical processing for non-canonical sentences,’ or ‘articulation rate.’ AG = angular gyrus; AOS = apraxia of speech; GP = globus pallidus; IFG = inferior frontal gyrus; Ins = insula; ITG = inferior temporal gyrus; MFG = middle frontal gyrus; MTG = middle temporal gyrus; MOG = middle occipital gyrus; PoCG = post-central gyrus; PrCG = precentral gyrus; SOG = superior occipital gyrus; Thal = thalamus.

858

| BRAIN 2018: 141; 848–862

Figure 3 Univariate and multivariate analyses of RLSM and CLSM data derived from the PCA analysis of speech comprehension and speech production tests/tasks. The top panel shows results results from compone component nt 1 and the bottom panel shows component 2. The left images show univariate results whereas the right images show multivariate results. Note that the scales form the univariate (Z-scores) and multivariate (R 2) analys analyses es are different. different.

physiological, physiolog ical, determines determines the pattern of spee speech ch and language impai impairmen rment. t. Figure Figure 3 shows shows th that at da dama mage ge to th thee dorsal dor sal reg region ionss has maj major or eff effect ectss on spe speech ech pro produc ductio tion, n, including articulation rate and apraxia of speech severity. Not sur surpri prisin singly gly,, apr apraxi axia a of spe speech ech,, an imp impair airmen mentt of speech spe ech mot motor or pla planni nning, ng, is pre predic dicted ted by dam damage age to lin links ks betwee bet ween n mul multip tiple le cor cortic tical al reg region ions, s, alm almost ost exc exclus lusive ively ly within the dorsal stream. Conversely, patients’ inability to comprehend speech is far more related to damage to the ventral vent ral strea stream. m. Impa Impaired ired gram grammatic matical al proce processing ssing of sentence ten ces, s, whi which ch has been sug sugges gested ted to rel rely y on the dorsal dorsal stre st ream am an and d no nott th thee ve vent ntra rall st stre ream am (Frie Friederic derici, i, 2009 2009;; et al. , Wilson et 2011; Bornkess Bornkessel-Sc el-Schlese hlesewsky wsky and Schlesewsky, 2013; 2013; Mesulam et al ., ., 2015), 2015), was primarily explained by damage to regions such as the posterior STG and Broca’s area in the current dataset. Whereas patients with wit h fro fronta ntall dam damage age hav havee bee been n sho shown wn to hav havee dif difficu ficulty lty with wi th pr proc oces essi sing ng gr gram amma mati tica call lly y co comp mple lex x se sent nten ence cess (Caramazza and Zurif, 1976; 1976 ; Tyler et al ., ., 2011), 2011), this impairme pai rment nt see seems ms lik likely ely to be rel relate ated d to dis discon connec nectio tion n of fronta fro ntall lob lobee reg region ionss fro from m tem tempo poral ral lob lobee str struct ucture uress (se (seee also Tyler also Tyler and Marslen-Wilson, 2008). 2008). Indeed, a functional network netw ork analy analysis sis by by Den Den Ouden et al. (2012) (2012) suggests suggests that the ventral and dors dorsal al stre streams ams cont contribu ribute te to gramm grammatica aticall processing in healthy speakers, in an interactive manner. Although the dorsal and ventral streams provide an organizatio ganiz ational nal fram framewor ework k on whic which h huma human n comm communic unication ation relies rel ies,, not all the cortical cortical reg region ionss wit within hin the str stream eamss are equall equ ally y imp import ortant ant for spe speech ech and lan langua guage ge pro proces cessin sing. g. This notion is demonstrated by the RLSM results that suggest damage to pars opercularis, angular gyrus, and posterior STG is more harmful, overall, to speech and language processing compared to damage to other regions. The univariat var iatee and mul multiv tivari ariate ate con connec nectom tomee res result ultss rev reveal ealed ed a



similar, albeit not identical, similar, identical, pict picture ure with damag damagee to link linkss that included regions such as Broca’s area (pars opercularis and pars triangularis), SMG/angular gyrus, as well as MTG having particularly negative effects. These regions could be considered especially important hubs in the cortical speech and language network as damage to connections that terminate here has proportionally greater effects on the different aspects of speech and language processing assessed by clinical tests of aphasia. Area Spt (Sylvian fissure, parietaltempor tem poral al jun juncti ction on), ), a pos poster terior ior reg region ion and a par partt of the dorsal dor sal str stream eam (Hickok et al ., ., 200 2003 3, 2009), 2009), is lo loca cate ted d in the JHU atlas at the junction of the posterior STG, SMG, and angular gyrus. This area has been proposed to play an important impo rtant role in coor coordina dinating ting acti activity vity across the dors dorsal al and an d ve ven ntra rall st strrea eam ms (Hi Hick ckok ok an and d Po Poep eppe pel, l, 20 2007 07;; Hickok et al ., ., 201 2011 1; Hickok, Hickok, 2012 2012). ). As su such ch,, it is no nott surpri sur prisin sing g tha thatt dam damage age to thi thiss reg region ion has gen genera erall eff effect ectss on spe speec ech h and language language.. The roles roles of Bro Broca’ ca’ss are area a and the posterior STG in speech and language processing have been amply discussed elsewhere and readers are referred to a rich literature on the neurobiology of language for further details regarding human communication (Hillis (Hillis et al ., ., 2001, 2001, 2006;; DeLeon et al ., 2006 ., 2007; 2007; Cloutman et al ., ., 2009; 2009; Binder and Desai, 2011; 2011; Rogalsky and Hickok, 2011; 2011 ; Friederici, 2012;; Poeppel et al ., 2012 ., 2012; 2012; Dick et al ., ., 2014; 2014; Fedorenko and Thompson-Shill, 2014; 2014; Hagoort and Indefrey, 2014). 2014). In relation to patients, damage to any of the hubs identified here should result in greater overall impairment of speech and language. However, this does not suggest that the remaining regions of interest analysed here are not crucial for specific speci fic proc processes esses.. For exam example, ple, atrophy of the temporal pole has been associated with impaired semantic processing (Hodges et al ., ., 1992; 1992; Mummery et al. , 2000; 2000; Galton et al ., ., 2001;; Ogar et al ., 2001 ., 2011; 2011; Faria et al ., ., 2014) 2014) and damage to the inferior temporal gyrus predicts poor single word comprehension (Bonilha (Bonilha et al ., ., 2017). 2017). Nevertheless, damage to the hubs identified here seems likely to have long-term effects on overall communication ability. As such, future studies should consider damage to these regions as important for prognosis of aphasia. In addition to highlighting areas that could be considered hubs for speech and language processing, the current data also als o emp emphas hasize ize th thee rol rolee of sho shorte rterr whi white te mat matter ter fibr fibres. es. Explicitly, it is clear from Figs 1 and 2 that many of the links that, when damaged, give rise to speech and language impairment connect adjacent cortical regions. To be clear, there is noth nothing ing that disad disadvant vantages ages longe longer-ran r-range ge conn connecections tio ns in our analyses analyses.. It wou would ld see seem m tha thatt sho shorte rterr fibr fibres, es, for exa exampl mple, e, bet betwee ween n the ang angula ularr gyr gyrus, us, pos poste terio riorr STG STG,, and MTG probably play a crucial role in some aspects of speech and language processing. This notion was specifically raised in a paper by Mesulam and colleagues (2015). (2015). Moreov Mor eover, er, it wou would ld see seem m odd if onl only y the most distal distal regions of the major tracts (e.g. arcuate fasciculus, superior longitudinal fasciculus, and inferior fronto-occipital fasciculus) were crucial for speech and language processing instead of also connections between more proximal regions.

Anatomy of aphasia

BRAIN 2018: 141; 848–862

Table 5 The top 10 connections surviving multivariate connectome analyses, along with corresponding R2 and R2 change values Test/scale

Connections

Aphasia Quotient

IFG triangularis $ IFG opercularis Posterior STG $ AG ITG $ IFG orbitalis STG pole $ PrCG Putamen $ Ins IFG orbitalis $ IFG opercularis Ins $ AG GP $ MOG Posterior Ins $ MTG Posterior Ins $ Ins

Spee Sp eech ch Fl Flue uenc ncyy

Speech Repetition

Auditory Word Recognitiona

IFG IF G tr tria iang ngul ular aris is $ IFG opercularis IFG orbitalis $ IFG opercularis Posterior ITG $ GP AG $ PoCG Ins $ STG pole Putamen $ Ins L ITG $ IFG orbitalis Posterior STG $ MTG STG pole $ IFG ope opercu rcular laris is Ins $ PoCG Ins

R2 change

0.22

0.22

0.28 0.32 0.35 0.38 0.41

0.06 0.04 0.03 0.03 0.03

0.40 0.43 0.45 0.47

0.01 0.03 0. 0.02 0.02



0.25

0.25

0.31

0.05

0.33 0.37 0.39 0.41 0.43 0.45 0.44 0.4 4 0.46

0.03 0..04 0 0.03 0.02 0.02 0.02 0.01 0 .01 0.

0.20

0.20

Medial lemniscus $ MFG Posterior Ins $ MTG pole Posterior STG $ AG AG $ PoCG PrCG $ PoCG STG pole $ PrCG AG $ IFG opercularis SMG $ PrCG Posterior Ins $ STG Putamen $ IFG tri rian angu gula lari riss

0.26 0.29 0.33 0.36 0.39 0.38 0.41 0.44 0.45 0.4 .47 7

0.06 0.04 0.03 0 .03 0. 0.03 0.01 0.03 0.02 0.02 0.0 .01 1

Posterior STG

0.17

0.17

0.21

0.04

0.25 0.28 0.30 0.33 0.36 0.38

0.03 0.03 0.03 0.03 0.03 0.02

0.32

0.32

0.39 0.44 0.49 0.49

0.07 0.04 0.05 0.00

0.54

0.05

$

ITG

$

MTG

IFG triangularis $ IFG opercularis IFG orbitalis $ MFG Posterior MTG $ ITG MTG pole $ MFG Posterior Ins $ MTG pole Posterior STG $ PrCG Posterior MTG $ IFG orbitalis Readinga

R2

IFG triangularis $ IFG opercularis Posterior Ins $ GP Posterior MTG $ pSTG Putamen $ MFG IFG triangularis $ IFG orbitalis ITG $ STG pole

(continued)


|

859


Connections

Posterior MTG

$

STG

R2

R2 change

0.60

0.06

Writinga

PrCG $ IFG opercularis Putamen $ MTG SMG $ MFG IFG orbitalis $ MFG IFG triangularis $ MFG STG pole $ PrCG GP $ IFG opercularis Putamen $ Ins

0.40 0.47 0.51 0.56 0.54 0.58 0.62 0.65

0.40 0.07 0 .04 0. 0.05 0.02 0.04 0.04 0.04

PPTTa

PrCG

0.12

0.12

PNT PN T Co Corr rrec ectt

Post Po ster erio iorr ST STG G $ PrCG IFG opercularis $ MFG Posterior STG $ AG Ins $ PrCG Posterior MTG $ GP Ins $ MTG pole Posterior MTG $ MTG pole Ins $ IFG opercularis Posterior Ins $ MTG pole STG pole $ PrCG

0.18 0.22 0.26 0.28 0.32 0.37 0.40

0.18 0.04 0.04 0.03 0. 0.04 0.05 0.04

0.43 0.46 0.48

0.03 0.03 0.02

ITG

0.13

0.13 0.

MOG $ AG Posterior STG $ MOG MTG pole $ STG SMG $ MFG Putamen $ MOG ITG $ STG pole MTG $ STG pole

0.19 0.25 0.28 0.33 0.36 0.40 0.38

0.06 0. 0.05 0.04 0 .04 0. 0.03 0.05 0.02

GP

0.08

0.08

MTG pole $ AG Ins $ ITG pMTG $ ITG

0.15 0.19 0.23

0.07 0.04 0.04

AG

SMG

0.26

0.26 0.

GP $ IFG orbitalis ITG $ MTG STG $ PrCG Ins $ MTG GP $ STG

0.38 0.46 0.53 0.59 0.64

0.12 0.09 0. 0.07 0.06 0. 0.04

Ins

0.26

0.26

Posterior MTG $ PoCG Putamen $ Ins Putamen $ STG pole SMG $ IFG triangular is is MTG pole $ IFG triangularis

0.34 0.40 0.45 0.43 0.50

0.08 0.07 0.05 0.02 0.07

Non-Canonicala

STG pole $ SMG GP $ Putamen

0.12 0.18

0.12 0.07

AOSa

PrCG

0.16

0.16

PNT Semantic Errorsa

PNT Phonological Errorsa

Syllable Identificationa

Sentence Comprehensiona

$

$

MTG

Putamen

$

$

$

MFG

ITG

$

MFG

(continued)

860

| BRAIN 2018: 141; 848–862



Connections

STG pole Speech Rate

Articulation Rate

$

MFG

R2

R2 change

0.22

0.07

SMG $ PoCG Posterior STG $ MOG Posterior STG $ MTG pole IFG triangularis $ IFG opercularis STG pole $ IF IFG G op oper erccul ular ariis MTG $ PrCG Putamen $ PoCG GP $ IFG opercularis MTG pole $ PrCG Posterior Ins $ MFG

0.16 0.22 0.29 0.33

0.16 0.06 0.07 0.04

0.4 .41 1 0.45 0.49 0.53 0.56 0.59

0.0 .07 7 0.05 0.04 0.04 0.04 0.03

SMG

0.12

0.12

0.21 0.27 0.32 0.37 0.42 0.4 .47 7 0.50 0.53

0.09 0.06 0.05 0.05 0.05 0.0 .05 5 0.03 0.03

0.55

0.03

$

PoCG

Posterior Ins $ ITG MOG $ IFG opercularis MOG $ PrCG Putamen $ PoCG AG $ IFG orbitalis STG pole $ IF IFG G op oper erccul ular ariis IFG orbitalis $ MFG IFG triangularis $ IFG opercularis pMTG $ STG a

Behaviours that were predicted by 10 or fewer connections; in these cases, all predictive connections are listed. AG = angular gyrus; AOS = apraxia of speech; GP = globus pallidus; IFG = inferior frontal gyrus; Ins = insula; ITG = inferior temporal gyrus; MFG = middle frontal gyrus; MTG = middle temporal gyrus; MOG = middle occipital gyrus; PoCG = post-central gyrus; PrCG = precentral gyrus.

It could be the case that larger lesions that cause the most severe impairments do so not only because of greater grey matter damage but also because of destruction of a number of und underl erlyin ying g whi white te mat matter ter tra tracts cts,, whi which ch are cru crucia ciall for speech and language processing. The reading and writing subtests on the Western Aphasia Battery provide a somewhat shallow picture of alexia and agraph agr aphia, ia, res respec pectiv tively ely.. Mu Much ch mo more re sen sensit sitive ive tas tasks ks are needed to understand specifically why a given patient struggles with reading and writing (Kay ( Kay et al ., ., 1996; 1996; Rapcsak et al ., ., 200 2007 7). Ye Yet, t, bot both h the RLSM and CL CLSM SM ana analys lyses es yielded yield ed stati statistica stically lly signi significan ficantt resul results ts for the readi reading ng and writing subtests on the Western Aphasia Battery. Some of the same regions and connections that predict poor reading performance also predict poor writing. As both subtests tax a broad cortical network, brain damage that causes reading and writi writing ng impai impairmen rmentt can invol involve ve seve several ral diff differen erentt regions and connections. The numb number er of diff differen erentt stat statistic istically ally signi significan ficantt corti cortical cal areas and links varied considerably across the 16 different subtests included here. One reason why this was the case is that some tests or tasks simply recruit input and coordination ati on of a lar larger ger set of cor cortic tical al reg region ionss rel relati ative ve to oth other er


tests. For example, tests. example, it is reaso reasonable nable to expe expect ct that spee speech ch fluency, a construct that theoretically relies on many different processes, would recruit more regions than speech syllable discrimination. However, other possibilities need to be considered consi dered.. As demo demonstr nstrated ated in Table Table 1, th thee nu numb mber er of part pa rtic icip ipan ants ts wh who o co comp mple lete ted d ea each ch te test st or ta task sk va vari ried ed.. Therefore, statistical power was not equal across all univariate analyses. Nevertheless, the number of regions and links revealed in the univariate RLSM and CLSM analyses, respectively, was not associated with the number of participants that completed each test or task. This suggests that the number of regions and links associated with poor task performance was not primarily driven by statistical power. It is also possible that error responses on individual tasks reflected impaired processes rooted in different cortical locations. A case in point is ‘phonological naming errors’: it is not so straightforward to determine the source of a given sound error. It could be the case that an error arose because of impaired phonological retrieval or assembly. Yet, a simila sim ilarr err error or cau caused sed by imp impair aired ed mot motor or pla planni nning ng may soun so und d si simi mila larr an and d sc scor orin ing g on th thee PN PNT T is no nott ty typi pica call lly y based on deta detailed iled acoustic analyses. analyses. Ther Therefor efore, e, the nonresultt for the univariate resul univariate RLSM analysis analysis of phon phonolog ological ical naming errors may reflect different sources of impairment rather than insufficient statistical power. Nevertheless, the multivaria mult ivariate te RLSM analysis analysis did reveal an assoc associatio iation n between twe en pos poster terior ior STG dam damage age and pho phonol nologi ogical cal nam naming ing erro er rors rs,, a re resu sult lt th that at is no nott in ag agre reem emen entt wi with th Schwartz and colleagues (2012), (2012), who revealed a relationship between moree ant mor anteri erior or dor dorsal sal str stream eam dam damage age and pho phonol nologi ogical cal naming errors. The current study is somewhat unique in that it relied on univariate and multivariate analyses of both lesion and connectome data. One of the differences between univariate and multivariate analyses is that the univariate analysis does not take into acco account unt covar covariance iance across pred predictor ictorss (regi (regions ons of interest) inte rest).. Ther Therefor efore, e, in the curr current ent stud study, y, adjac adjacent ent regio regions ns were quite often predictors predictors of perf performan ormance ce on a given test in the univariate analyses whereas this rarely occurred in the multivariate analyses. This is because the extent of damage to adjacent regions is correlated in stroke. For example, patients who wh o ha have ve da dama mage ge to pa pars rs op oper ercu cula lari riss al also so te tend nd to ha have ve damage dam age to par parss tri triang angula ularis ris,, and vic vicee ver versa, sa, whi which ch mea means ns that these regions are highly correlated as predictors in our statistical analyses. Therefore, it was unlikely that both pars triangula trian gularis ris and pars oper opercula cularis ris woul would d be inclu included ded in the samee pre sam predic dictio tion n mo model del in a mu multi ltivar variat iatee ana analys lysis is du duee to theirr high covariance. thei covariance. This noti notion on is demo demonstr nstrated ated in Fig. 2, whic which h clea clearly rly demo demonstr nstrates ates that the mult multivari ivariate ate RLSM analyses analy ses rarel rarely y iden identifie tified d spat spatially ially cont contiguou iguouss regio regions. ns. We suggest the univariate and multivariate results should be considered complementary instead of contradictory. The univariate analyses highlight larger contiguous regional clusters that, when damaged, are very likely to cause impairment. In contrast, the multivariate analyses reveal smaller overall damage but more distal modules as being independent predictors of performance on a given test.

Anatomy of aphasia

In conclusion, our findings reveal that clinical tests typically used to assess aphasia recruit a cortical network compose po sed d of th thee do dors rsal al an and d ve vent ntra rall st stre ream amss un unde derl rlyi ying ng phonol pho nologi ogical cal fo formrm-toto-art articu iculat lation ion (do (dorsa rsall str stream eam)) and phonolog phon ological ical form form-to-to-mean meaning ing (ven (ventral tral strea stream), m), resp respectectively. Speech production is impaired primarily as a result of damage to the dorsal stream whereas speech comprehension is more likely associated with ventral stream damage. Nevertheless, many clinical tests of aphasia involve multiple proce pro cesse ssess tha thatt rel rely y on bot both h str stream eams, s, whi which ch can result result in poor performance performance due to damag damagee affe affectin cting g diff differen erentt section ti onss of th thee co cort rtic ical al sp spee eech ch an and d la lang ngua uage ge ne netw twor ork. k. Damage Dam age to spe specifi cificc co corti rtical cal hu hubs bs suc such h as Bro Broca’ ca’ss are area, a, SMG/angular gyrus, and posterior STG affects performance at least 6 months after stroke on several different aphasia tests and should be explored in future studies of prognosis in aphasia.

Funding This wo This work rk wa wass su supp ppor orte ted d by th thee Na Nati tion onal al In Inst stit itut utee on De Deaf afne ness ss an and d Ot Othe herr Co Comm mmun unic icat atio ion n Di Diso sord rder erss (Fri (F ridr drik ikss sson on:: P5 P50 0 DC DC01 0146 4664 64,, U0 U01 1 DC DC01 0117 1739 39,, R2 R21 1 DC01 0141 4170 70;; Bon onil ilh ha: R01 DC DC14 140 021 21;; Hil illi liss: R01 DC05375, DC05 375, P50 0146 014664; 64; Basil Basilakos akos:: T32 DC01 DC014435 4435)) and by the American Heart Association (Bonilha: SFDRN2603000)

Supplementary material Supplementary material is available at Brain online.

References Amunts K, Schleicher A, Bu ¨ rgel U, Mohlberg H, Uylings H, Zilles K. ¨ rgel Broca’s region revisited: cytoarchitecture and intersubject variability. J Comp Neurol 1999; 412: 319–41. Ashburner J, Friston KJ. Unified segmentation. Neuroimage 2005; 26: 839–51. Behr Be hren enss TE TE,, Be Berg rg HJ HJ,, Jb Jbab abdi di S, Ru Rush shwo wort rth h M, Wo Wool olri rich ch M. Probabilistic diffusion tractography with multiple fibre orientations: what can we gain? Neuroimage 2007; 34: 144–55. Binder JR, Desai RH. The neurobiology of semantic memory. Trends Cogn Sci 2011; 15: 527–36. Boni Bo nilh lha a L, Hi Hill llis is AE AE,, Hi Hick ckok ok G, de den n Ou Oude den n DB DB,, Ro Rord rden en C, Fridriks Frid riksson son J. Temp Temporal oral lobe netw networks orks supp supporti orting ng the comp comprehe rehennsion of spoken words. Brain 2017; 140: 2370–80. Boni Bo nilh lha a L, Ne Nesl slan and d T, Ro Rord rden en C, Fi Fill llmo more re P, Ra Ratn tnay ayak akee RP RP,, Fridriks Frid riksson son J. Mapp Mapping ing remo remote te subc subcortic ortical al rami ramificat fications ions of inju injury ry after ischemic strokes. Behav Neurol 2014a; 2014: 215380. Bonilha Boni lha L, Rord Rorden en C, Frid Fridriks riksson son J. Asse Assessin ssing g the clinical effect of residu res idual al cor cortic tical al dis discon connec nectio tion n aft after er isc ischem hemic ic str stroke okes. s. Str Stroke oke 2014b; 45: 988–93. Bornkessel-Schlesewsky I, Schlesewsky M. Reconciling time, space and function func tion:: a new dorsal–ventr dorsal–ventral al strea stream m mode modell of sente sentence nce comp comprerehension. Brain Lang 2013; 125: 60–76. Caramazza A, Zurif EB. Dissociation of algorithmic and heuristic processes cess es in lang language uage comprehension comprehension:: evide evidence nce from aphasia. aphasia. Brai Brain n Lang 1976; 3: 572–82.


BRAIN 2018: 141; 848–862

|

861

Caviness V, Makris N, Montinaro E, Sahin N, Bates J, Schwamm L, et al. An Anato atomy my of str stroke oke,, Par Partt I an MRI MRI-ba -based sed topograp topographic hic and volumetric system of analysis. Stroke 2002; 33: 2549–56. Cloutm Clo utman an L, Got Gottes tesma man n R, Cha Chaudh udhry ry P, Dav Davis is C, Kle Kleinm inman an JT, Pawlak Paw lak M, et al. Where Where (in the bra brain) in) do sem seman antic tic errors errors com comee from? Cortex 2009; 45: 641–9. DeLeon J, Gottesman RF, Kleinman JT, Newhart M, Davis C, HeidlerGary Gar y J, et al. Neu Neural ral reg regions ions essential essential for dist distinct inct cognitive cognitive pro process cesses es underlying picture naming. Brain 2007; 130 (Pt 5): 1408–22. den Ouden DB, Saur D, Mader W, Schelter B, Lukic S, Wali E, et al. Netw Ne twor ork k mo modu dula lati tion on du duri ring ng co comp mple lex x sy synt ntac acti ticc pr proc oces essi sing ng.. Neuroimage 2012; 59: 815–23. Dick AS, Bernal B, Tremblay P. The language connectome: new pathways, new concepts. Neuroscientist 2014; 20: 453–67. Dronkers Dron kers NF, Plais Plaisant ant O, Iba-Z Iba-Zizen izen MT, Caba Cabanis nis EA. Paul Broca’s Broca’s historic cases: high resolution MR imaging of the brains of Leborgne and Lelong. Brain 2007; 130(Pt 5): 1432–41. Engelter ST, Gostynski M, Papa S, Frei M, Born C, Ajdacic-Gross V, et al. Epidemiology of aphasia attributable to first ischemic stroke. Stroke 2006; 37: 1379–84. Faria AV, Joel SE, Zhang Y, Oishi K, van Zjil PC, Miller MI, et al. Atlas-based analysis of resting-state functional connectivity: evaluation for reproducibility and multi-modal anatomy-function correlation studies. Neuroimage 2012; 61: 613–21. Faria AV, Sebastian R, Newhart M, Mori S, Hillis AE. Longitudinal imaging and deterioration in word comprehension in primary progressive aphasia: potential clinical significance. Aphasiology 2014; 28: 948–63. Fedorenk Fedo renko o E, Hsie Hsieh h P-J, Nieto Nieto-Cas -Castan tan Whitfie tfieldld-Gab Gabrie rieli li S, ˜ ´ o´ n A, Whi ˜ o Kanwisher Kanw isher N. New method for fMRI inve investiga stigations tions of lang language uage:: definin defi ning g reg region ionss of int intere erest st fun functi ction onall ally y in ind indivi ividua duall sub subjec jects. ts. J Neurophysio Neurophysioll 2010; 104: 1177–94. Fedorenko E, Thompson-Schill SL. Reworking the language network. Trends Cogn Sci 2014; 18: 120–6. Fischl B, Rajendran N, Busa E, Augustinack J, Hinds O, Yeo BT, et al. Cortical Cort ical fold folding ing patte patterns rns and pred predictin icting g cyto cytoarch architect itecture. ure. Cere Cereb b Cortex 2008; 18: 1973–80. Fridriksson J, Fillmore P, Guo D, Rorden C. Chronic Broca’s Aphasia Is Cau Caused sed by Da Damag magee to Bro Broca’ ca’ss and Wer Wernic nicke’ ke’ss Are Areas. as. Cer Cereb eb Cortex 2014; 25: 4689–96. Fridriksson J, Yourganov G, Bonilha L, Basilakos A, Den Ouden D-B, Rorden Rord en C. Reve Revealing aling the dual streams of speec speech h proc processin essing. g. Proc Natl Acad Sci USA 2016; 113: 15108–13. Friederici AD. Pathways to language: Fiber tracts in the human brain. Trends Cogn Sci 2009; 13: 175–81. Friederici AD. The cortical language circuit: from auditory perception to sentence comprehension. Trends Cogn Sci 2012; 16: 262–8. Galton CJ, Patterson K, Graham K, Lambon-Ralph M, Williams G, Anto An toun un N, et al al.. Di Diff ffer erin ing g pa patt tter erns ns of te temp mpor oral al at atro roph phy y in Alzheimer Alzh eimer’s ’s disea disease se and sema semantic ntic deme dementia. ntia. Neur Neurolog ology y 2001 2001;; 57: 216–25. Geschwin Gesc hwind d N. The organization organization of lang language uage and the brain. Scien Science ce 1970; 170: 940–4. Goodglas Good glasss H, Wing Wingfield field A. Anom Anomia: ia: neur neuroana oanatomi tomical cal and cogn cognitive itive correlates. San Diego, CA: Academic Press; 1997. Hagoort P, Indefrey P. The neurobiology of language beyond single words. Annu Rev Neurosci 2014; 37: 347–62. Helm-Est Helm -Estabro abrooks oks N, Albe Albert rt M. Manu Manual al of apha aphasia sia ther therapy. apy. Austin, TX: PRO-ED; 1991. Hickok Hick ok G. The corti cortical cal orga organizat nization ion of spee speech ch proc processin essing: g: feedb feedback ack control and predictive coding the context of a dual-stream model. J Com Dis 2012; 45: 393–402. Hickok G, Buchsbaum B, Humphries C, Muftuler T. Auditory–motor interaction revealed by fMRI: speech, music, and working memory in area Spt. J Cogn Neurosci 2003; 15: 673–82. Hickok G, Houde J, Rong F. Sensorimotor integration in speech processing: computational basis and neural organization. Neuron 2011; 69: 407–22.

862

| BRAIN 2018: 141; 848–862

Hickok Hicko k G, Ok Okada ada K, Ser Serenc ences es JT. Area Spt in the human human pla planum num temporale supports sensory-motor integration for speech processing. J Neurophys 2009; 101: 2725–32. Hickok G, Poeppel D. Dorsal and ventral streams: a framework for unders und erstan tandin ding g asp aspect ectss of the fun functi ctiona onall an anato atomy my of lan langua guage. ge. Cognition 2004; 92: 67–99. Hickok G, Poeppel D. The cortical organization of speech processing. Nat Rev Neurosci 2007; 8: 393–402. Hillis AE, Kleinman JT, Newhart M, Heidler-Gary J, Gottesman R, Barker Bark er PB, et al. Restoring Restoring cerebral blood flow reveals neural regions critical for naming. J Neurosci 2006; 26: 8069–73. Hillis AE, Wityk RJ, Tuffiash E, Beauchamp NJ, Jacobs MA, Barker PB, et al. Hypoperfusion of Wernicke’s area predicts severity of semantic deficit in acute stroke. Ann Neurol 2001; 50: 561–6. Hodges Hodg es JR, Patterson Patterson K, Oxbu Oxbury ry S, Funn Funnell ell E. Sema Semantic ntic dementia. dementia. Brain 1992; 115: 1783–806. Holland A, Fromm D, Forbes M, MacWhinney B. Long-term recovery in stro stroke ke acco accompan mpanied ied by apha aphasia: sia: a reco reconsid nsiderati eration. on. Apha Aphasiolo siology gy 2017; 31: 152–65. Hope TM, Leff AP, Prejawa S, Bruce R, Haigh Z, Lim L, et al. Right hemisphere structural adaptation and changing language skills years after left hemisphere stroke. Brain 2017; 140: 1718–28. Howard D, Patterson K. The pyramids and palm trees test: a test of semant sem antic ic acc access ess fro from m wor words ds and pic pictur tures. es. Ca Cambr mbridg idge: e: Tha Thames mes Valley Test Company; 1992. Kay J, Lesser R, Coltheart M. Psycholinguistic assessments of language processing in aphasia (PALPA): an introduction. Aphasiology 1996; 10: 159–80. Kertes Ker teszz A. Wes Wester tern n aph aphas asia ia ba batte ttery. ry. New York, NY NY:: Gr Grune une & Stratton; 1982. Lazar RM, Mohr J. Revisiting the contributions of Paul Broca to the study of aphasia. Neuropsychol Rev 2011; 21: 236. Lichtheim L. On aphasia. Brain 1885; 7: 433–84. Magnusdottir S, Fillmore P, Den Ouden D, Hjaltason H, Rorden C, Kjartansson O, et al. Damage to left anterior temporal cortex predicts impa impairmen irmentt of comp complex lex synta syntactic ctic proc processi essing: ng: a lesio lesion—sy n—sympmptom mapping study. Hum Brain Mapp 2013; 34: 2715–23. Mesula Mes ulam m M-M M-M,, Tho Thomps mpson on CK, Wei Weintr ntraub aub S, Rog Rogals alski ki EJ. The Wernicke conundrum and the anatomy of language comprehension in primary progressive aphasia. Brain 2015; 138: 2423–37. Mohr J, Pessin M, Finkelstein S, Funkenstein H, Duncan G, Davis K. Broca aphasia Pathologic and clinical. Neurology 1978; 28: 311– 24. Mumme Mum mery ry CJ CJ,, Pat Patter terson son K, Pri Price ce C, Ash Ashbur burner ner J, Fra Fracko ckowia wiak k R, Hodges JR. A voxel-based morphometry study of semantic dementia:: rel tia relati ations onship hip bet betwee ween n tem tempor poral al lob lobee atr atrop ophy hy and sem seman antic tic memory. Ann Neurol 2000; 47: 36–45. Nach Na chev ev P, Co Coul ulth thar ard d E, Ja gerr HR HR,, Ke Kenn nnar ard d C, Hu Husa sain in M. ¨ ge ¨ Enan En anti tiom omor orph phic ic no norm rmal aliz izat atio ion n of fo foca call lly y le lesi sion oned ed br brai ains ns.. Neuroimage 2008; 39: 1215–26. Naeser Naes er MA, Palumbo Palumbo CL, Prete MN, Fitzp Fitzpatric atrick k PM, Mimura M, Samara Sam arawee weera ra R, et al. Vis Visibl iblee cha change ngess in les lesion ion bo borde rders rs on CT scan after five years poststroke, and long-term recovery in aphasia. Brain Lang 1998; 62: 1–28.


J. Fridriksson et al . Ogar JM, Baldo JV, Wilson SM, Brambati SM, Miller BL, Dronkers NF, et al. Sema Semantic ntic dementia and pers persistin isting g Werni Wernicke’s cke’s aphasia: aphasia: linguistic and anatomical profiles. Brain Lang 2011; 117: 28–33. Ojemann Ojem ann GA, Whita Whitaker ker HA. Lang Language uage loca localizat lization ion and vari variabili ability. ty. Brain Lang 1978; 6: 239–60. Poeppel Poep pel D, Emmo Emmorey rey K, Hick Hickok ok G, Pylk Pylkka ka Toward ardss a new ¨ nen L. Tow ¨ nen neurobiology of language. J Neurosci 2012; 32: 14125–31. Rapcsak SZ, Henry ML, Teague SL, Carnahan SD, Beeson PM. Do dual-route models accurately predict reading and spelling performance an ce in in indi divi vidu dual alss wi with th ac acqu quir ired ed al alex exia ia an and d ag agra raph phia ia?? Neuropsychologia 2007; 45: 2519–24. Rauschecker JP, Scott SK. Maps and streams in the auditory cortex: nonhu no nhuma man n pri primat mates es ill illum umina inate te hum human an spe speech ech pro proces cessin sing. g. Na Natt Neurosci 2009; 12: 718–24. Rauschecker JP, Tian B. Mechanisms and streams for processing of ‘what’ ‘wh at’ and ‘where’ ‘where’ in aud audito itory ry cor cortex tex.. Pro Procc Na Natl tl Aca Acad d Sci US USA A 2000; 97: 11800–6. Roach Roa ch A, Sch Schwar wartz tz MF, Martin Martin N, Gre Grewa wall RS, Brecher Brecher A. The Phil Ph ilad adel elph phia ia na nami ming ng te test st:: sc scor orin ing g an and d ra rati tion onal ale. e. Cl Clin inic ical al Aphasiology 1996; 24: 121–34. Rogalsky C, Hickok G. The role of Broca’s area in sentence comprehension. J Cogn Neurosci 2011; 23: 1664–80. Schwa Sch wartz rtz MF, Faseyita Faseyitan n O, Kim J, Co Cosle slett tt HB HB.. The dor dorsal sal str stream eam contri con tribut bution ion to pho phonol nolog ogica icall ret retrie rieval val in obj object ect nam naming ing.. Bra Brain in 2012; 135(Pt 12): 3799–814. Strand EA, Duffy JR, Clark HM, Josephs K. The apraxia of speech rating rat ing sca scale: le: a too tooll for dia diagn gnosi osiss and des descri cripti ption on of apr aprax axia ia of speech. J Commun Disord 2014; 51: 43–50. Tourvi Tou rville lle JA, Guenther Guenther FH. The DIV DIVA A mod model: el: a neu neural ral theory theory of speech spee ch acqu acquisiti isition on and prod producti uction. on. Lang Cogn Proc Process ess 2011 2011;; 26: 952–81. Tyler LK, Marslen-Wilson W. Fronto-temporal brain systems supporting spoken language language comp comprehe rehensio nsion. n. Philo Philoss Tran Transs R Soc Lond B Biol Sci 2008; 363: 1037–54. Tyler LK, Marsl Marslen-Wi en-Wilson lson WD, Rand Randall all B, Wrig Wright ht P, Deve Devereux reux BJ, Zhuang Zhua ng J, et al. Left infer inferior ior frontal cortex and synt syntax: ax: function, function, structure stru cture and beha behaviou viourr in patie patients nts with left hemis hemispher pheree dama damage. ge. Brain 2011; 134: 415–31. Ueno T, Saito S, Rogers TT, Lambon Ralph MA. Lichtheim 2: synthesizing aphasia and the neural basis of language in a neurocomputational model of the dual dorsal-ventral language pathways. Neuron 2011; 72: 385–96. Wernicke C. Der aphasische Symptomencomplex: eine psychologische Studie Stu die auf an anato atomis mische cherr Ba Basis sis.. Bre Bresla slau: u: M. Coh Cohn n un und d Wei Weiger gert; t; 1874. Wilson Wilso n SM, Galantucci Galantucci S, Tarta Tartaglia glia MC, Risin Rising g K, Patt Patterso erson n DK, Henry ML, et al. Syntactic processing depends on dorsal language tracts. Neuron 2011; 72: 397–403. Yourgano Your ganov v G, Fridr Fridriksso iksson n J, Rord Rorden en C, Glei Gleichge chgerrch rrchtt E, Boni Bonilha lha L. Multivari Mult ivariate ate conn connectom ectome-ba e-based sed symp symptom tom mapp mapping ing in post post-stro -stroke ke patients: pati ents: networks networks supp supporti orting ng lang language uage and spee speech. ch. J Neur Neurosci osci 2016; 36: 6668–79. Yourganov G, Smith KG, Fridriksson J, Rorden C. Predicting aphasia type from brai brain n dama damage ge meas measured ured with stru structura cturall MRI. Cortex 2015; 73: 203–15.

Anatomy of Aphasia Revisited (2018)

Recommend Documents