Functional.Lesion Investigation of Developmental ...ric.uthscsa.edu/personalpages/lancaster/SPM_Class/Lecture_20/Fox... · Science Center San Antonio Janis C ... the dominant general

Journal of Speech and Hearing Research, Volume 39, 1208-1227, December 1996

Functional.Lesion Investigation ofDevelopmental Stuttering WithPositron Emission Tomography

Roger J. InghamDepartment of Speechand Hearing Sciences

University of California,Santa Barbara

Peter T. FoxResearch Imaging Center

University of Texas HealthScience Center

San Antonio

Janis C. InghamDepartment of Speechand Hearing Sciences


Frank ZamarripaCharles Martin

Paul JerabekResearch Imaging Center

University of Texas HealthScience Center

San Antonio

John CottonDepartment of Psychology


Positron emission tomographic (PET) H25 0 measurements of resting-state regional cerebralblood flow (CBF) were obtained in 29 right-handed men, 10 of whom stuttered and 19 of whomdid not. PET images were analyzed by sampling 74 regions of interest (ROls), 37 perhemisphere. ROI placement was guided both physiologically and anatomically. PhysiologicalROI placement was based on speech motor activations. Anatomical ROls were positioned byreference to a stereotactic, neurosurgical atlas with positions confirmed and finely adjusted byco-registered magnetic-resonance images (MRIs). For all subjects, PET and MR images werenormal to visual inspection. Highly significant (p < 0.0001) between-region and between-hemisphere effects were found for both groups, as have been previously reported for normalsubjects, but no significant between-group differences were found for any regional CBF values.Analysis by a laterality index found a weakly significant between-groups effect (p = 0.04) thatwas isolated to five regions, four of which are implicated in speech or hearing. However, theseregional laterality effects showed no consistent directionality, nor did these regions haveabsolute differences in regional blood flow between groups. Present findings do not supportrecent suggestions that developmental stuttering is associated with abnormalities of brainblood flow at rest. Rather, our findings indicate an essentially normal functional brain terrainwith a small number of minor differences in hemispheric symmetry.

KEY WORDS: stuttering, PET (positron emission tomography), functional-lesion

Currently, the dominant general hypothesis governing stuttering research andtheory is that developmental stuttering is a product of central nervous systemdysfunction, possibly with genetic origins. The strength of this hypothesis is reflectedin a plethora of recent studies into the neurophysiology and physiology of stuttering(see Bloodstein, 1995; Boberg, 1993; Cooper, 1990; Peters, Hulstijn, & Stark-weather, 1991) and the influence of theories postulating hemispheric laterality and/ormotor system dysfunction (Webster, 1993; Zimmermann, 1980). There is somerecognition (see, for example, Rosenfield & Nudelman, 1987; Smith, 1990) thatstuttering must also be affect- and environmentally sensitive, but this view stillconcedes primacy to a neurologic disorder hypothesis. The most favored viewappears to be that the core problem is either a neural dysfunction or deficiency thatis expressed most obviously in stuttering behavior, but equally significantly througha variety of relative deficiencies in general speech motor skills of persons who stutter(see Peters et al., 1991), albeit mainly among adults. These theories raise thequestion of whether the brains of persons who stutter are in any way fundamentallydifferent from the brains of persons who do not stutter-that is, whether people whostutter display signs of cortical dysfunction or brain lesion as opposed to displayingdifferences primarily associated with speech production and/or language process-ing. This is the question addressed in the current study. In this regard, Moore (1993,p. 46) summarized his own research findings to conclude that "... differences

C 1996, American Speech-Language-Hearing Association 0022-4685/96/390612081208

Ingham et al.: PET Functional-Lesion Investigation of Stuttering 1209

observed between stutterers and nonstutterers are not re-lated to differences in brain morphology. . ." but rather todifferences in the way the brains of people who stutterprocess linguistic stimuli.

Whether apparent brain differences between people whostutter and those who do not are reflective of the existenceof subtle brain lesions is best addressed through studiesthat examine or infer patterns of brain activity during non-speech tasks, ideally during resting states. In this context,Moore's (1993) above-stated conclusion appears contradic-tory to his own findings, obtained during rest conditions, ofsignificant EEG alpha power differences between 9 adultstuttering speakers and adult controls (Wells & Moore,1990). Between-groups differences occurred at both anteriorand posterior temporoparietal electrode placements. Fur-ther, unlike the control subjects, stutterers' brains showedno differences in posterior alpha power between restingstate and linguistic tasks.

Additional support for the position that the brains ofpeople who stutter are fundamentally different from normallyfluent speakers' brains derives from Webster's behavioralresearch on finger-tapping performance (cf. Webster, 1993).Finger-tapping tasks were selected based on the assump-tion that the neural mechanisms of speech motor controloverlap other, nonspeech motor control phenomena (andperhaps because finger-tapping seems especially relevantdue to the close proximity of mouth and finger on the motorstrip homunculus). Webster attributes the significant be-tween-group differences he observed in complex finger-tapping tasks to "some underlying structural and functionalproblem with the circuitry" (p. 91) of people who stutter,which he ascribes primarily to a fragile speech motor controlsystem characterized by "structural weakness of the leftSMA [supplementary motor area]" (pp. 99-100). Caruso's(1991) review of the neuromotor processes underlying stut-tering continues the theme that nonspeech neuromotormechanisms may be abnormal in people who stutter. In thiscontext he reports findings from Weinstein, Caruso, Sever-ing, and VerHoeve (1989) that uncovered visual timing/sequencing and movement speed deficits in people whostutter, thus further indicating the existence of a generalized(i.e., not speech-specific) movement disorder. Caruso at-tributes this to problems located in SMA and/or the basalganglia.

More support for the position that the brains of peoplewho stutter might be fundamentally different is presented byYeudall et al. (1993), who administered to 17 adult personswho stuttered and a control group a battery of neuropsy-chological measures designed to assess the functionalintegrity of the frontal regions of both hemispheres. Thebattery contained behavioral tests and quantitative EEGmeasures of brainstem and middle latency responses. Re-sults indicated that subjects could be divided into five"clusters," each showing different patterns of aberrant func-tion. In more general terms, however, Yeudall et al. claimthat the findings from a majority of the subjects who stut-tered "provide evidence of a fundamental disturbance of theleft hemisphere motor systems... [which provides] evi-dence for a neurological substrate in the etiology of stutter-ing..." (p. 160). Stub, Black, and Naeser (1987) used a

similar research approach with two sibling adult personswho stuttered, utilizing a different battery of neuropsycho-logical tests that included CT scans. They concluded thatboth subjects displayed anomalous cerebral dominance thatwas both functional and structural. Theories of (lack of)cerebral dominance have waxed and waned over the yearssince first proposed by Orton (1928) and Travis (1931, 1978)and are thriving once again as technologies for investigationof brain activity improve.

Establishing the location, extent, severity, and sense (i.e.,is neural activity increased or decreased?) of brain abnor-malities underlying stuttering is requisite for neurologicallysophisticated hypotheses and models of stuttering. Lesionlocations (in stuttering) inferred from performance studies,such as an SMA lesion postulated because of difficultieswith complex movements (Caruso, 1991; Webster, 1993),are merely putative and require confirmation. Similarly, elec-troencephalographic (EEG) studies also indicate brain dys-function in persons who stutter, both at rest and during taskperformance (Blood & Blood, 1984; Decker, Healey, &Howe, 1982; Wells & Moore, 1990; Yeudall et al., 1993);EEG, however, has such poor spatial resolution that virtuallyno inferences can be made regarding lesion location. Thecurrent generation of neuroimaging technology offers theonly conclusive, noninvasive means for localizing and char-acterizing brain lesions. Once lesions are well characterized,knowledge from other scientific domains can be systemat-ically applied. Information on neural connectivity, neuro-transmitter distributions, regional function-even computa-tional models of involved systems-can be brought to bear.This cross-disciplinary strategy, using neuroimaging as thebridge between basic science and clinical disorder, hasbeen successfully used in a broad range of diseases. Forexample, Grafton, Sutton, Couldwell, Lew, and Waters(1994) have created a computational model of Parkinson'sdisease by applying to neuroimaging data a model of thecircuitry of the basal ganglia, thalamus, and motor cortexderived from a large body of primate electrophysiologic andanatomic research.

The objective of the present neuroimaging study is toidentify and characterize the regional brain abnormalitiesassociated with developmental stuttering. The long-termgoal is to lay the groundwork for the development of equallysophisticated models for stuttering. In the absence of com-pelling evidence to the contrary, we hypothesized thatabnormalities of both structure and function would be found.Based on the diverse nature of the performance abnormal-ities associated with stuttering, we hypothesized that stut-tering would prove to be a disorder affecting several neuralsystems and structures. For this reason, we imaged virtuallythe entire brain. To avoid prematurely restricting the focus ofthis study to a single task or class of behaviors, imaging wasperformed at rest.

Neuroimaging modalities can be categorized as structural,chiefly imaging anatomy, and functional, chiefly imagingphysiology. Structural imaging modalities include x-raycomputed tomography, ultrasonography, and magnetic res-onance imaging (MRI). Functional imaging modalities in-clude electroencephalography (EEG), magnetoencephalog-raphy, positron-emission tomography (PET), single-photon

1210 Journal of Speech and Hearing Research

emission computed tomography (SPECT) and, most re-cently, functional MRI. In the present study, imaging modal-ities combining high spatial resolution and high sensitivity forboth brain structure (MRI) and brain function (PET) wereapplied. To our knowledge, this is the first study to reportMR imaging in developmental stuttering. Functional imaginghas been previously reported in persons who stutter, as willbe discussed.

Functional imaging studies fall into two broad categoriesbased on experimental design: (a) functional-lesion studiesand (b) functional-activation studies. Functional-lesion stud-ies are intended to detect regional physiological abnormal-ities that are disease specific, but not due to gross structuralabnormalities. Functional lesions in the absence of gross,structural lesions are thought to reflect microstructural ab-normalities such as neural agenesis, dysgenesis, or degen-erations. Functional-lesion neuroimaging studies, then, canidentify regional abnormalities associated with psychiatric,developmental, and neuro-degenerative disorders. Exam-ples of functional lesions that have been detected by PETand SPECT include frontal hypometabolism in schizophrenia(reviewed in Gray, Feldon, Rawlins, Hemsley, & Smith,1991), caudate hypometabolism in Huntington's disease(Mazziotta, Frackowiak, & Phelps, 1992), limbic and paral-imbic hypometabolism and hypoperfusion in depression(Baxter, Schwartz, & Phelps, 1989; Mayberg, 1994), andtemporal lobe hypometabolism and hypoperfusion (interic-tally) in partial complex epilepsy (Gaillard et al., 1995).

Functional-activation studies map the neural systemsunderlying specific behaviors. PET activation studies arecharacterized by the use of short-lived tracers (e.g., 150water; half-life 123 s), multiple scans per session, andexperimental designs in which task states and control statesare contrasted within session and within subject. Functionalactivation studies can be used to isolate neural systems ofnormal behaviors as well as behaviors selectively altered bydisease. For example, these studies permit the investigationof neural activity during speaking tasks and have alreadybroadened our understanding of the neural substrates ofspeech and language (reviewed in Liotti, Gay, & Fox, 1994).The literature mapping the neural systems of disorderedbehavior is smaller, but includes cognitive (Gray, 1995;Weinberger, Aloia, Goldberg, & Berman, 1994), emotional(Drevits & Raichle, 1995), motor (Grafton et al., 1994), andperceptual (Neville, 1990) disorders.

Both functional-activation and functional-lesion studieshave been conducted with persons who stutter. The firstpublished functional-activation study was reported byWood, Stump, McKeehan, Sheldon, and Proctor (1980) andattempted to record regional cerebral blood flow (rCBF)during a stuttering and a reduced stuttering condition. Thisstudy reported inadequate left cerebral dominance forspeech production, complementing a variety of non-rCBFstudies showing unusual hemispheric laterality among per-sons who stutter (see Moore, 1990). More recently, somepreliminary functional activation studies using PET have alsostarted to investigate neural systems of persons who stutter(Braun, Ludlow, Varga, Shulz, & Stager, 1994; Braun et al.,1995; DeNil, Kroll, Kapur, & Houle, 1995; Fox et al., 1996;Ingham, Fox, & Ingham, 1994, 1995; Wu, Maguire, & Riley,

1995; Wu et al., 1994, 1995). All of these studies haveassociated atypical neural activations with the speech ofpersons who stutter.

As mentioned above, however, functional-lesion studiesare more applicable to the question addressed in the currentstudy. Such an investigation using SPECT technology wasreported by Pool, Devous, Freeman, Watson, and Finitzo(1991). Resting-state rCBF was measured in 20 adults whostutter and in 78 nonstutterering controls. This study re-ported that the at-rest pattem of hemispheric asymmetriesdiffered between the experimental and control groups. Spe-cifically, Pool et al. reported abnormal asymmetries in re-gions associated with speech motor control, thus inferring astutterer-specific trait.' The same subjects were studied byFinitzo, Pool, Freeman, Devous, and Watson (1991) viaauditory evoked potentials acquired from the temporal cor-tex. These results complement the rCBF findings and ledFinitzo et al. to conclude that " ... stuttering is a neurologicdisorder with focal and diffuse functional lesions at cortex"(p. 259).

The present study is also a functional-lesion study ofdevelopmental stuttering. Its purpose is to determinewhether persons who stutter have abnormalities of brainfunction or structure. The present study serves as well as asystematic replication (Cook & Campbell, 1979; Sidman,1960) of Pool et al. (1991), using the more recent and moresensitive neuroimaging technologies of PET and MRI.

Methods

Subjects

Participants in this study were paid volunteers recruitedby advertisements in local newspapers, posted at localschools, and by word-of-mouth. All participants gave in-formed consent in the manner prescribed by and with formsapproved by the Institutional Review Board of the Universityof Texas Health Science Center. Ten adult men who stutterand 19 similarly aged men who do not stutter were studied.All subjects were physically well, unmedicated, and with nomedical, neurologic, or psychiatric diagnoses. All partici-pants had complete right-body dominance, as scored with

'The SSseffect is best established using a model with hypothesized cell .values, but it is more convenient to use a post hoc estimate, the observedestimate for the effect in question. Power for the Trial x Hemisphereinteraction (see p. 1217) was calculated with SSeffect = 72.388 and MS-for-Error = 13.642, making lambda = 5.30626, with MS-for-Error being the meansquare for Trials x Hemisphere x Subjects pooled over groups. ThisMS-for-Error usage, standard for repeated measurements designs with thesewithin-subjects factors, is not available in O'Brien and Muller's programswithout some modifications.

For a significance level of 0.05, df effect = 2 and df error = 54, all thatremains is to calculate the theoretical F-value for significance, F(.05), 2, 54,and the corresponding power, the probability of attaining that F or a higher Fwhen there is no interaction in the population. Many equivalent procedurescan be used here; ours employs essentially the SAS program as shownbelow:

data power; var2 = 'TRIAL by HE'; f5pcnt_f = finv(.95, 2, 54, 0); lambda =72.388/13.642; power = 1 - probf(f5pcnt f, 2, 54, lambda); proc printdata = power; run;

Output from this program tells us FPCNT__F = 3.16825, LAMBDA =5.3026, and Power = 0.50527. The SAS functions "finv" and "probf" arecentral to this power calculation.

39 1208-1227 December 1996

Ingham et al.: PET Functional-Lesion Investiganon of Stuttering 1211

the Harris Test of Lateral Dominance (Harris, 1958). This testassesses laterality of use of the eyes, hands, and feet.

The 10 men who stutter ranged in age from 22 to 46 years,with a mean age of 33 years. All carried a diagnosis ofchronic stuttering, with childhood onset. No subject hadreceived treatment for stuttering within the previous 3 years.Five subjects reported a family history of stuttering. All themen who stutter underwent a spontaneous speech screen-ing task, administered by the investigators. In this task,stuttering severity was judged by two independent judges(RJI; JCI) using a 3-point rating scale: 1 = mild; 2 =moderate; 3 = severe. Their scores were: mild, 4 subjects;moderate, 2 subjects; severe, 4 subjects. Their stutteringfrequency during the experiment, measured as the numberof 4-s intervals of oral reading judged to contain stuttering(Ingham, Cordes, & Finn, 1993), ranged from 6 to 29 out of30 possible stuttered intervals (3 x 40-s scans). These datawere obtained from a PET functional-activation study ofstuttered speech, from which the present resting state datawere obtained (Fox et al., 1996).

The 19 control subjects ranged in age from 21 to 55 years,with a mean of 29 years. None reported a history of speechor hearing problems. Ten were participants in the above-mentioned functional-activation study (Fox et al., 1996); theremaining 9 were drawn from two concurrent studies thatalso used three eyes-closed rest scans as the control state.The mean ages of the two groups did not differ (t = 0.86;p > 0.05).

Subject Preparation

PET and MRI scans were acquired with each subjectsupine and with his head supported in a foam-padded,hemicylindrical head holder. For PET imaging, head positionwas adjusted with the aid of laser alignment beams toapproximate a plane 15 degrees (rotated clockwise, whenviewed from the subject's left) from a horizontal planepassing between the anterior and posterior commissures(AC-PC plane; Talairach & Toumoux, 1988). At this orienta-tion, the PET planes of section spanned all known speechmotor regions, including the supplementary motor area andcerebellum. Following positioning, the head was immobi-lized within a tightly fitted, thermally molded, plastic facialmask (Fox, Perlmutter, & Raichle, 1985). The mask extendedfrom the hairline to the chin, with openings for eyes, ears,nose, and mouth. An antecubital venous catheter wasplaced for administrations of the PET radiotracer (below). In20 subjects (10 in each group), a catheter was placed in theradial artery, for determining the arterial time-activity func-tion. Catheters were removed prior to MR imaging. Time-activity curves were successfully obtained in 14 subjects (6men who stutter and 8 controls).

PET Imaging

PET images were acquired with a GE/Scanditronix 4096camera. This PET camera simultaneously acquires 15 par-allel slices with a center-to-center interslice distance of 6.5mm and a transaxial field of view (FOV) of 10.0 cms. Images

were reconstructed at an in-plane resolution of 7 mmfull-width at half-maximum (FWHM) and an axial resolutionof 7 mm FWHM. Before emission scanning, a 15-slicetomographic transmission image (8Ge/ 68Ga) was obtainedin order to calculate regional attenuation coefficients.

Water labeled with 15-oxygen (H250, half-life 123 s) wasused as a blood flow tracer (Herscovitch, Markham, &Raichle, 1983; Raichle, Martin, & Herscovitch, 1983). Be-tween 45 and 65 millicuries of H2

150 in 5-10 cc of sterilesaline were delivered as an intravenous bolus. Data acqui-sition was triggered by arrival of the tracer bolus in the fieldof view (15-20 s after tracer injection) and continued for 90 s.Each subject underwent a series of 9 measurements of brainblood flow. Subjects were scanned during 3, 40-s eyes-closed rest conditions; 3, 40-s periods of continuous read-ing (Task 1); and 3, 40-s periods of chorus reading (Task 2).Each subject experienced these scans in one of two coun-terbalanced orders: Rest-Task 1-Task 2-Rest-Task 2-Task1-Rest-Task 1-Task 2 or Rest-Task 2-Task 1-Rest-Task1-Task 2-Rest-Task 2-Task 1. For the questions addressedin this study, only data from the rest condition, which areentirely independent of the activation task conditions, arepertinent. A 10-15 min interscan interval occurred, whichwas sufficient for isotope decay (5 half-lives) and to reestab-lish resting levels of CBF before the subsequent trial (Ma-lonek & Grinvald, 1996). Prior to regional analysis, PETimages were value normalized to a whole-brain CBF of 50ml/1 00 gm/min (Fox, Burton, & Raichle, 1987; Fox & Raichle,1984).

MRI Imaging

MRI data were acquired on an Elscint Gyrex 2T-DLX. Thissystem operates at 1.9 tesla and has an inner diameter of 90cm. A three-dimensional, fast spin echo, T-1 weightedsequence was used. Voxel size was 0.9 x 0.9 x 0.9 mm.The entire brain was imaged.

Spatial Normalization

PET and MR images were spatially normalized (Fox et al.,1985) relative to the atlas of Talairach and Toumoux (1988),a now widely accepted procedure for interpreting imagingdata (Evans, Collins, & Holmes, 1996; Ford, 1995; Friston,1995; Strother, Kanno, & Rottenberg, 1995; Woods, 1996).Spatial normalization was performed with the SN software,using the method of Lancaster et al., (1995). Parameterscharacterizing slice location and orientation were extractedfrom each modality independently. Parameters characteriz-ing brain dimensions were obtained from the MR image andapplied to both the MR and PET images. Using theseextracted parameters, images were transformed into 3D,spatially normalized image volumes, with isomorphic voxels,2 mm x 2 mm x 2 mm (Figures 1A and 1B).

Regional Sampling

PET images were regionally sampled using two, indepen-dent, mutually complementary strategies: physiological tar-


FIGURE 1. Group-mean MR images for right-handed men who stutter (A; n = 10) and a control group of age-matched,right-handed nonstuttering men (B; n = 19). There were no gross differences in brain morphology between the two groups, asis evident from visual inspection of these images. The slight blurring of these images, in comparison to the MRI of asingle-subject (see Figure 2), is an accurate index of the residual anatomical variance among subjects (within group) that wasnot removed by the spatial normalization process. The planes shown are a subsampling of a 3D image volume. The numberbeneath each plane is its distance in millimeters above (+) or below (-) the AC-PC reference plane.

39 1208-1227 December 996


1214 Jouman! Spech and Hearing Research

geting and anatomical targeting. Physiologically targetedregions of interest (ROls) were defined by speech activationfindings acquired from 20 of these 29 subjects. Anatomicallytargeted regions were defined with reference to the Ta-lairach and Toumoux (1988) atlas. Using these two strate-gies, a total of 74 ROls, 37 per hemisphere, were defined.ROls were either 216 cubic mm (3 x 3 x 3 voxels) or 1000cubic mm (5 x 5 x 5 voxels), where voxels are cubic areaswith sides 2 mm in length. As much as possible, ROlssampled grey matter, avoiding white matter and cerebrospi-nal fluid. ROI names and locations are listed in Table 1.

Physiologial regions of interest. In 20 of the 29 sub-jects (10 men who stutter and 10 who do not stutter), PETimages during two speech tasks, chorus and nonchorusreading, were acquired (Fox et al., 1996). By statisticalcontrast with the resting state, brain regions engaged during

speech were identified in group-mean images created foreach group. Eighteen (9 per hemisphere) regions activatedduring speech output were identified. These regions were:superior supplementary motor area (SSMA); inferior SMA(ISMA); superior, lateral premotor cortex (SLPrM); primarymotor cortex (M1); inferior, lateral premotor cortex (ILPrM);primary auditory cortex (Al); secondary auditory cortex (A2);anterior insula (INS); and superior, anterior cerebellum(CBM). The mean location coordinate (x-y-z) for each regionwas determined. This mean coordinate was then projectedback onto the images of each individual in order to extractregional values, trial-by-trial, and person-by-person.

Anatomical regions of interest. Anatomical regions ofinterest were chosen to broadly sample the brain, includingregions outside the areas implicated in speech production.In cerebral cortex, regions were selected from all cytoarchi-

TABLE 1. The 37 regional sampling areas (ROts) from each heiphere for PET data (providing74 ROIs), which were identified as a result of either physidogical activation or anatoicalinformation (see Methods). The ROls are referenced to ther Brodmann Areas (BA), wherepossible, plus lobe, gyrus, and region name as specified in the Talairach and Toumnoux (1988)atlas. The mean atlas coordinates for ROis are also shown (x = right[+k-); y = Warior[+]/posterior -]; and z = superior [+]/inferr [-. The 11 ROls ampled in Pool et al. (1991) areidentified by astersks (*) with those showing significant betwern-up dlerences highlighted

by double asterisks (*). (See Appendix for explanation of abbrevatons).

BA Lobe Gyrus Region ( z

Physiological ROls6 Frontal GFd SMAi 5 2 486 Frontal GFd SMAs 5 2 586 Frontal GPrC SLPrM 51 -4 384 Frontal GPrC MI 48 -9 36

44 Frontal GPrC ILPrM 56 -6 1941 Temporal GTs Al 50 -25 622 Temporal GTs A2 56 -21 2na Insula INS INS 34 12 -2na Quadrangular na CBM 6 -64 -16

Anatomical ROls10 Frontal GFs SFG* 10 60 832 Limbic ant GC AG** 6 40 810 Frontal GFm MFG* 34 48 846 Frontal GFi IFG* 44 32 8

6 Frontal GPrC P/PG' 50 2 1022 Temporal GTs S/STG * 54 -18 621 Temporal GTm MTG** 52 -50 437 Temporal GTi ITG* 44 -70 018 Occipital Cu O* 16 -88 1623 Occipital PCu C/P * 10 -46 24na Sub-lobar na thalamus* 6 -14 8na Sub-lobar na caudate 12 12 8na Sub-lobar na putamen 26 6 4na Sub-lobar na GP 18 2 0

8 Frontal GFs GFs 10 32 489 Frontal GFs GFs 10 52 30

1/2/3 Parietal GPoC GPoC 44 -24 524 Frontal GPrC GPrC 32 -16 52

40 Parietal LPi LPi 48 -46 4217 Occipital Cu/GL Cu/GL 6 -88 425 Umbic Gs Gs 6 10 -1035 Temporal Gh Gh 24 -30 -1030 Umbic GC GC 8 -48 1038 Temporal GTs GTs 48 6 -1039 Temporal GTm GTm 44 -66 1843 Parletal GPoC GPoC 54 --16 1844 Frontal GFi GFi 48 6 2647 Frontal GFi GFi 44 30 - 6

39 1208-1227 D~ecemer 996


tectonic zones, insofar as possible. Cytoarchitectonic zonesemployed were those of Brodmann, as localized byTalairach and Tournoux (1988). In all, 27 (of 41) Brodmannareas (BA) were sampled using 24 regions. BAs were un-sampled if they fell beyond the planes of section in anysubject. Excluded areas were: posterior, superior parietalcortex, orbito-frontal cortex, and inferior temporal cortex.None of the unsampled areas fell within known speechmotor areas. In Table 1 regions falling in or near the 11 brainareas sampled by Pool et al. (1991) are identified by aster-isks. Four major subcortical nuclei were also sampled:thalamus, caudate, putamen, and globus pallidus.

Anatomically defined ROls were initially positioned relativeto the BA and nuclear designations of the Talairach andToumoux atlas (1988). ROls were positioned within theirrespective anatomical regions so as to be placed within thecenter of the anatomical region and free of partial-volumeinteractions with neighboring ROls. Once positioned relativeto the atlas, the coordinates for each ROI were read from theatlas. Using these coordinates, ROls were projected ontothe spatially normalized MR images of each subject. Onceplaced onto the MRI, the ROI was confirmed and adjusted,if necessary. Repositioning was limited to centering ROlswithin the grey-matter. No region was moved more than twovoxels in any direction. ROI locations are illustrated inFigure 2.

Statistical Analysis

Regional sampling provided 6,438 values: 37 ROls, for 2Hemispheres, for 29 Subjects (forming 2 groups), for threeTrials per subject. Each independent value was the mean of27 (for ROls of size 3 x 3 x 3 voxels) or 125 (for ROls of size5 x 5 x 5 voxels) voxel values. Values for each subject wereindependent from each other and deviations from sphericity(noted below) were minimal. Analysis was progressive, be-ginning with a multiple-factor omnibus analysis (Winer,Brown, & Michels, 1991). Factors not identified as significantby omnibus testing were not further tested. The likelihoodthat any true differences between experimental groups weremissed was assessed using a power analysis.

Omnibus ANOVA. A repeated measures, multiple factorsanalysis of variance (ANOVA) was performed using theSuperANOVA statistical package (Gagnon et al., 1989) run-ning on a Macintosh Quadra. The ANOVA design was"one-between, three-within," where the "between" factorwas experimental Group (men who stuttered and men whodid not stutter); and the "within" factors were ROI, Hemi-sphere, and Trial. Main effects and interactions were testedusing the Huynh-Feldt correction for sphericity effects in arepeated measures ANOVA (Huynh & Feldt, 1970).

Power analysis Power analyses were conducted to esti-mate the sensitivity of the Omnibus Analyses. Power wascomputed for all nonsignificant interactions with a p < 0.1,using the procedure of O'Brien and Muller (1993) with anadaptation for the repeated measures design. Absent a prioriestimates of variance, the appropriate Mean Squares for errorand degrees of freedom (see Table 2) were applied using analpha level of 0.05. The protected effect sizes (lambda

values) were calculated as follows: (SSeffts)/(MSfor.error) byanalogy to O'Brien and Muller's (1993) equation 8.39.2

Laterality Index ANOVA. Because the omnibus ANOVAfound a main effect of Hemisphere for both groups, andbecause prior studies have found unusual regional asymme-tries among persons who stutter (cf., Moore, 1990), alaterality analysis was performed, as follows. For each ROI,a Laterality Index was computed as: 100 (right - left)/(right + left)/2. Laterality Indices were entered into a re-peated measures, "one-between, two-within" ANOVA,where the "between" factor was Group and the "within"factors were Trial and ROI.

Results

The most striking finding was a lack of evidence forfunctional (or anatomical) lesions in the brains of men whostutter. This absence of functional lesions was obtaineddespite high statistical sensitivity, as indicated by: (a) thedetection of very slight regional and hemispheric asymme-tries in both groups, and (b) formal power analyses.

Visual inspection of the PET and MRI scans for eachindividual revealed no gross abnormalities in any brain areafor any subject. Visual inspection of the group-mean MRI(Figures 1A and 1B) and the group-mean PET (Figures 3Aand 3B) revealed no differences between the groups foreither mode.

Omnibus Analysis

Two significant (p < 0.05) main effects and one significantinteraction were identified (Table 2). The main effects andthe interaction were common to both groups. There were nosignificant between-group interactions.

Main effects. Highly significant main effects of ROI andHemisphere were observed for both groups. The ROI effectsreflect the visually obvious (Figures 3A and 3B) and well-known differences in resting state blood flow among differ-ent regions of the brain.

The Hemisphere effects indicate a difference in the meanblood flow values between the two hemispheres. Thesedifferences were minor, less than 2%, but highly significant

2Pool et al. (1991) also reported (but did not discuss or interpret) a morestatistically significant and more controversial finding: whole-brain blood flowthat was consistently lower (20% group-mean difference) in the persons whostutter than in the nonstuttering controls. (See Fox, Lancaster, & Ingham[1993] and Pool, Finitzo, Devous, Watson, & Freeman [1993] for a critique anddiscussion of this finding and other aspects of Pool et al.). When using H2O

50as a radiotracer, analysis of whole-brain blood flow requires kinetic modelingusing time-activity values determined from rapid, repeated sampling of thearterial blood during the scan. In the current study, time-activity curves wereobtained and whole-brain blood flow was quantified (Herscovitch et al., 1983;Raichle et al., 1983) in 6 of the men who stutter (mean = 56.0 ml/1 00 gm/min;SD = 15.81) and 8 of the men who do not stutter (mean = 47.2 ml/100gm/min; SD = 9.82). The two groups did not differ, by unpaired t-test (t =0.567; p > 0.1). In addition, their compared mean whole-brain blood flowvalues were in the opposite direction to those reported by Pool et al. However,a limitation of the present study is that we were unable successfully to insertarterial lines in all of our subjects, hence the statistical power of this analysisis too low to confidently contradict the finding of Pool et al. The power is lowbecause a whole-brain analysis uses only a single value from each subject,thus greatly limiting the available degrees of freedom.

--- --

1216 Journal of Speech and Hearing Research 39 1208-1227 December 1996

FIGURE 2. MRI of a single subject. The planes shown are a subsampling of a 3D image volume. The white squares on the imagesrepresent ROls, and accurately reflect ROI locations and sizes. ROls not falling on the illustrated planes are not shown. ROIlabels are explained in Table 1. ROI locations were established as described in the text.


TABLE 2. Summary table for 3-factor repeated measures (unequal groups) ANOVA using ROIdata. The three factors are Group, Region of Interest (ROI), and Hemisphere (HE), with repeatedmeasures across Trials. F probabilities, corrected for sphericity effects using a Huynh and Feldt(1970) procedure, are shown in the final column (H-F).

Source of variance df SS MS F p H-F

GROUP 1 224.79 224.79 0.82 0.37Subject (Group) 27 7409.87 274.44Trial 2 22.2 11.1 0.27 0.76 0.76Trial x GROUP 2 112.68 56.34 1.38 0.26 0.26Trial x Subject (Group) 54 2203.29 40.8ROI 36 149862.46 4162.85 34.66 0 0ROI x GROUP 36 3667.12 101.86 0.85 0.72 0.66ROI x Subject (Group) 972 116738.78 120.1HE 1 1985.2 1985.2 32.35 0 0HE x GROUP 1 119.19 119.19 1.94 0.17 0.17HE x Subject (Group) 27 1657.05 61.37Trial x ROI 72 1423.35 19.77 1.37 0.02 0.07Trial x ROI x GROUP 72 971.63 13.5 0.93 0.64 0.59Trial x ROI x Subject (Group) 1944 28118.89 14.46Trial x HE 2 72.39 36.19 2.65 0.08 0.08Trial x HE x GROUP 2 2.19 1.1 0.08 0.92 0.92Trial x HE x Subject (Group) 54 736.67 13.64ROI x HE 36 3940.95 109.47 2.01 0 0ROI x HE x GROUP 36 2891.11 80.31 1.48 0.04 0.06ROI x HE x Subject (Group) 972 52912.28 54.44Trial x ROI x HE 72 791.51 10.99 1.01 0.47 0.46Trial x ROI x HE x GROUP 72 672.65 9.34 0.86 0.8 0.75Trial x ROI x HE x Subject (Group) 1944 21251.09 10.93

for each group (p < 0.0001). In both groups the direction-ality of the effect was the same, right > left, which is to beexpected (Perlmutter, Herscovitch, Powers, Fox, & Raichle,1985; Perlmutter, Powers, Herscovitch, Fox, & Raichle,1987). In controls the asymmetry was 2.5% (58.6, left; 60.1,right); in persons who stutter, the asymmetry was 1.5%(59.3, left; 60.2, right). This difference between groups wasnot significant (Hemisphere x Group interaction, p = 0.175).(Note that this right > left asymmetry pertains to "wholebrain" laterality rather than the commonly acknowledgedleft > right laterality of the language system.)

Interactions. The ROI x Hemisphere interaction washighly significant (p < 0.002). This effect indicates that theoverall Hemisphere effect (right > left) was not uniformthroughout the brain. Rather, the directionality of the asym-metries differed across regions. This finding is further ex-plored in the Laterality-Index Analysis (see below).

Power Analysis

Three near-significant interactions (0.1 < p > 0.05) werefound: Trial x ROI, Trial x Hemisphere, and ROI x Hemi-sphere x Group (Table 2).

For the Trial x ROI interaction, the power analysis re-vealed that confidence in nonsignificance was p = 0.9999.That is, there was less than a one-in-1 0-thousand chance ofmissing true trial x ROI interaction effects of size lambda orgreater. Because each of the three resting-state scans wasseparated by at least 45 min, the lack of significant Trialeffects indicates the consistency of multiple measurementsin the resting state. This consistency validates both the useof multiple trials as a means of increasing the statisticalpower of this investigation and the use of a resting state asthe measurement condition.

For the Trial x Hemisphere interaction, confidence innonsignificance was p = 0.51. This low power likely is due tothe much smaller degrees of freedom (df) for this compari-son (54) than for the Trial x ROI analysis (df = 1944) or forthe ROI x Hemisphere x Group analysis (df = 972). Thus, aTrial x Hemisphere interaction cannot be excluded.

For the ROI x Hemisphere x Group interaction, confi-dence in nonsignificance was p = 0.994. That is, there is aless than a six-in-one-thousand chance of missing trueeffects of size lambda or greater. Thus, it can be concludedwith high confidence that the experimental group (of menwho stutter) did not have resting state regional differences inbrain blood flow (functional lesions) when compared to anage-matched control group of men who do not stutter.

Laterality Index ANOVA

One significant (p < 0.05) main effect (ROI) and onesignificant interaction (ROI x Group) were identified(Table 3). The main effect replicates the ROI x Hemisphereinteraction observed in the Omnibus ANOVA. The ROI xGroup interaction indicates that regional asymmetries dif-fered between the groups for some brain regions. Theregions responsible for this effect were identified by t-test,as follows.

Laterality Index t-Tests

Between-group differences in hemispheric symmetrywere identified for five brain regions (Table 4). Because theinput values for this analysis are asymmetry indices, thesefindings do not contradict the absence of a Group x ROI x


FIGURE 3. Group-mean PET images for right-handed men who stutter (A; n = 10) and a control group of age-matched,right-handed nonstuttering men (B; n = 19). The extensive field-of-view of the PET image data is visually apparent. There wereno gross differences in regional brain physiology between the two groups, as is evident from visual inspection of these images.The blurring of these images (greater than in Figures 1 or 2) reflects the combined effects of unremoved anatomical variance(see Figure 1 legend) and the limited spatial resolution of PET (here, 7 mm FWHM). The planes shown are a subsampling of a3-D image volume. The number beneath each plane is its distance in millimeters above (+) or below (-) the AC-PC referenceplane.

39 1208-1227 December 996



TABLE 3. Summary table for 2-factor repeated measures (unequal groups) ANOVA usingLaterality Index scores. The two factors are Group and Region of Interest (RO), with repeatedmeasures across Trials. F probabilities, corrected for sphericity effects using t Huynh and Feldt(1970) procedure, are shown in the final column (H-F).

Source df SS MS F p H-F

GROUP 1 725.71 725.71 2.00 0.17Subject (Group) 27 9777.35 362.12Trial 2 398.93 199.46 2.57 0.09 0.09Trial x GROUP 2 4.15 2.08 0.03 0.97 0.97Trial x Subject (Group) 54 4196.84 77.72ROI 36 25652.09 712.56 2.25 0.00 0.00ROI * GROUP 36 17185.55 477.38 1.51 0.03 0.05ROI * Subject (Group) 972 308181.31 317.06Trial * ROI 72 4362.68 60.59 0.95 0.59 0.56Trial * ROI * GROUP 72 3787.22 52.60 0.83 0.85 0.80Trial * ROI * Subject (Group) 1944 123419.23 63.49

Hemisphere effect in the Omnibus Analysis. Rather, thisanalysis detects left-right balance, even when the inputvalues are in the range of normal. Further, these five regionalsymmetry differences were isolated using a significance

level of p < 0.05, without a correction for multiple compar-isons, which none of these findings withstood. Nonetheless,in order to consider even the most remote possibility that thestatistical analyses somehow obscured meaningful between-

TABLE 4. Means and standard deviations of Laterality Index scores for the subjects whostuttered (n = 10) and subjects who did not stutter (n = 19) in 37 ROls. Negative seres indicateleft laterality. Significant p values (without Bonferroni correction) are marked by pluses (+),indicating ROIs where there were signs of asymmetry differences between the groups. The 11ROls sampled in Pool et al. (1991) are identified by asterisks (*), with those showing significantbetween-group differences highlighted by double asterisks (*). (See Appendix for explanationof abbreviations.)

Stutterers Controls

Region Mean (SD) Mean (SD) p value

SMAi 8.2 (7.50) 7.2 (8.24) 0.699SMAs 1.0 (5.86) 3.2 (6.03) 0.265SLPrM 14.6 (9.40) 12.2 (11.85) 0.518MIl 2.7 (5.39) 0.7 (7.00) 0.367ILPrM 6.9 (7.31) -0.2 (9.71) 0.023+Al -5.1 (10.45) 1.0 (6.86) 0.035+A2 1.1 (8.67) 1.9 (7.10) 0.761INS -0.2 (6.92) 0.9 (8.12) 0.651CBM -0.6 (9.84) 2.4 (8.30) 0.308SFG' 0.7 (5.62) 0.9 (9.37) 0.952AG** -2.2 (20.01) 5.6 (20.09) 0.285MFG* -4.5 (11.07) 3.9 (10.74) 0.020+IFG' -3.2 (14.96) 3.5 (14.02) 0.180P/PG' 5.3 (16.51) -0.8 (9.88) 0.135S/STG** -1.9 (14.86) 2.2 (11.70) 0.334MTG** 9.1 (15.57) 8.4 (15.67) 0.911ITG* -4.3 (10.72) 1.5 (17.10) 0.291O* 5.3 (14.34) 6.1 (11.74) 0.865C/P* -3.1 (11.48) 3.1 (15.32) 0.221thalamus* 4.4 (10.27) 2.2 (10.96) 0.548caudate 6.7 (17.69) 0.0 (12.05) 0.168putamen - 1.2 (11.26) 0.4 (10.47) 0.625GP 4.5 (13.65) 2.8 (11.35) 0.042+GFs 1.1 (8.05) 3.6 (11.06) 0.399GFs -1.3 (11.36) 3.5 (11.39) 0.199GPoC -2.3 (9.49) 2.2 (9.19) 0.146GPrC 2.4 (9.84) -2.7 (14.93) 0.261LPi 1.9 (10.16) 2.6 (10.18) 0.841Cu/GL 1.9 (15.11) 1.2 (8.24) 0.866Gs 0.6 (10.40) 3.7 (12.23) 0.347Gh -3.5 (14.84) 3.7 (12.64) 0.099GC -1.4 (13.26) 9.0 (10.53) 0.003+GTs 0.0 (19.47) 1.6 (13.07) 0.778GTm 0.8 (11.28) 5.4 (11.86) 0.115GPoC 0.4 (18.88) 5.2 (14.70) 0.431GFi 6.7 (11.29) -1.6 (15.65) 0.111GFi 5.2 (13.13) 3.3 (13.32) 0.676

39 1208-1227 December 1996


group differences, these near-significant regions were fur-ther examined.

Differences in Laterality Indices were largely limited tobrain areas implicated in speech production, but did notshow a consistent directionality. ILPrM (BA 44), an area ofthe cerebral cortex involved in motor planning (Fox, 1995;Parsons et al., 1995), was strongly right lateralized (+6.9) inthe men who stutter and weakly left lateralized in thecontrols (-0.2). Al (BA 41), the primary auditory area of thecerebral cortex, showed the reverse effect, with strong leftlaterality (-5.1) in the men who stutter and weak rightlaterality (+1.0) in the controls. Middle frontal gyrus (MFG,lateral BA6), an area implicated in motor planning (Roland,1993, pp. 260-262), was left lateralized (-4.5) in the menwho stutter, but right lateralized (+3.9) in the controls.Globus pallidus, a subcortical nucleus involved in motorcontrol, was right lateralized in men who stutter (+4.5), butleft lateralized (-2.8) in controls (Roland, 1993, pp. 262-264;Strick, Dum & Picard, 1995). Finally, posterior cingulategyrus (BA 29/30), an area implicated in visual attention and

imagery (Roland, 1993, pp. 260-361), was left lateralized(-1.4) in the men who stutter and strongly right lateralized(+9.0) in the controls. Thus, inconsistent laterality differ-ences were noted, with three of the five areas more leftlateralized in the men who stutter.

Values x ROI. The sources of the regional asymmetriesdescribed in the preceding paragraph can be identified byinspection of the actual ROI values. The mean values andstandard deviation for all ROls, for each Hemisphere andGroup are provided in Table 5.

Inspection of each ROI value by comparison to the meanof the two values (right and left) for the controls reveals thatthe only consistent finding is that the men who stutter nevershowed a bilateral decrease in blood flow. For two ROls(ILPrM and cingulate [GC]), both values were higher than themean of the controls, but asymmetrically so. For the remain-ing three regions, the left-sided value was higher thancontrols in two instances (Al and MFG) and the right-sidedvalue was higher than controls in one instance (GP). Thus, aswith the Laterality Indices, no consistent lateralizing effect

TABLE 5. Mean cerebral blood flow (rCBF) values for the subjects who stuttered (n = 19) and subjects who did not stutter (n = 19)in 37 ROls in both hemispheres. Bilateral mean scores ([L + R]/2) for the controls are shown for purposes of comparison with scoresby the subjects who stutter. The 11 ROls sampled in Pool et al. (1991) are identified by asterisks (*), with those showing significantbetween-group differences highlighted by double asterisks (**). ROI asymmetry differences indicated in Table 4 are marked by pluses(+). (See Appendix for explanation of abbreviations).

Stutterers Controls

Left Right Bilateral Left RightRegion mean (SD) mean (SD) mean mean (SD) mean (SD)

SMAi 57.8 (4.18) 62.6 (3.13) 61.6 59.3 (3.57) 63.9 (4.91)SMAs 54.3 (5.10) 54.8 (4.94) 56.2 55.4 (5.89) 57.1 (6.12)SLPrM 50.8 (9.16) 58.4 (7.77) 52.7 49.6 (5.42) 55.9 (4.25)MI 60.7 (4.68) 62.4 (5.21) 61.7 61.5 (3.49) 61.9 (3.16)ILPrM 55.0 (7.67) 58.8 (7.66) 54.3 54.4 (3.17) 54.3 (4.41)+Al 65.9 (6.92) 62.4 (4.76) 62.8 62.4 (4.45) 63.1 (4.52)+A2 62.5 (4.55) 63.1 (3.78) 61.3 60.8 (3.82) 61.9 (2.93)INS 65.8 (4.90) 65.6 (3.59) 67.7 67.3 (3.73) 68.0 (5.55)CBM 60.7 (5.65) 60.4 (5.70) 62.0 61.2 (4.99) 62.7 (5.50)SFG* 56.3 (3.21) 56.7 (2.60) 55.6 55.3 (3.62) 55.9 (3.83)AG** 65.0 (6.82) 64.0 (9.54) 61.8 60.3 (10.27) 63.3 (7.01)MFG* 66.6 (5.71) 63.7 (5.77) 63.5 62.3 (5.83) 64.7 (6.42)+IFG* 59.6 (4.73) 57.9 (6.43) 59.6 58.6 (5.67) 60.7 (5.59)P/PG* 62.6 (7.13) 66.4 (9.60) 67.0 67.3 (5.88) 66.7 (5.57)S/STG** 64.3 (6.95) 63.0 (6.13) 62.2 61.6 (5.53) 62.9 (5.60)MTG** 54.6 (6.54) 59.9 (8.12) 57.3 54.9 (6.46) 59.7 (5.77)ITG* 57.0 (7.16) 54.5 (6.20) 56.8 56.4 (7.70) 57.2 (8.09)0* 49.0 (6.98) 51.6 (6.94) 50.9 49.4 (5.21) 52.5 (5.19)C/P* 49.6 (5.10) 48.3 (6.97) 47.8 47.0 (5.51) 48.6 (6.49)thalamus* 73.0 (8.94) 76.2 (8.02) 73.6 72.7 (6.21) 74.4 (7.63)caudate 58.2 (10.73) 61.4 (6.25) 61.1 61.1 (5.70) 61.1 (5.64)putamen 68.2 (7.05) 67.5 (7.31) 67.6 67.5 (6.01) 67.7 (5.66)GP 58.3 (7.98) 61.0 (7.90) 59.4 60.2 (5.58) 58.5 (6.04)+GFs 53.9 (5.97) 54.3 (4.56) 56.8 55.8 (6.11) 57.7 (4.74)GFs 55.2 (4.92) 54.6 (5.36) 56.2 55.2 (5.74) 57.2 (6.00)GPoC 54.5 (5.08) 53.3 (5.65) 55.6 54.9 (4.11) 56.2 (4.99)GPrC 53.9 (5.41) 55.2 (4.52) 53.9 54.6 (5.50) 53.3 (6.53)LPi 61.2 (5.56) 62.4 (5.57) 61.9 61.1 (5.26) 62.8 (6.58)Cu/GL 65.1 (6.94) 66.4 (7.67) 64.3 64.0 (6.87) 64.6 (4.96)Gs 61.3 (4.55) 61.8 (6.15) 60.7 59.7 (6.56) 61.8 (5.24)Gh 57.5 (8.33) 55.1 (5.12) 53.8 52.8 (5.53) 54.8 (4.96)GC 56.5 (11.46) 55.5 (9.90) 53.6 51.1 (5.23) 56.1 (6.58)+GTs 55.8 (6.42) 56.5 (10.48) 53.8 53.3 (5.82) 54.3 (6.54)GTm 60.0 (4.24) 60.6 (5.38) 59.9 61.5 (6.47) 58.4 (6.74)GPoC 64.5 (10.77) 64.5 (8.92) 62.1 60.5 (7.02) 63.8 (7.76)GFi 58.0 (7.11) 61.8 (4.86) 59.1 59.6 (7.75) 58.7 (7.83)GFi 61.7 (6.16) 65.0 (6.17) 60.3 59.3 (5.21) 61.4 (5.48)


was observed that differentiated men who stuttered fromthose who did not.

DiscussionThe present study, which is methodologically and statis-

tically powerful, was unable to provide support for thehypothesis that brain blood flow in the resting state differ-entiates men who stutter from men who do not. By visualinspection, all PET and MRI images for both subject groupswere normal. The MRI images used in this study were of veryhigh spatial resolution (voxels of <1 mm in all dimensions),high signal-to-noise, and high contrast-to-noise. Visual in-spection of these images conclusively rules out small infarc-tions due to focal ischemic lesions (such as a postulated byFinitzo et al., 1991) and structural developmental abnormal-ities such as hemispheric atrophy, dysplastic gyri, or corticalmigrational failures (such as have been reported in dyslexiaby Geschwind and Galaburda, 1984, and others). Nonethe-less, because detailed morphometric analyses of theseMRIs were not conducted, it is possible, although unlikely,that subtle differences in brain anatomy may exist. By ROIanalysis, the only statistically significant findings were thebetween-region and between-hemisphere-but not be-tween group-differences that constitute the normal physi-ological terrain of the brain. Only in the Laterality IndexAnalysis-an analysis targeting regional hemispheric asym-metries-was a weakly significant (p < 0.045) between-groups effect detected and isolated to 5 (of 37) brain regions.Although only one of these regions was confirmed in thisgroup as being active during speech, four of five regionswere brain areas likely related to speech and hearing. Theseregional asymmetry differences were too weak to withstanda correction for multiple comparisons and varied in theirsense. Inspection of the mean rCBF values (left and right) ofthese same regions again confirmed that regional CBF didnot differentiate men who stutter from those who do not.

Present findings, then, do not support theories or previousresearch findings using other methodologies indicating thatdevelopmental stuttering is associated with or due to focal,functional lesions (e.g., Pool et al., 1991). 3 Nor do thepresent findings support the postulate that developmentalstuttering is associated with an absence of normal asymme-try between the cerebral hemispheres (Travis, 1978; Web-ster, 1993), normal asymmetry being readily detected inboth groups. In view of the largely negative findings of thisstudy, a methodological critique would seem to be in order.

Methodological Critique of the Present Study

Overall, the present study has methodological advantagesover prior functional-lesion studies of stuttering (see below).A key factor contributing to this strength was meticuloussubject selection; an attendant limitation of this degree of

31t is of interest that the findings of the present study have been recentlyreported to have been replicated (Braun, 1996) in another H150 PET study(Braun et al., 1995).

selectivity was a relatively small cohort. Another importantstrength of this study was the use of H2

15 0 PET to obtainhighly accurate measurements of brain bloodflow in a re-peated-measures design. Finally, regional sampling wasboth extensive and exact. ROls were placed by the com-bined use of physiological and stereotactic guidance.

Subject selection. Subject selection was quite rigorous.Experimental and control groups were closely age-matchedand limited to right-handed, adult men. These restrictionswere imposed to avoid confounding by known and possibleregional effects on brain blood flow of handedness, sex, andage (Duara, Lowenstein, & Barker, 1990; Gur et al., 1995).The experimental group was limited to persons with chronic,developmental stuttering, because late onset of stuttering isfrequently associated with identifiable brain damage(Rosenbek, 1984) and its relationship to the more commondevelopmental stuttering is unknown.

The chief limitation imposed by the stringent subjectselection criteria was a relatively small number of subjects.This limitation was overcome, in part, by performing multi-ple, independent scans on each subject. The present studydid not include a test for any possible relationship betweenstuttering severity and regional brain blood flow. This is animportant topic for future investigations, although a largercohort of subjects spanning the range of stuttering severitywill likely be needed.

PET imaging. Although the PET methods employed areamong the most modern, they have inherent limitations infield of view and spatial resolution. The field of view (10.0cm) failed to sample the superior aspects of the parietallobes and the most inferior aspects of the frontal andtemporal lobes in some subjects. Brain areas not included inall subjects were not included in the analysis. By design, theunsampled areas were areas not known to be involved inspeech or hearing. Thus, this limitation should be of little orno practical consequence.

The spatial resolution of the present study was 8 mm. ForPET and SPECT, the sensitivity with which a functionallesion can be detected is directly related to the spatialresolution (FWHM) of the imaging device. Higher spatialresolution (smaller FWHM) allows less profound abnormali-ties to be detected (Hoffman, Huang, & Phelps, 1979;Huang, Hoffman, Phelps, & Kuhl, 1980; Mazziotta, Phelps,Huang, & Kuhl, 1982). Thus, although spatial resolution inthe present study was quite good, use of a PET camera withstill higher spatial resolution could further improve its sensi-tivity.

Three additional benefits of the present PET study overSPECT studies should be noted. By measuring regionalattenuation coefficients in each subject and by using morepenetrant (higher energy) gamma emitters, PET has far moreuniform sensitivity and a better signal-to-noise ratio through-out the brain. By using a short-lived radiotracer (H21'50), PETpermitted multiple, independent measurements to be madein each subject, thereby increasing statistical power.

Data analysis. In the present study, analysis was re-stricted to 74 (37 per hemisphere) rather small ROls. ROlswere made small to minimize mixing of values from hetereo-geneous tissues and, thereby, to increase sensitivity. SmallROls, if well positioned, are more sensitive to regional

39 1208-1227 December 996


abnormalities than are large ROls. If poorly positioned, theycan be less sensitive than larger ROls. The sensitivity of thisstudy, then, depends on its ROI placement. Rather than relyupon a single ROI placement strategy, two were used.Physiological ROI guidance exploited the fact that 20 of the29 subjects had participated in a functional-activation studyduring which their speech motor systems were mapped.Physiological ROls were placed specifically on these areas.

Anatomical ROI placement used registration techniquesdeveloped for stereotactic neurosurgery. ROls were distrib-uted throughout the brain, sampling most of the 41 Brod-mann cytoarchitectonic regions, as indicated in the Ta-lairach and Toumoux atlas (1988). Great care was taken toposition ROIs sufficiently far from one another to assureindependence of sampling. For large cytoarchitectonic ar-eas, ROls were placed in locations likely to be involved inspeech or hearing or to approximate ROI placements in priorstudies. For every ROI in every subject, placement wasconfirmed and refined by use of a coregistered, high reso-lution MRI.

The use of ROIs has inherent limitations. Optimal sensi-tivity can only be obtained when the entire ROI lies wellwithin the bounds of the abnormal area. (A distance of twicethe FWHM between the inner border of the abnormality andthe outer border of the ROI is most sensitive (Hoffman et al.,1979; Huang et al., 1980; Mazziotta et al., 1982.) Suchplacement is attainable when regional abnormalities arevisually apparent (e.g., in focal epilepsy) or the location ofthe pathology is well defined (e.g., caudate nucleus hypo-metabolism in Huntington's disease). When regional abnor-malities are not known or seen, ROI placement must rely onother strategies, as is done here. Despite the care exercisedin ROI placement in the present study, the possibility re-mains that regional abnormalities do exist in areas that wereunsampled.4

The number of ROls employed in this study might becriticized as being too large. Inclusion of large numbers ofROls from normal brain areas can potentially dilute theobservations and weaken the power of an omnibus analysis.Two findings mitigate this criticism. First, the five regionalasymmetry (marginal) differences observed between groupswere contributed by each of the ROI placement strategiesand were found in very different brain areas: physiologicallyand anatomically defined ROls, cortical and subcorticalstructures, speech and nonspeech areas. Thus, any reduc-tion of our sampling would likely have reduced our findings.Second, power analyses (performed on interactions near thesignificance threshold) indicated that present sensitivity for

4A strategy for physiological ROI placement, considered but not used, was toplace and adjust the location of each ROI on a per-person basis, so as toexactly match individual activation locations, rather than the group-meanlocations. This strategy might have increased the precision of placement by afew mm, the typical between-subjects standard deviation of speech motorareas (Fox & Pardo, 1991), but would have limited the number of subjects andareas studied. This strategy was not applicable to 9 of the control subjects,because they did not perform the speech motor tasks. Even in the subjectsperforming the speech motor task, not every physiological area was readilyidentified in every subject. The most reliable and least restrictive strategy wasemployed.

analyses at the ROI level was extremely high due to a largenumber of degrees of freedom and a very low variance.

An alternative to the use of ROI-based analyses is avoxel-by-voxel between-groups contrast. Whereas thisstrategy has been applied to explore the neural dysfunctionassociated with psychiatric disorders (Andreasen et al.,1994; Bench et al., 1992; Drevits & Raichle, 1995), it isextremely sensitive to any imprecision in spatial normaliza-tion and registration. Further, this method has not yet beenformally described or validated. Nevertheless, it may wellprove possible to apply such an analysis to the present dataat a later date.

Prior Functional Imaging Studies

Only one prior neuroimaging study can be considered atrue functional-lesion study, that of Pool et al. (1991). How-ever, Wu et al. (1995) also attempted to use their findings toaddress the issue of fundamental brain abnormalities inpeople who stutter.

Pool et al. (1991). As described previously, this was aSPECT study that imaged, during rest, three separate "slic-es" of the brain (one of which is reported) for a group ofpeople who stuttered and a larger group of nonstutteringcontrol subjects. In producing findings of significant be-tween-group differences in certain areas, their findings uni-formly contradict those of the present study. Therefore, acomparison of the two studies on various dimensions seemsin order. Beginning with subject characteristics, Pool et al.'scohort of persons who stutter was larger (n = 20), but lessselective, than that of the present study. Whereas thatcohort was mainly right-handed men with developmentalstuttering, one woman, two left-handers, and two personswith "postpuberty onset" were included. Sex and handed-ness are strongly associated with cerebral asymmetries(Geschwind & Galaburda, 1984; Gur et al., 1995). Acquiredstuttering is strongly associated with known brain lesions(Rosenbek, 1984). Thus, up to 25% of their experimentalsubjects had confounding conditions likely to yield atypicalpatterns of brain blood flow, relative to the remainder of theexperimental group.

The control subjects (n = 78) used by Pool et al. were notacquired concurrently with the experimental group, beingdrawn from a larger pool of subjects used in a studypublished 5 years earlier (Devous, Stokely, Chehabi, &Bonte, 1986). Fox, Lancaster, and Ingham (1993) haveargued that this delay was technically problematic and, atleast in part, could account for the reported differences.

The regional analysis of Pool et al. was restricted to anasymmetry index. The expected Hemisphere effect (right >left) (Perlmutter et al., 1985, 1987), which was confirmed inthe present study, was not addressed. Rather, abnormalasymmetries in 3 (of 11) regions were reported: anteriorcingulate gyrus, superior temporal gyrus, and middle tem-poral gyrus. Although these regions were sampled in thepresent study, they produced no between-group differences(see Table 4: AG, S/STG, MTG). In fact, whereas Pool et al.reported a consistent left < right laterality in these regionsamong persons who stuttered, in the present study the(insignificant) asymmetries were in the opposite direction in


two of the three regions. Thus, the present study confirmednone of the observations of Pool et al.

This lack of confirmation raises the issue of methodolog-ical differences between the two studies. Pool et al. mea-sured brain blood flow in a single, low-resolution (1.7-1.9cm) plane using 33Xe SPECT. This single plane was ori-ented "by eye," without the aid of ancillary stereotactic orstructural imaging (e.g., CT or MRI). The ROls applied werebased on a standard template rather than being optimizedphysiologically or anatomically to target brain areas involvedin speech and hearing. Further, the ROls were large andcontiguous, which made no allowances for the partial vol-ume averaging known to be problematic in such low-resolution images. Because no attenuation correction wasapplied, all measurements, particularly those deep to thesurface of the brain, could be considered quantitativelyquestionable.5

Wu et al. (1995). This is a report of an 8FDG PET studythat the authors believe addresses the question of whetherpeople who stutter display a "nonreversible trait difference"(p. 501). Their conclusion, that stutterers indeed show a"permanent left caudate hypometabolism which is a possi-ble trait marker for stuttering . . ." (p. 504), was not repli-cated in the present study. They studied 8 right-handedadults: 4 persons with developmental stuttering and 4 nor-mal controls. Both subject groups contained 3 men and 1woman, roughly approximating the sex ratio found in thepopulation of persons who stutter. The scan conditions werereading aloud and choral reading, which were directly com-pared, without a rest condition to establish whether func-tional lesions exist or to isolate the task-associated activa-tions specific to each subject group. Such "hybrid" designscannot clearly distinguish trait (i.e., the physiological terrainof the brain at rest) from state (i.e., the activation patternsspecific to the required task). Wu et al. draw their conclusionregarding the caudate from the observation that caudatemetabolism of the subjects who stutter did not normalizeduring the (supposedly fluent) choral reading condition. Nodata regarding stuttering frequency during the experimentare reported.

Wu et al. spatially normalized images relative to Talairachand Tournoux (1988), as we advocate. Co-registered MRls,however, were not employed. Further, direct between-groupcomparisons of activations were made voxel-by-voxel, aprocedure for which no validations have been published.This type of analysis is inherently susceptible to falselypositive "abnormalities," especially at boundaries betweenareas of high and low metabolism (or blood flow). Thus, theirfinding of an abnormality restricted to the caudate nucleusmust be viewed with caution. The caudate, a grey-matterstructure with a high metabolic rate, abuts the lateral ven-tricle, a cerebrospinal fluid space with a metabolic rate ofzero. Further, the lateral ventricles vary greatly in size fromindividual to individual, are differentially enlarged by normal

5Many of the preceding methodological criticisms apply to two subsequentreports based on the same subjects and data acquisition procedures(Watson, Pool, Devous, Freeman, & Finitzo, 1992; Watson et al., 1994).Hence, such derivative analyses should be regarded with caution.

aging (Barron, Jacobs, & Kinkel, 1976; Glydensted, 1972;Talairach et al., 1967). Variations in the size of the lateralventricle, in turn, directly affect the position of the caudate,within the normalized space (Talairach et al., 1967). Mis-alignment of even a single subject in either group would besufficient to account for the decreased caudate metabolismobserved. In the study reported herein, confirmation that thecaudate ROI was well within the caudate nucleus and wellremoved from the ventricle was obtained by within-subject,co-registered MRI in every instance. The caudate nucleishowed no abnormalities of regional blood flow or regionalasymmetry. Additionally, in a functional-activation study onthe same subjects used in the present study (Fox et al.,1996) caudate activation was not observed. Thus, the find-ing of Wu et al. must await replication.

ConclusionsThe present study, then, is an essentially negative func-

tional-lesion report on developmental stuttering. In the ROIanalysis, the only significant findings were between-regionand between-hemisphere differences common both to per-sons who do and do not stutter. These variations are bestunderstood as the brain's normal terrain. In the LateralityIndex analysis, between-group differences were relativelyminor variations in the expected regional asymmetries.These were not consistent in sense or direction, and werenot associated with regional between-group differences.Although the methods of the present study were not beyondcriticism, they were stronger-often greatly so-than priorneuroimaging studies of developmental stuttering. Giventhese findings, it appears that promising directions for futurework are: (a) functional activation studies contrasting stut-tering to (induced) fluency, and (b) a functional-lesion studyrestricted to severe developmental stuttering, this being thepopulation in which focal lesions are most likely to be found,if they exist.

AcknowledgmentsSenior authorship is shared between the first two authors. The

authors thank several colleagues whose contributions were funda-mental to the success of this study: Barbara Alkek and Karen Myersfor recruiting and screening the subjects who stuttered; Betty Heyl,Ralph Evans, and Traci Hirsch for assistance in image acquisition;Jack Lancaster, Thomas Glass, and J. Hunter Downs for developingthe software used for image analysis; Mario Liotti and Peter Frankfor expert assistance with statistical analyses of the data. This studywas supported in part by research grant number 5 R01 DC 00060from the National Institute on Deafness and Other CommunicationDisorders, National Institutes of Health, awarded to the first author.

ReferencesAndreason, N. C., Arnmdt, S., Swayze, V., II, Cizadlo, T., Flaum, M.,

O'Leary, D., Ehrhardt, J. C., & Yuh, W. T. C. (1994). Thalamicabnormalities in schizophrenia visualized through magnetic reso-nance image averaging. Science, 266, 294-298.

Barron, S. A., Jacobs, L., & Kinkel, W. R. (1976). Changes in size

39 1208-1227 December 1996


of normal lateral ventricles during aging determined with comput-erized tomography. Neurology, 26, 1011-1013.

Baxter, L. R., Schwartz, J. M., & Phelps, M. (1989). Reduction ofprefrontal cortex glucose metabolism common to three types ofdepression. Archives of General Psychiatry, 46, 243-250.

Bench, C. J., Friston, K. J., Brown, R. G., Scott, L., Frackowiak,R. S. J., & Dolan, R. J. (1992). The anatomy of melancholia: Focalabnormalities of cerebral blood flow in major depression. Psycho-logical Medicine, 22, 607-615.

Blood, . M., & Blood, G. W. (1984). Relationship between stutteringseverity and brainstem-evoked response testing. Perceptual andMotor Skills, 59, 935-938.

Bloodstein, 0. (1995). A handbook on stuttering. San Diego, CA:Singular Publishing Company.

Boberg, E. (Ed.) (1993). The neuropsychology of stuttering. Ed-monton, Alberta, Canada: University of Alberta Press.

Braun, A. R. (1996). Personal communication, June 6.Braun, A. R., Ludlow, C. L., Varga, M., Shulz, G. M., & Stager, S.

(1994). Central correlates of speech motor control in stuttering:An H2

150 PET study. Neurology, 44 (Suppl 2), A261.Braun, A. R., Varga, M., Stager, S., Schulz, G., Selbie, S., Maisog,

J. M., Carson, R. E., & Ludlow, C. L. (1995, December). Alteredpatterns of cerebral activity during speech and language produc-tion in developmental stuttering. An H2 '5 0 PET study. Paper readat the annual meeting of the American Speech-Language-HearingAssociation, Orlando, FL.

Caruso, A. J. (1991). Neuromotor processes underlying stuttering.In H. F. M. Peters, W. Hulstijn, & C. W. Starkweather (Eds.),Speech motor control and stuttering (pp. 101-116). Amsterdam:Excerpta Medica.

Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation:Design and analysis issues for field settings. Chicago: RandMcNally.

Cooper, J. A. (1990). Research needs in stuttering: Roadblocks andfuture directions. ASHA Reports, 18.

Decker, T. N., Healey, E. C., & Howe, S. W. (1982). Brainstemauditory electrical response characteristics of stutterers andnonstutterers: A preliminary report. Journal of Fluency Disorders,7, 385-401.

De Nil, L. F., Kroll, R. M., Kapur, S., & Houle, S. (1995, December).Silent and oral reading in stuttering and nonstuttering adults: Apositron emission tomography study. Paper read at the annualmeeting of the American Speech-Language-Hearing Association,Orlando, FL.

Devous, M. D., Sr., Stokely, E. M., Chehabi, H. H., & Bonte, F. J.(1986). Normal distribution of regional cerebral blood flow mea-sured by dynamic single-photon emission tomography. Journal ofCerebral Blood Flow and Metabolism, 6, 95-104.

Drevits, W. C., & Raichle, M. E. (1995). Positron emission tomo-graphic imaging studies of human emotional disorders. In M. S.Gazzaniga (Ed.), The cognitive neurosciences (pp. 1153-1164).Cambridge, MA: MIT Press.

Duara, R., Lowenstein, D. A., & Barker, W. W. (1990). Utilization ofbehavioral activation paradigms for positron emission tomogra-phy studies in normal young, and elderly subjects and in demen-tia. In R. Duara (Ed.), Positron emission tomography in dementia(pp. 131-148). New York: Wiley-Liss.

Evans, A. C., Collins, D. L., & Holmes, C. J. (1996). Computationalapproaches to quantifying human neuroanatomical variability. InA. W. Toga & J. C. Mazziotta (Eds.), Brain mapping: The methods(pp. 343-361). San Diego: Academic Press.

Finitzo, T., Pool, K. D., Freeman, F. J., Devous, M. D., Sr., &Watson, B. C. (1991). Cortical dysfunction in developmentalstutterers. In H. F. M. Peters, W. Hulstijn, & C. W. Starkweather(Eds.), Speech motor control and stuttering (pp. 251-261). Am-sterdam: Excerpta Medica.

Ford, . (1995). Some nonontological and functionally unconnectedviews on current issues in the analysis of PET datasets. Journal ofCerebral Blood Flow and Metabolism, 15, 371-377.

Fox, P. T. (1995). Broca's Area-motor encoding in somatic space.Behavioral and Brain Sciences, 18, 344-345.

Fox, P. T., Burton, H., & Raichle, M. D. (1987). Mapping human

somatosensory cortex with positron emission tomography. Jour-nal of Neurosurgery, 67, 34-43.

Fox, P. T., Ingham, R. J., Ingham, J. C., Hirsch, T., Downs, H.,Martin, C., Jerabek, P., Glass, T., & Lancaster, J. L. (1996). APET study of the neural systems of stuttering. Nature, 382,158-162.

Fox, P. T., Lancaster, J. L., & Ingham, R. J. (1993). On stutteringand global ischemia. Archives of Neurology, 50, 1287-1288.

Fox, P. T., & Pardo, J. V. (1991). Does inter-subject variability incortical functional organization increase with neural "distance"from the periphery? In Ciba Foundation Symposium 163 Explor-ing brain functional anatomy with positron tomography (pp. 125-144). Chichester, England: John Wiley.

Fox, P. T., Perlmutter, J. S., & Raichle, M. E. (1985). A stereotacticmethod of anatomical localization for positron emission tomog-raphy. Journal of Computer Assisted Tomography, 9, 141-153.

Fox, P. T., & Raichle, M. E. (1984). Stimulus rate dependence ofregional cerebral blood flow in human striate cortex, demon-strated by positron emission tomography. Journal of Neurophys-iology, 51, 1109-1120.

Friston, K. J. (1995). Statistical parametric mapping: Ontology andcurrent issues. Journal of Cerebral Blood Flow and Metabolism,15, 361-370.

Gagnon, J., Roth, J., Finzer, W. F., Hofmann, R., Haycock, K. A.,Feldman, D. S., & Simpson, J. (1989). SuperANOVA [software].Berkeley, CA: Abacus Concepts.

Gaillard, W. D., Bhatia, S., Bookheimer, S. F., Fazilat, S., Sato, S.,& Theodore, W. H. (1995). FDG-PET and volumetric MRI in theevaluation of patients with partial epilepsy. Neurology, 45, 123-126.

Geschwind, N., & Galaburda, A. M. (1984). Cerebral dominance:The biological foundations. Cambridge, MA: Harvard UniversityPress.

Glydensted, C. (1972). Measurements of the normal ventricularsystem and hemispheric sulci of 100 adults with computedtomography. Neuroradiology, 14, 183-192.

Grafton, S. T., Sutton, J., Couldwell, W., Lew, M., & Waters, C.(1994). Network analysis of motor system connectivity in Parkin-son's disease: Modulation of thalamocortical interactions afterpallidotomy. Human Brain Mapping, 2, 45-55.

Gray, J. (1995). A model of the limbic system and basal ganglia. InM. S. Gazzaniga (Ed.), The cognitive neurosciences (pp. 1165-1176). Cambridge, MA: The MIT Press.

Gray, J. A., Feldon, J., Rawlins, J. N. P., Hemsley, D. R., & Smith,A. D. (1991). The neuropsychology of schizophrenia. Behavioraland Brain Sciences, 14, 1-20.

Gur, R. C., Mozley, L. H., Mozley, P. D., Resnick, S. M., Karp,J. S., Alavi, A., Arnold, S. E., & Gur, R. E. (1995). Sex differencesin regional cerebral glucose metabolism during a resting state.Science, 267, 528-531.

Harris, A. J. (1958). Harris test of lateral dominance: Manual fordirections for administration and interpretation. San Antonio,TX:Psychological Corporation.

Herscovitch, P., Markham, J., & Raichle, M. E. (1983). Brain bloodflow measured with intravenous H2

150. I. Theory and erroranalysis. Journal of Nuclear Medicine, 24, 782-789.

Hoffman, E. J., Huang, S. C., & Phelps, M. E. (1979). Quantitationin positron emission tomography: 1. Effect of object size. Journalof Computer Assisted Tomography, 3, 299-308.

Huang, S. C., Hoffman, El J., Phelps, M. E., & Kuhl, D. E. (1980).Quantitation in positron emission computed tomography: 3. Effectof sampling. Journal of Computer Assisted Tomography, 4, 819-826.

Huynh, H., & Feldt, L. S. (1970). Conditions under which meansquares ratios in repeated measures designs have exact F-distributions. Journal of the American Statistical Association, 65,1582-1589.

Ingham, R. J., Cordes, A. K., & Finn, P. (1993). Time-intervalmeasurement of stuttering: Systematic replication of Ingham,Cordes, and Gow (1993). Journal of Speech and Hearing Re-search, 36, 1168-1176.

Ingham, R. J., Fox, P. T., & Ingham, J. C. (1994, November). Brainimage investigation of the speech of stutterers and nonstutterers.


Paper read at the American Speech-Language-Hearing Associa-tion Annual Convention, New Orleans, LA.

Ingham, R. J., Fox, P. T., & Ingham, J. C. (1995, December). Areport on a functional-activation and functional-lesion PET studyof stuttering in adults. Paper read at the American Speech-Language-Hearing Association Annual Convention, Orlando, FL.

Lancaster, J. L., Glass, T. G., Lankipalli, B. R., Downs, H.,Mayberg, H., & Fox, P. T. (1995). A modality-independentapproach to spatial normalization of tomographic images of thehuman brain. Human Brain Mapping, 3, 209-223.

Liotti, M., Gay, C. T., & Fox, P. T. (1994). Functional imaging andlanguage: Evidence from positron emission tomography. Journalof Clinical Neurophysiology, 11, 175-190.

Malonek, D., & Grinvald, A. (1996). Interactions between electricalactivity and cortical microcirculation revealed by imaging spec-troscopy: Implications for functional brain mapping. Science, 272,551-554.

Mayberg, H. S. (1994). Frontal lobe dysfunction in secondarydepression. Journal of Neuropsychiatry and Clinical Neuro-sciences, 6, 428-442.

Mazziotta, J. C., Frackowiak, R. S. J., & Phelps, M. E. (1992). Theuse of positron emission tomography in the clinical assessment ofdementia. Seminar in Nuclear Medicine, 22, 233-246.

Mazziotta, J. C., Phelps, M. E., Huang, S. C., & Kuhl, D. E. (1982).Quantitation in positron emission tomography: 5. Physical-ana-tomical effects. Journal of Computer Assisted Tomography, 5,734-743.

Moore, W. H., Jr. (1990). Pathophysiology of stuttering: Cerebralactivation differences in stutterers vs. nonstutterers. ASHA Re-ports, 18, 72-80.

Moore, W. H., Jr. (1993). Hemisphere processing research. In E.Boberg (Ed.), Neuropsychology of stuttering (pp. 39-72). Edmon-ton, Alberta: The University of Alberta Press.

Neville, H. J. (1990). Intermodal competition and compensation indevelopment. Evidence from studies of the visual system incongenitally deaf adults. Annals of the New York Academy ofSciences, 608, 71-87.

O'Brien, R. G., & Muller, K. E. (1993). Unified power analysis fort-tests through multivariate hypotheses. In L. K. Edwards (Ed.),Applied analysis of variance in the behavioral sciences (pp.297-344). New York: Marcel Decker.

Orton, S. T. (1928). A physiological theory of reading disability andstuttering in children. New England Journal of Medicine, 198,1045-1052.

Parsons, L. M., Fox, P. T., Downs, J. H., Glass, T., Hirsch, T.,Martin, C., Jerabek, P., & Lancaster, J. L. (1995). Use of implicitmotor imagery for visual shape discrimination as revealed by PET.Nature, 375, 54-59.

Perlmutter, J. S., Herscovitch, P., Powers, W. J., Fox, P. T., &Raichle, M. E. (1985). A standardized mean regional method forcalculating global positron emission tomographic measurements.Journal of Cerebral Blood Flow and Metabolism, 5, 476-480.

Perlmutter, J. S., Powers, W. Herscovitch, P., Fox, P. T., &Raichle, M. E. (1987). Regional asymmetries of cerebral bloodflow, blood volume, oxygen utilization and extraction in normalsubjects. Journal of Cerebral Blood Flow and Metabolism, 7,644-667.

Peters, H. F. M., Hulstijn, W., & Starkweather, C. W. (1991).Speech motor control and stuttering. Amsterdam: Exerpta Med-ica.

Pool, K. D., Devous, M. D., Sr., Freeman, F. J., Watson, B. C., &Finitzo, T. (1991). Regional cerebral blood flow in developmentalstutterers. Archives of Neurology, 48, 509-512.

Pool, K. D., Finitzo, T., Devous, M. D., Sr., Watson, B. C., &Freeman, F. J. (1993). In reply. Archives of Neurology, 50,1289-1290.

Raichie, M. E., Martin, W., & Herscovitch, P. (1983). Brain bloodflow measured with intravenous H2

15 0. II. Implementation andvalidation. Journal of Nuclear Medicine, 24, 790-798.

Roland, P. E. (1993). Brain activation. New York: Wiley-Liss.Rosenbek, J. C. (1984). Stuttering secondary to nervous system

damage. In R. F. Curlee & W. H. Perkins (Eds.), Nature and

treatment of stuttering: New directions (pp. 31-48). San Diego,CA: College-Hill Press.

Rosenfield, D. B., & Nudelman, H. B. (1987). Neuropsychologicalmodels of speech dysfluency. In L. Rustin, H. Purser, & D. Rowley(Eds.), Progress in the treatment of fluency disorders (pp. 3-42).London: Taylor & Francis.

Sidman, M. (1960). Tactics of scientific research. New York: BasicBooks.

Smith, A. (1990). Toward a comprehensive theory of stuttering: Acommentary. Journal of Speech and Hearing Disorders, 55,398-401.

Strick, P. L., Dum, R. P., & Picard, N. (1995). Macro-organizationof the circuits connecting the basal ganglia with the cortical motorareas. In J. Houk, J. Davis, & D. Beiser (Eds.), Models ofinformation processing the basal ganglia (pp. 117-130). Cam-bridge, MA: The MIT Press.

Strother, S. C., Kanno, I., & Robbenberg, K. A. (1995). Principalcomponent analysis, variance partitioning, and "functional con-nectivity." Journal of Cerebral Blood Flow and Metabolism, 15,353-360.

Strub, R. L., Black, F. W., & Naeser, M. A. (1987). Anomalousdominance in sibling stutterers: Evidence from CT scan asymme-tries, dichotic listening, neuropsychological testing, and handed-ness. Brain and Language, 30, 338-350.

Talairach, J., Szikla, G., Toumoux, P., Prossalentis, A., Bordas-Ferrer, M., Covello, L., Jacob, M., & Mempel, E. (1967). Atlasd'anatomie stereotaxique du telencephale. Etudes anatamo-ra-diologiques. Paris: Masson.

Talairach, J., & Tournoux, P. (1988). Co-planar stereotaxic atlas ofthe human brain. Verlag: Thieme.

Travis, L. E. (1931). Speech pathology. New York: Appleton-Century-Crofts.

Travis, L. E. (1978). The cerebral dominance theory of stuttering:1931-1978. Journal of Speech and Hearing Disorders, 43, 278-281.

Watson, B. C., Freeman, F. J., Devous, M. D., Sr., Chapman,S. B., Finitzo, T., & Pool, K. D. (1994). Linguistic performanceand regional cerebral blood flow in persons who stutter. Journalof Speech and Hearing Research, 37, 1221-1228.

Watson, B. C., Pool, K. D., Devous, M. D., Sr., Freeman, F. J., &Finitzo, T. (1992). Brain blood flow related to acoustic laryngealreaction time in adult developmental stutterers. Journal of Speechand Hearing Research, 35, 555-561.

Webster, W. G. (1993). Hurried hands and tangled tongues. Impli-cations of current research for the management of stuttering. In E.Boberg (Ed.), Neuropsychology of stuttering (pp. 73-127). Ed-monton, Alberta, Canada: The University of Alberta Press.

Weinberger, D. R., Aloia, M. S., Goldberg, T. E., & Berman, K. F.(1994). The frontal lobes and schizophrenia. Journal of Neuropsy-chiatry and Clinical Neurosciences, 6, 419-427.

Weinstein, J. M., Caruso, A. J., Severing, K., & VerHoeve, J.(1989). Abnormalities of oculomotor control in stutterers. Investi-gative Ophthalmology and Visual Science, 30 [Abstract].

Wells, B. G., & Moore, W. H., Jr. (1990). EEG alpha asymmetries instutterers and nonstutterers: Effects of linguistic variables onhemispheric processing and fluency. Neuropsychologia, 28,1295-1305.

Winer, B. J., Brown, D. R., & Michels, K. M. (1991). Statisticalprinciples in experimental design (3rd ed.). New York: McGraw-Hill.

Wood, F., Stump, D., McKeehan, A., Sheldon, S., & Proctor, J.(1980). Patterns of regional cerebral blood flow during attemptedreading aloud by stutterers both on and off haloperidol medica-tion: Evidence for inadequate left frontal activation during stutter-ing. Brain and Language, 9, 141-144.

Woods, R. P. (1996). Correlation of brain structure and function. InA. W. Toga & J. C. Mazziotta (Eds.), Brain mapping: The methods(pp. 313-341). San Diego, CA: Academic Press.

Wu, J. C., Maguire, G., & Riley, G. (1995, December). A PET studyof stuttering using FDG. Paper read at the American Speech-Language-Hearing Association Annual Convention, Orlando, FL.

Wu, J. C., Maguire, G., Riley, G., Fallon, J., LaCase, L, Chin, S.,Klein, E., Tang, C., Cadwell, S., & Lottenberg, S. (1994, No-

39 1208-1227 December 996


vember). A positron emission tomography study of developmentalstuttering. Paper read at the American Speech-Language-Hear-ing Association Annual Convention, New Orleans, LA.

Wu, J. C., Maguire, G., Riley, G., Fallon, J., LaCase, L., Chin, S.,Klein, E., Tang, C., Cadwell, S., & Lottenberg, S. (1995). Apositron emission tomography (F-18) deoxyglucose study ofdevelopmental stuttering. NeuroReport, 6, 501-505.

Yeudall, L. T., Manz, L., Ridenour, C., Tani, A., Lind, J., & Fedora,0. (1993). Variability in the central nervous system of stutterers. InE. Boberg (Ed.), Neuropsychology of stuttering (pp. 129-163).Edmonton, Alberta, Canada: The University of Alberta Press.

Zimmermann, G. (1980). Stuttering: A disorder of movement.Journal of Speech and Hearing Research, 23, 122-136.

Received February 15, 1996Accepted July 31, 1996

Contact author: Peter T. Fox, MD, Research Imaging Center,University of Texas Health Science Center, 7703 Floyd Curl Drive,San Antonio, TX 78284-6240. Email: [email protected]

Appendix

Definitions of Talairach and Tournoux (1988) BrainAtlas Abbreviations (See Tables 1, 4, and 5)

BA Lobe Gyrus Region

6 Frontal GFd: Gyrus frontalis medialis SMAi: Inferior supplementary motor area6 Frontal GFd: Gyrus frontalis medialis SMAs: Superior supplementary motor area6 Frontal GPrC: Gyrus precentralis SLPrM: Superior lateral premotor area4 Frontal GPrC: Gyrus precentralis Ml: Primary motor area

44 Frontal GPrC: Gyrus precentralis ILPrM: Inferior lateral premotor area41 Temporal GTs: Gyrus temporalis superior Al: Primary auditory area22 Temporal GTs: Gyrus temporalis superior A2: Associated auditory areana Insula INS: Insula INS: Insulana Quadrangular na CBM: Cerebellum10 Frontal GFs: Gyrus frontalis superior SFG: Superior frontal gyrus32 Limbic ant GC: Gyrus cinguli AG: Anterior cingulate gyrus10 Frontal GFm: Gyrus frontalis medius MFG: Medial frontal gyrus46 Frontal GFi: Gyrus frontalis inferior IFG: Inferior frontal gyrus

6 Frontal GPrC: Gyrus precentralis P/PG: Precentral/Postcentral gyrus22 Temporal GTs: Gyrus temporalis superior S/STG: Supramarginal/Superior temporal21 Temporal GTm: Gyrus temporalis medius MTG: Medial temporal gyrus37 Temporal GTi: Gyrus temporalis inferior ITG: Inferior temporal gyrus18 Occipital Cu: Cuneus O: Occipital area23 Occipital Pcu: Precuneus C/P: Cingulate/Precuneus areana Sub-lobar na Thalamusna Sub-lobar na Caudatena Sub-lobar na Putamenna Sub-lobar na GP: Globus pallidus8 Frontal GFs: Gyrus frontalis superior GFs: Superior frontal gyrus9 Frontal GFs: Gyrus frontalis superior GFs: Superior frontal gyrus

1/2/3 Parietal GPoC: Gyrus postcentralis GPoC: Postcentral gyrus4 Frontal GPrC: Gyrus precentralis GPrC: Precentral gyrus

40 Parietal LPi: Lobulus parietalis inferior LPi: Inferior parietal lobe17 Occipital Cu/GL: Cuneus/Gyrus lingualis Cu/GL: Cuneus/Lingual gyrus25 Limbic Gs: Gyrus subcallosus Gs: Subcallosal gyrus35 Temporal Gh: Gyrus parahippocampi Gh: Hippocampal gyrus30 Limbic GC: Gyrus cinguli GC: Cingulate gyrus38 Temporal GTs: Gyrus temporalis superior GTs: Superior temporal syrus39 Temporal GTm: Gyrus temporalis medius GTm: Medial temporal gyrus43 Parietal GPoC: Gyrus postcentralis GPoC: Postcentral gyrus44 Frontal GFi: Gyrus frontalis inferior GFi: Inferior frontal gyrus47 Frontal GFi: Gyrus frontalis inferior GFi: Inferior frontal gyrus

Documents

Functional.Lesion Investigation of Developmental ...ric.uthscsa.edu/personalpages/lancaster/SPM_Class/Lecture_20/Fox... · Science Center San Antonio Janis C ... the dominant general