Plasticity of Language-Related Brain Function During
Recovery From Stroke
Keith R. Thulborn, MD, PhD; Patricia A. Carpenter, PhD; Marcel A. Just, PhD
Background and Purpose—This study was undertaken to correlate functional recovery from aphasia after acute stroke
with the temporal evolution of the anatomic, physiological, and functional changes as measured by MRI.
Methods—Blood oxygenation level– dependent contrast and echo-planar MRI were used to map language comprehension
in 6 normal adults and in 2 adult patients during recovery from acute stroke presenting with aphasia. Perfusion,
diffusion, sodium, and conventional anatomic MRI were used to follow physiological and structural changes.
Results—The normal activation pattern for language comprehension showed activation predominately in left-sided
Wernicke’s and Broca’s areas, with laterality ratios of 0.8 and 0.3, respectively. Recovery of the patient confirmed as
having a completed stroke affecting Broca’s area occurred rapidly with a shift of activation to the homologous region
in the right hemisphere within 3 days, with continued rightward lateralization over 6 months. In the second patient, in
whom mapping was performed fortuitously before stroke, recovery of a Wernicke’s aphasia showed a similar increasing
rightward shift in activation recruitment over 9 months after the event.
Conclusions—Recovery of aphasia in adults can occur rapidly and is concomitant with an activation pattern that changes
from left to a homologous right hemispheric pattern. Such recovery occurs even when the stroke evolves to completion.
Such plasticity must be considered when evaluating stroke interventions based on behavioral and neurological
measurements. (Stroke. 1999;30:749-754.)
Key Words: magnetic resonance imaging n aphasia n cerebral infarction n stroke outcome

Downloaded from http://ahajournals.org by on April 11, 2024

T

Functional MRI (fMRI), a noninvasive method of measuring human brain function during cognition,6 –9 has shown that
a network of cortical areas contributes to language processing.10,11 For example, the difficulty of language comprehension during reading has been demonstrated to correlate with
the degree of cortical activation.10 This association between
task difficulty and fMRI activation has been generalized to
other tasks including visuospatial processing.12 This reading
comprehension also includes brain activation associated with
eye movement.13 Such cognitive paradigms are now being
used to map eloquent cortical areas before surgical treatment
of epilepsy.14 Such methodology allows the network of brain
areas to be characterized not simply by network locations but
also in terms of workload distribution between each location.
A modified10 simple sentence–reading paradigm for fMRI
has been standardized in a group of normal subjects. Two
separate clinical cases have been studied with fMRI to show
the changes in activation patterns during the first months of
the recovery period from stroke involving either Broca’s area
or Wernicke’s area. The first patient with acute onset of
expressive aphasia from left middle cerebral artery (MCA)
stroke showed rapid recovery over the first few days. Although normal controls are useful, the question of language

he development of language in right-handed and most
left-handed subjects involves many areas of the brain,
including the superior temporal (Wernicke’s area) and inferior frontal (Broca’s area) gyri of the left hemisphere.1 This
view derives from many sources, including lesion studies and
Wada tests.1 Although knowledge of language processing
during development remains rudimentary, the left hemispheric regions are believed to dominate language comprehension early in the process. If the left hemisphere becomes
dysfunctional during development, neuroplasticity allows
homologous regions in the right hemisphere to perform the
same processes.2 Recent neuroimaging studies with positron
emission tomography (PET) of adult patients recovered from
aphasia after left-hemisphere stroke have demonstrated rightsided activation for language processing.3 Similar results
were reported for adult patients recovered from auditory
agnosia.4 Functional recovery appears to be connected to
activation of homologous regions of the contralateral hemisphere. Although some ipsilateral activation may also aid
recovery, the possible bilateral representation before stroke
has been raised as a possible explanation.5 Such previous
studies were completed well after recovery from the clinical
symptoms, leaving the temporal course of such changes
uncharted.

Received October 21, 1998; final revision received January 27, 1999; accepted January 27, 1999.
From the MR Research Center (K.R.T.), University of Pittsburgh Medical Center, and the Department of Psychology (P.A.C., M.A.J.), Carnegie Mellon
University, Pittsburgh, Pa.
Correspondence to Keith R. Thulborn, MD, PhD, B855, MR Research Center, Presbyterian University Hospital, 200 Lothrop St, Pittsburgh PA 15213.
E-mail keith@mrctr.upmc.edu
© 1999 American Heart Association, Inc.
Stroke is available at http://www.strokeaha.org

749

750

fMRI of Aphasia Recovery After Stroke

representation in a patient before stroke requires a study of
the rare patient who has been mapped before stroke. The
second patient was examined before and after stroke suffered
during surgery for left temporal lobe epilepsy. The implications of these neuroimaging results are discussed in terms of
recovered brain function and plasticity of large-scale
networks.5,15,16

Subjects and Methods
Informed consent followed the guidelines of the Institutional Review
Board.

Control Group
Six healthy right-handed male college graduates (aged 26 to 31
years) served as control subjects .

Case 1

Downloaded from http://ahajournals.org by on April 11, 2024

The patient was a 45-year-old right-handed man, on chronic anticoagulation medication for aortic valve replacement 7 years previously.
He was a high school graduate with additional technical school
training. He presented to medical attention with abrupt onset of a
dense expressive aphasia and right-sided weakness. A left MCA
stroke was suspected. Emergency CT showed no cerebral hemorrhage. Intravenous tissue plasminogen activator (tPA) was initiated
within 3 hours of onset of symptoms. An MRI study was performed
at 5 hours after onset of symptoms and more than 2 hours after
commencement of thrombolytic therapy. Diffusion imaging confirmed the left MCA stroke, and perfusion imaging demonstrated
persistent arterial occlusion. The patient recovered considerable
language function over the next 3 days. Another comprehensive
fMRI examination was performed at 76 hours and again at 6 months,
by which time minor word-finding difficulties and a slight balance
impairment were the only symptoms.

Case 2
The patient was a 34-year-old right-handed male college graduate
with chronic epilepsy since sustaining a left-sided closed head injury
at age 11 years. He had been poorly maintained on antiepileptic
medication because of adverse effects. A CT examination demonstrated focal cortical calcification in the mid left superior temporal
gyrus. When surgical intervention was contemplated to control his
seizures to avoid the use of medications, fMRI using the language
paradigm was performed before surgical placement of grid electrodes for cortical recording. Agreement between fMRI and cortical
grid mapping was excellent, with Wernicke’s area being placed
about 2 centimeters posterior to the well-defined localized lesion in
the left temporal lobe. The patient proceeded to surgery with limited
resection of the small focal region of calcifications but not extending
posteriorly into the area of language function documented by cortical
electrodes. He awoke with a dense receptive aphasia. Functional
MRI studies at 3 and 9 months, during which time language
comprehension skills improved, were compared with the preoperative study. The pathology of the resected tissue diagnosed a ganglioglioma. The patient returned to work but still acknowledges having
language difficulties since his surgery.

Imaging Protocol
All fMRI imaging was performed with a commercial quadrature
head coil on either a 1.5- or 3.0-T whole-body scanner equipped with
echo-planar imaging and large data-handling capabilities.8,9,17
The comprehensive MRI protocol for patient studies, involving
anatomic, diffusion, perfusion, and sodium imaging, has been
reported elsewhere.18
Acquisition parameters for fMRI using gradient-echo echo-planar
imaging at 3.0 T were as follows: TR54000 ms; TE525 ms; field of
view (FOV), 40320; voxel size, 3.133.133.0 mm3; slice thickness/
gap, 3/1 mm; and n518 slices. The 1.5-T parameters differed only in
that TE was 50 ms and slice thickness was 5 mm.

The language comprehension paradigm8,10 consisted of 5 cycles
of 2 conditions: (1) central fixation (rest condition, 30 s); and
(2) silently reading simple sentences (mean length, 5.5 words) each
followed by a question requiring a “true” or “false” answer indicated
by pushing 1 of 2 finger switches (sentence condition, 30 s). The
accuracy of responses (at least 79% in each session) to the comprehension questions assured that the patients were performing the
required task. An additional condition, silent reading of word strings,
is not reported because it produced similar results to the sentencereading condition but provided no assurance of comprehension
processing. This paradigm was chosen because it activates both
Broca’s and Wernicke’s areas and can be used to detect changes in
both expressive and receptive aphasias.
The fMRI data were analyzed using customized software,13,19
resulting in a 2-tailed paired t test comparison of the rest and active
conditions of the paradigm for each voxel. The t test is a conservative
statistical approach to analysis of fMRI data and emphasizes the
most robust responses within the data. The volume of each activated
region for each subject was measured as the number of voxels
exceeding the chosen threshold (t.3.0). This threshold minimized
the number of randomly activated voxels while maximizing the
number of reproducibly activated voxels for the subjects of the
normal group. Higher thresholds did not change the patterns of
activations, only decreasing the number of reproducibly activated
voxels. Head motion was assessed by visual inspection of the images
presented in a closed display loop. Three-dimensional display of the
functional data was performed with software called AFNI19 to ensure
that regions were appropriately labeled according to Talairach
coordinates.20 From our previous experience with eye movement13
and language10 paradigms in normal subjects, the areas of activation
are Broca’s area, Wernicke’s area, supplementary eye fields, superior and inferior frontal eye fields, intraparietal sulcus, visual cortex,
and prefrontal cortex. By counting the number of voxels in each
region of interest, volumes of activation were determined. These
volumes allowed hemispheric asymmetries to be expressed as the
laterality ratio (LR) for each area: LR5(left2right)/(left1right),
where LR equal to 1, 0, or 21 shows that activation is only in the left
hemisphere, symmetrically distributed, or restricted to the right
hemisphere, respectively.

Results
Normal Reading Comprehension
Figure 1 shows a representative activation map for the control
subjects that identifies 8 regions bilaterally. This activation
pattern represents a large-scale network15,16 that includes the
2 main language areas, Wernicke’s and Broca’s areas, which
show left-sided asymmetry, irrespective of right- or lefthandedness, as expected for adult language comprehension.1,5
The Table provides the means and standard deviations of
these laterality ratios (LR) for these areas, showing that the
asymmetry is particularly marked for Wernicke’s area (LR,
0.860.3) and less so for Broca’s area (LR, 0.460.4). By
contrast, the activation associated with eye movement (superior frontal eye fields, inferior frontal eye fields, and intraparietal sulcus) showed laterality ratios close to 0, indicating
bilateral symmetry.

Case 1
Structural Analysis
Selected anatomic, physiological, and metabolic images from
the first (5 hours), second (76 hours), and third (6 months)
MRI examinations after the abrupt onset of a dense expressive aphasia are shown in Figure 2. Although the T2weighted image shows little change acutely (Figure 2a), an
extensive cortical region of signal hyperintensity was present

Thulborn et al

April 1999

751

Downloaded from http://ahajournals.org by on April 11, 2024

Figure 1. Representative axial activation map of a normal control for sentence reading comprehension showing the largescale network of multiple active areas. Areas are labeled as 1,
Broca’s area; 2, Wernicke’s area; 3, intraparietal sulcus; 4,
superior frontal eye fields; 5, inferior frontal eye fields; 6, supplementary eye fields; 7, prefrontal cortex; and 8, visual cortex.
Radiological convention (viewing the supine patient from the
foot of the bed) has been used in all images, with the right-hand
side of the subject on the left-hand side of the page.

Figure 2. Images of patient 1 documenting the left-sided stroke
(arrow) at 5 hours (top row), 76 hours (middle row), and 6
months (bottom row) after the event on representative slices of
T2-weighted images (a, d, g), diffusion-weighted images (b, e,
h), and sodium images (c, f, i). The lesion is diagnosed early by
diffusion imaging. Infarction is documented by the elevated
sodium levels by 76 hours.

on diffusion-weighted imaging (Figure 2b) that showed a
reduced apparent diffusion coefficient (ADC [lesion],
0.5860.09310 23 mm 2 /s; ADC [normal gray matter],
0.9760.0931023 mm2/s) from normal gray matter. Acutely,
there was a persistent perfusion defect with prolonged tissue
transit time (TTT [lesion], .20 s; TTT [contralateral], 10 s).
The tissue sodium image showed little change between the
tissue sodium concentration of 45 mmol/L in the lesion and
that on the contralateral side (Figure 2c). The subject was
unable to perform the language paradigm during this examination at 5 hours after the event.
By 76 hours, considerable clinical recovery from the initial
aphasia had occurred. The spin-echo images show markedly
T2-hyperintense cortex along the left Sylvian fissure (Figure
2d) whereas little change occurred to the region of hyperin-

tense signal on the diffusion-weighted imaging (Figure 2e),
except for further reduction in ADC (ADC [lesion],
0.4760.0531023 mm2/s; ADC [contralateral], 0.9760.23
1023 mm2/s). The sodium image (Figure 2f) showed increased
TSC (TSC [lesion], 70 mmol/L; TSC [contralateral],
44 mmol/L) over an extensive area encompassing the abnormal ADC.
Selected images from the third MRI examination at 6
months after stroke, by which time the patient had only a few
word-finding difficulties, are shown in Figures 2g, 2h, and 2i.
The spin-echo images show markedly T2-hyperintense encephalomalacic change along the left Sylvian fissure and left
inferior frontal region (Figure 2g). The region of signal
hyperintensity on the diffusion-weighted imaging (Figure 2h)
showed increased ADC as expected with old stroke (ADC

Laterality Ratios (6SD) for Activation in Wernicke’s Area and Broca’s Area for
Reading Comprehension for Control Subjects (n56) and Patients 1 and 2 at
Different Times From Stroke
Case 1

Case 2

Brain Area

Control
Subjects

76 hours

6 months

Before Stroke

3 mo

9 mo

Wernicke’s area

0.860.3

1.0

0.9

0.8

0.3

20.6

Broca’s area

0.460.4

20.8

21.0

20.2

0

0.2

752

fMRI of Aphasia Recovery After Stroke

Figure 3. Activation maps of patient 1 at (a) 76 hours and (b) 6
months using the same paradigm and activation labels as in
Figure 1.

[lesion], 1.8660.3231023 mm2/s; ADC [contralateral],
0.9860.2531023 mm2/s). The TSC abnormality (TSC [lesion], 74 mmol/L; TSC [contralateral], 49 mmol/L) was
similar to the second examination. Thus, structural changes
for patient 1 can be attributed definitively to a large stroke in
the left MCA territory, as indicated by anatomic, diffusion,
perfusion, and sodium imaging performed acutely and confirmed at follow-up, despite rapid clinical improvement over
3 days.

Downloaded from http://ahajournals.org by on April 11, 2024

fMRI Analysis
The significant clinical recovery of patient 1 permitted fMRI
studies with language to be performed at 76 hours and at 6
months, as shown in Figures 3a and 3b, respectively. The
fMRI results show that this right-handed subject had an
abnormal laterality ratio for Broca’s area (the region affected
by the stroke) by 76 hours, showing strong right dominance,
in contrast to the normal left dominance. By 6 months, this
abnormal cerebral dominance pattern progressed to being
totally right-sided. Wernicke’s area, which was structurally
undamaged, was completely left dominant at 76 hours and
remained strongly left dominant at 6 months. The changes
involving several areas measured at 2 times support the
hypothesis that clinical recovery was associated with rapid
redistribution of the task over an existing large-scale network
to allow rapid initial recovery within days, followed by
consolidation of the new pattern over subsequent months. No
new nodes of activation other than those observed in normal
subjects were identified. Activation in areas related to eye
movement were normal, indicating that not all of the areas
activated during sentence reading had been disturbed by the
lesion.

Case 2
Structural Analysis
Selected anatomic MR images before and at 3 months after
stroke with resultant receptive aphasia are shown in Figure 4.
Initially (Figure 4a and 4b), the epileptic lesion in the left
temporal lobe was a focal region of hypointensity on T1- and
T2-weighted images. At 3 months, an extensive cortical
region of signal hyperintensity with encephalomalacia was
present on T2-weighted images (Figure 4c). This corresponded to a T1-hypointense region (Figure 4d) consistent

Figure 4. Images of patient 2 documenting the left-sided epileptic focus (arrow) before surgery (upper row) and after the
stroke (arrow) at 3 months follow-up (lower row) using representative slices of T2-weighted images (a, c) and diffusion-weighted
images (b, d).

with stroke. Differential damage to white matter surrounding
the lateral angle of the frontal horn of the lateral ventricle and
of the isthmus of the left temporal lobe have been related to
the differential rates of recovery.21 The lateral angle was
spared in patient 1, indicating that cingulate-supplementary
motor area connections to the caudate nucleus were intact. In
contrast, the isthmus of the left temporal lobe of patient 2 was
at least partially involved in the stroke. Thus, patient 2 was
confirmed as having a focal stroke in the superior temporal
gyrus by anatomic imaging at 3 months after the event, during
which time progressive recovery had occurred.
fMRI Analysis
The fortuitous availability of pre-event fMRI data from this
subject afforded a rare glimpse of a documented shift in the
hemispheric dominance after stroke. At 3 and 9 months, the
subject participated in the same fMRI language study as used
before surgery. The activation maps from these studies are
shown in Figure 5. The laterality ratio of Wernicke’s area
changed progressively from strong left hemisphere dominance before the event to weak right dominance after 3
months and considerable right hemisphere dominance by 9
months. Although the stroke did not involve Broca’s area,
there was a slight shift from weak right dominance before the
event to weak left dominance by 9 months. In general, the
large-scale network seemed to be reestablishing function in
an interactive way among its member nodes. No new regions
of activation other than those observed in normal subjects
were identified. Activation in the areas related to eye movement were normal, indicating again that not all of the areas

Thulborn et al

Figure 5. Activation maps of patient 2(a) before stroke with epileptic focus (vertical focus) and at (b) 3 months and (c) 9
months after stroke (oblique arrows) using the same paradigm
and activation labels as in Figure 1.

activated during sentence reading were disturbed by the
lesion. These observations support the hypothesis that recovery was associated with a redistribution of workload over the
existing large-scale network, with consolidation occurring
over many months.

Discussion
Downloaded from http://ahajournals.org by on April 11, 2024

Because language function, like other higher level cognitive
processes, is subserved by multiple brain areas,10,11 focal
brain damage and subsequent recovery constitute a change to
the large-scale network supporting such cognition. In the case
of patient 2, with a long standing underlying disease process
that may have influenced the development and maturation of
the language network, we caution that this pattern of redistribution may not generalize to recovery patterns of patients
with simple acute embolic stroke. However, the observations
are that when a key node of a large-scale cortical network is
damaged by a stroke, undamaged network components
(namely, contralateral homologs) are increasingly recruited to
increase their workload. Hence, as the workload of the
remaining network is modified, a shift in cognitive workload
can occur toward the contralateral hemisphere. There may
also exist other patterns of compensatory cortical activation
that have been observed in chronic patients years after stroke,
such as increased activation in areas immediately adjacent to
the lesion.5 Any of these long-term adaptations of a network
constitutes a form of plasticity that is associated with recovery of language function after brain damage.5
The contribution of fMRI to the understanding of such
phenomena is its ability to measure and characterize the
activity of a large-scale cortical network and to noninvasively
monitor any changes in its organization during the course of
spontaneous or intervention-based recovery from stroke. The
language paradigm used for these cases is a particularly
useful one in that abnormal activation patterns were detected
for both expressive and receptive aphasias. The evaluation of
the effects of therapies to promote recovery from stroke,

April 1999

753

whether they be pharmacological, surgical, or behavioral, can
be guided by functional neuroimaging.
Involvement of contralateral areas of the brain after stroke
in adults has been documented previously in aphasia recovery
weeks after the event at low spatial resolution by cortical DC
potential changes 22 and in motor output recovery weeks to
months after the event by magnetic transcranial stimulation,23,24 blood flow responses,25 and motor-evoked responses.25 The PET studies were performed months to years after
recovery.3,4 The current study demonstrates a spontaneous
redistribution of function to the right hemisphere that occurred within days and continued over months as performance normalized during recovery from aphasia. The results
indicate the organizational flexibility of the cortical systems
that underlie higher-level function.15,16 This information may
be useful in designing future rehabilitation strategies that can
exploit this flexibility.21 Given that even the adult has
mechanisms of plasticity that can produce rapid functional
recovery despite the evolution of infarction, functional neuroimaging is essential for distinguishing the success of acute
stroke interventions from these innate compensatory
mechanisms.

Acknowledgments
The authors acknowledge financial support from Public Health
Service grants PHS-PO1-NS35949 and PHS-RO1-CA63661; National Institute of Mental Health Scientist Awards MH-00661 and
MH-00662; and General Electric Medical Systems.

References
1. Stromswold K. The cognition and neural bases of language acquisition.
In: Gazzaniga MS, ed. The Cognitive Neurosciences. Cambridge, Mass:
MIT Press; 1996:855– 870.
2. Muller RA, Rothermel RD, Behen ME, Muzik O, Mangner TJ,
Chakraborty PK, Chugani HT. Brain organization of language after early
unilateral lesion: a PET study. Brain Lang. 1998;62:422– 451.
3. Weiller C, Isensee C, Rijntjes M, Huber W, Muller S, Bier D, Dutschka
K, Woods RP, Noth J, Diener HC. Recovery from Wernicke’s aphasia: a
positron emission tomographic study. Ann Neurol. 1995;37:723–732.
4. Engelien A, Silbersweig D, Stern E, Huber W, Doring W, Frith C,
Frackowiak RS. The functional anatomy of recovery from auditory
agnosia: a PET study of sound categorization in a neurological patient and
normal controls. Brain. 1995;118:1395–1409.
5. Frackowiak RS, Friston KJ, Frith CD, Dolan RJ, Mazziotta JC. The
cerebral basis of functional recovery. In: Human Brain Function. San
Diego, Calif: Academic Press; 1997:275–299.
6. Kwong KK, Belliveau JW, Chesler DA, Chesler DA, Goldberg IE,
Weisskoff RM, Poncelet BP, Kennedy DN, Hoppel BE, Cohen MS,
Turner R, Cheng H-M, Brady TJ. Dynamic magnetic resonance imaging
of human brain activity during primary sensory stimulation. Proc Natl
Acad Sci U S A. 1992;89:5675–5679.
7. Ogawa S, Tank DW, Menon RS, Ellerman JM, Kim SG, Merkle H,
Ugurbil K. Intrinsic signal changes accompanying sensory stimulation:
functional brain mapping using MRI. Proc Natl Acad Sci U S A. 1992;
89:5951–5955.
8. Thulborn KR, Davis D, Erb P, Strojwas M, Sweeney JA. Clinical fMRI:
implementation and experience. Neuroimage. 1996;4:S101–S107.
9. Thulborn KR, Voyvodic J, Chang S, Strojwas M, Sweeney JA. New
approaches to cognitive function by high field functional MRI. In: Yuasa
T, Prichard JW, Ogawa S, eds. Current Progress in Functional Brain
Mapping: Science and Applications. Niigata, Japan/London, UK: Nishimura/Gordon-Smith; 1998:15–23.
10. Just MA, Carpenter PA, Keller TA, Eddy WF, Thulborn KR. Brain
activation modulated by sentence comprehension. Science. 1996;274:
114 –116.
11. Binder JR, Rao SM, Hammeke TA, Yetkin FZ, Jesmanowicz A,
Bandettini PA, Wong EC, Estkowski LD, Goldstein MD, Haughton VM,

754

fMRI of Aphasia Recovery After Stroke

Hyde JS. Functional magnetic resonance imaging of human auditory
cortex. Ann Neurol. 1994;35:662– 672.
12. Carpenter PA, Just MA, Keller T, Eddy W, Thulborn KR. Graded functional activation in the visuo-spatial system with the amount of task
demand. J Cogn Neurosci. 1999;11:9 –24.
13. Luna B, Thulborn KR, Strojwas MH, McCurtain BJ, Berman RA,
Genovese CR, Sweeney JA. Dorsal cortical regions subserving visually
guided saccades in humans: an fMRI study. Cereb Cortex. 1998;8:40 – 47.
14. Hertz-Pannier L, Gaillard WD, Mott SH, Cuenod CA, Bookheimer SY,
Weinstein S, Conry J, Papero PH, Schiff SJ, LeBihan D, Theodore WH.
Noninvasive assessment of language dominance in children and adolescents with functional MRI. Neurology. 1997;48:1003–1012.
15. Mesulam M-M. Large-scale neurocognitive networks and distributed processing for attention, language, and memory. Ann Neurol. 1990;28:
597– 613.
16. Mesulam M-M. From sensation to cognition. Brain. 1998;121:
1013–1052.
17. Thulborn KR, Gillen JS, McCurtain B, Betancourt C, Sweeney JA.
Functional magnetic resonance imaging of the human brain. Bull Magn
Reson. 1996;18:37– 42.
18. Thulborn KR, Uttecht S, Betancourt C, Talagala SL, Boada FE, Shen GX.
A functional, physiological, and metabolic toolbox for clinical magnetic
resonance imaging: integration of acquisition and analysis strategies. Int
J Imaging Sys Technol. 1997;8:572–581.

19. Cox RW. AFNI: software for analysis and visualization of functional
magnetic resonance neuroimages. Comput Biomed Res. 1996;29:
162–173.
20. Talairach J, Tournoux P. Co-Planar Stereotaxic Atlas of the Human
Brain: 3D Proportional System: An Approach to Cerebral Imaging. New
York, NY: Georg Thieme Verlag; 1988.
21. Naeser M. Neuroimaging and recovery of auditory comprehension and
spontaneous speech in aphasia with some implications for treatment in
severe aphasia. In: Kertesz A, ed. Localization and Neuroimaging in
Neuropsychology. San Diego, Calif: Academic Press; 1994:245–295.
22. Thomas C, Altenmuller E, Marckmann G, Kahrs J, Dichgans J. Language
processing in aphasia: changes in lateralization patterns during recovery
reflect cerebral plasticity in adults. Electroencephalogr Clin Neurophysiol. 1997;102:86 –97.
23. Cicinelli P, Traversa R, Rossini PM. Post-stroke reorganization of brain
motor output to the hand: a 2– 4 month follow-up with focal magnetic
transcranial stimulation. Electroencephalogr Clin Neurophysiol. 1997;
105:438 – 450.
24. Netz J, Lammers T, Homberg V. Reorganization of motor output in the
non-affected hemisphere after stroke. Brain. 1997;120:1579 –1586.
25. Honda M, Nagamine T, Fukuyama H, Yonekura Y, Kimura J, Shibasaki
H. Movement-related cortical potentials and regional cerebral blood flow
change in patients with stroke after motor recovery. J Neurol Sci. 1997;
146:117–126.

Downloaded from http://ahajournals.org by on April 11, 2024