lesion location alters brain activation in chronically impaired stroke survivors
TRANSCRIPT
www.elsevier.com/locate/ynimg
NeuroImage 21 (2004) 924–935
Lesion location alters brain activation in chronically
impaired stroke survivors
Andreas R. Luft,a,b,* Sandy Waller,c Larry Forrester,c Gerald V. Smith,c Jill Whitall,c
Richard F. Macko,d,e Jorg B. Schulz,b and Daniel F. Hanleya
aDepartment of Neurology, Johns Hopkins University, Baltimore, MD 21287, USAbDepartment of Neurology, University of Tubingen, Tubingen 72076, GermanycDepartment of Physical Therapy, University of Maryland, Baltimore, MD 21201, USAdDepartment of Neurology, University of Maryland, Baltimore, MD, USAeDepartment of Veterans Affairs and the Baltimore VA Geriatric Research Education and Clinical Center (GRECC), USA
Received 25 June 2003; revised 2 October 2003; accepted 14 October 2003
Recovery of motor function after stroke is associated with reorganiza-
tion in central motor networks. Functional imaging has demonstrated
recovery-dependent alterations in brain activation patterns when
compared to healthy controls. These alterations are variable across
stroke subjects. Factors identified as contributing to this variability are
the degree of functional impairment, the time interval since stroke, and
rehabilitative therapies. Here, the hypothesis is tested that lesion
location influences the activation patterns. Using functional magnetic
resonance imaging, the objective was to characterize similarities or
differences in movement-related activation patterns in patients chroni-
cally disabled by cortical plus subcortical or subcortical lesions only.
Brain activation was mapped during paretic and non-paretic move-
ment in 11 patients with subcortical stroke, in nine patients with stroke
involving sensorimotor cortex, and in eight healthy volunteers. Patient
groups had similar average motor deficit as measured by a battery of
scores and strength measures. Substantial differences between patients
groups were found in activation patterns associated with paretic limb
movement: Whereas contralateral motor cortex, ipsilateral cerebellum
(relative to moving limb), bilateral mesial (cingulate, SMA), and
perisylvian regions were active in subcortical stroke, cortical patients
recruited only ipsilateral postcentral, mesial hemisphere regions, and
areas at the rim of the stroke cavity. For both groups, activation in
ipsilateral postcentral cortex correlated with motor function; in
subcortical stroke, the same was found for mesial and perisylvian
regions. Overall, brain activation in cortical stroke was less, while in
subcortical patients, more than in healthy controls. For non-paretic
movement, activation patterns were similar to control in cortical
patients. In subcortical patients, however, activation patterns differed:
the activation of non-paretic movement was similar to that of paretic
movement (corrected for side). The data demonstrate more differences
than similarities in the central control of paretic and non-paretic limb
movement in patients chronically disabled by subcortical versus
cortical stroke. Whereas standard motor circuitry is utilized in
subcortical stroke, alternative networks are recruited after cortical
stroke. This finding proposes lesion-specific mechanisms of reorgan-
1053-8119/$ - see front matter D 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.neuroimage.2003.10.026
* Corresponding author. Department of Neurology, Universitat Tubin-
gen, Hoppe-Seyler-Strasse 3, 72076 Tubingen, Germany. Fax: +49-7071-
967857.
E-mail address: [email protected] (A.R. Luft).
Available online on ScienceDirect (www.sciencedirect.com.)
ization. Optimal activation of these distinct networks may require
different rehabilitative strategies.
D 2004 Elsevier Inc. All rights reserved.
Keywords: Stroke; Hemiparesis; Sensorimotor cortex; Cerebellum; Func-
tional imaging; Chronic
Introduction
Recovery of motor function occurs in most patients after
hemiparetic stroke and is associated with reorganization of central
motor networks (Liepert et al., 2000; Nelles et al., 2001; Rijntjes
and Weiller, 2002). Activation patterns across these networks have
been characterized using functional MRI. Several studies described
the differences between brain activation patterns of patients and
healthy controls (Chollet et al., 1991; Cramer et al., 1997; Seitz et
al., 1998; Weiller et al., 1992, 1993). In recovered patients, three
patterns of activation have been identified likely representing
different mechanisms of reorganization: (1) enlarged activation in
primary motor cortex (M1) of the damaged hemisphere (contralat-
eral to movement). (2) Recruitment of secondary motor cortices
(SMA, premotor cortex, cingulate areas) or somatosensory cortex
in the lesioned hemisphere. (3) Recruitment of sensorimotor motor
cortex in the undamaged hemisphere ipsilateral to the moving limb
(Cao et al., 1998; Cramer et al., 1997; Rossini et al., 1998; Weiller
et al., 1993). During the phase of spontaneous recovery, lesioned
hemisphere (contralateral) motor cortex activation shrinks in a
subset of patients (Feydy et al., 2002) and may grow in others
(Cuadrado et al., 1999; Marshall et al., 2000).
Factors contributing to the variability of activation patterns
associated with hemiparetic movement in different subjects are
their individual degree of recovery (Ward et al., 2003) and the time
interval since stroke (Feydy et al., 2002). Some studies postulate or
indicate lesion location as another factor determining brain activa-
tion during paretic movement (Chen et al., 2000; Feydy et al.,
2002; Fries et al., 1993; Shelton and Reding, 2001). Several
A.R. Luft et al. / NeuroImage 21 (2004) 924–935 925
pathophysiological hypotheses have been proposed to explain this
influence: subcortical stroke may leave intracortical circuitry intact
thereby allowing for cortical reorganization to compensate func-
tional deficits (Byrnes et al., 2001). Cortical strokes may substan-
tially impair these mechanisms and require other brain areas to
mediate functional recovery. On the other hand, descending motor
pathways are more vulnerable at points of convergence in the
internal capsule and peduncles—regions injured in subcortical
stroke—as compared to corona radiata and motor cortex (Shelton
and Reding, 2001). In many studies mapping the motor system
during paretic movement, patients with various lesion distributions
and morphologies are being combined—in part due to the paucity
of eligible candidates for fMRI studies.
To gain a better understanding of the interrelation between
lesion location and the functioning of the lesioned central motor
system, subject samples corrected for other influencing factors—
time-since-stroke and degree of recovery—have to be studied.
Here, we investigate how lesion location (pure subcortical versus
cortical plus subcortical) affects the neural control of paretic limb
movement in patient groups with similar paresis and degree of
recovery at a time when all potential for spontaneous recovery has
been realized, hence, the patients are clinically stable.
Materials and methods
Subjects
Twenty patients (age: 62 F 9 years, mean F SD, 9 females, 11
males), who suffered a first-ever cortical or subcortical stroke 53.2F46.3 months (mean F SD) before inclusion, underwent fMRI of
bilateral elbow movement. All patients were hemiparetic with
residual movement from a single ischemic stroke, and were selected
based on the presence of residual hemiparesis, completion of all
conventional rehabilitation programs, and adequate language and
neurocognitive function to understand consenting and study-related
Fig. 1. For each patient group, the distribution of lesions is shown superimposed o
indicate in how many of 9 cortical stroke and 11 subcortical stroke patients, resp
instructions. Patients with more than one symptomatic stroke, other
neurological disease, chronic pain, emotional disorders, and/or
metal implants were excluded. All patients received comparable
conventional rehabilitation therapy during the initial 6 months after
their stroke. All subjects, recalling the time before their stroke, were
right-handed according to the Oldfield inventory.
Upper extremity motor function was assessed at the time of
study enrollment using a test battery including the upper-extremity
portion of the Fugl-Meyer Motor Performance assessment (scores
0–66, best) (Fugl-Meyer et al., 1975), the weight section of the
Wolf Motor Arm test (Wolf weight) (Wolf et al., 2001), the time to
complete the Wolf task battery (Wolf time), and strength of
shoulder and elbow extension measured as the peak force (in kg)
in three consecutive trials using a dynamometer. A combined
functional score consisting of the sum of the quotients individual
score/mean score per group was entered into a general linear
model (see below). Because medical records for patients with
distant strokes were not available, the degree of initial impairment
(acutely after stroke) was assessed by asking the patient to
subjectively categorize initial disabilities (‘‘no movement’’, ‘‘major
impairment/not able to lift arm’’, ‘‘minor impairment/able to lift
arm’’).
The patient sample was stratified according to whether the
lesion involved motor cortex F subcortical regions (further named
cortical group) or only subcortical regions (subcortical group,
subcortical regions were all areas medial to the insular cortex and
ventral to the corpus callosum, that is, basal ganglia and descend-
ing pathways). Fig. 1 shows the average stroke locations per group.
Table 1 summarizes the patient characteristics and details of lesion
location.
Patients were compared to a group of healthy volunteers (five
males, three females, age 55 F 8.5 years). Full written informed
consent was obtained from all subjects in accordance with the
Declaration of Helsinki. The study protocol and the consent forms
were approved by the participating institutions (Johns Hopkins
University and University of Maryland).
nto an averaged anatomical image derived from all patients. Shades of gray
ectively, a particular area was lesioned (brighter—injury more frequent).
Table 1
Patient characteristics
Case
number
Age Sex Side of
stroke
Months
since
stroke
Location of stroke Fugl-Meyer
score
Shoulder
strength
(kg)
Elbow
strength
(kg)
Wolf
weight
Wolf
time
(s)
Lesion
volume
(Al)
Subcortical
1 59 f R 50.9 posterior putamen, posterior limb of int. caps. 25 2.5 6.8 0.5 72.0 5135
2 68 f R 21.0 posterior putamen, posterior limb of int. caps. 42 6.4 6.1 0.0 29.5 2673
3 37 m L 9.2 pallidum, mid putamen, knee of int. caps. 31 10.2 22.3 0.9 53.1 3913
4 42 f R 60.1 putamen, int. caps. 49 6.1 8.2 2.7 11.8 9835
5 66 m L 9.6 putamen, ant limb of int. caps
(small extension into corona rad.)
56 15.4 13.6 4.1 18.7 6726
6 67 m R 33.2 ant. putamen, head of caudate, lateral thalamus 22 3.4 5.2 1.9 72.3 909
7 77 m R 91.2 upper pons (midline, extending to R more than to L) 54 14.5 20.4 4.1 3.7 981
8 64 f R 16.1 lower pons 25 5.4 5.0 0.0 70.7 776
9 75 m L 50.5 post limb of int. caps. 35 8.2 8.2 1.4 25.2 1664
10 74 m R 180.6 post limb of int. caps. (small extension into corona rad.) 26 3.4 6.6 0.0 33.7 162
11 71 f R 156.6 central corona radiata, upper pole of int. caps. 26 7.0 5.0 0.0 74.4 8018
Cortical
12 77 m L 38.2 parietal lobe, post. insula, int caps., putamen 36 6.1 8.0 0.0 66.2 248 556
13 61 f R 31.9 precentral, post. part of frontal lobe, insula,
putamen, int. caps.
21 13.0 0.0 1.6 105.0 104 988
14 61 m R 57.3 postcentral, corona rad., int. caps., putamen 28 10.7 13.2 0.5 21.0 48 880
15 76 m R 37.6 precentral, post. part of frontal lobe,
superior temp. gyrus,
corona rad., int. caps.
24 8.6 9.8 0.0 66.3 231 906
16 62 m R 79.2 pre-, postcentral, post. part of frontal lobe, int. caps.,
caudatum, putamen, (complete MCA territory)
27 5.0 0.0 0.0 55.9 238 308
17 54 m R 10.7 pre-, postcentral, caudatum 46 19.5 10.7 1.8 28.6 62 389
18 64 f L 71.5 premotor cortex, inf. frontal gyrus, ant. temporal lobe,
int. caps., caudatum, putamen
20 2.3 7.5 0.0 88.4 130 954
19 47 m R 9.8 pre-, postcentral, post. part of frontal lobe,
sup. temp. gyrus,
int. caps., caudatum, putamen (complete MCA territory)
23 3.2 6.6 0.0 89.2 245 709
20 75 m R 49.7 pre-, postcentral, premotor, insula, ant. putamen 43 6.6 7.3 1.8 19.5 63 128
A.R. Luft et al. / NeuroImage 21 (2004) 924–935926
fMRI technique and paradigms
fMRI technique and paradigms were described before (Luft et
al., 2002). In brief, a custom-made plexiglass scaffold was used
inside the scanner to ensure constant movement range and plane of
motion and to limit concomitant head motion by fixation of arm and
body. fMRI scanning was performed using a 1.5 Tesla scanner
(Gyroscan, Philips, Eindhoven, The Netherlands) at the F.M. Kirby
Center for Functional Brain Imaging, Kennedy Krieger Institute,
Baltimore. AT1-weighted 3D-MPRAGE sequence (TR 8.2 ms, TE
3.7 ms, flip angle 8j, NEX 1, matrix 256 � 256, voxel-resolution:
1 � 1 � 1 mm3) was obtained for anatomical mapping. BOLD-
weighted EPI sequences (TR 3 s, TE 40 ms, matrix: 64 � 64, slice
thickness 5 mm, pixel resolution 3.75 � 3.75, 30–36 slices to
obtain whole brain coverage, interleaved slice timing) were ac-
quired in axial orientation. For each movement paradigm, 60 scans
were acquired over a period of 3 min without interscan delay. One
measurement consisted of three cycles of rest and 10 elbow move-
ments beginning with rest. Elbow movements were acoustically
cued by computer-generated beeps (one every 3 s, beep duration 0.1
s). This low frequency of movement allowed all patients to perform
the movement without fatiguing, as tested before the scanning
procedure. Elbow flexion started at 45j and ended at 60j–75j and
was followed by return extension. The movement range was
adjusted according to the patient’s ability, again with the goal of
avoiding fatigue during the measurement. The plane of motion was
in parallel to the longitudinal axis of the supine body and angled
60j off the horizontal plane. Non-paretic and paretic limbs were
assessed separately. Likewise, control subjects moved their right
and left limbs separately. Performance was video-taped to monitor
for compliance. Subjects kept their eyes closed during scanning.
Before or after the fMRI session, biceps and deltoid EMG
activity were recorded bilaterally using surface electrodes while the
patient contracted the hemiparetic limb in a mock setup similar to
the fMRI setting. Neither was there evidence for cocontraction in
EMG tracings, nor did any patient show overt mirror movement
during the mock or the fMRI session (as monitored with contin-
uous video footage).
fMRI analysis
BrainVoyager (Brain Innovation B.V., Maastricht, The Nether-
lands) was used to analyze fMRI data according to standard
protocols.
Preprocessing
Functional images were corrected for motion and slice timing,
and temporally (high pass filter, linear trend removal) and spatially
(FWHM 4 mm) smoothed. Effectiveness of motion correction was
assessed by viewing post-correction movies of functional image
series. For none of the subjects did this procedure reveal any
visible head motion in the corrected data sets. Images were then
A.R. Luft et al. / NeuroImage 21 (2004) 924–935 927
transformed to Talairach space, first, by applying rigid body
transformations (translation, rotation), followed by manual defini-
tion of Talairach landmarks, and third, by scaling of 12 landmark-
defined subvolumes into Talairach space.
Statistical analysis
Movement and rest periods were modeled by a boxcar function
with hemodynamic response modification (predictor movement).
Multistudy general linear models (GLM) were used to extract foci
of activation. Several GLMs were computed:
(1) a GLM including either paretic or non-paretic movement
paradigms of all patients per group (cortical, subcortical). Image
Fig. 2. Average activation maps per movement paradigm and group are shown [mu
two rows, data from subcortical (yellow) and cortical patients (red) are superimp
outlined on images in the top row (pink line). During paretic movement in cortica
(filled red arrow), in the intact ipsilateral hemisphere (open red arrow), and in regio
show more overall activation, especially in contralateral motor cortex (filled yellow
perisylvian areas. During non-paretic limb movement (middle row), both groups
activation. Subcortical stroke patients recruit additional mesial hemisphere and per
(data from right- and left sided movement combined; data from right elbow mov
SMA, and bilateral cerebellar activation.
data of patients with left hemisphere lesions were flipped about the
mid-sagittal line, such that all subjects were considered to have a
right-sided lesion. Independent variables were movement and a
constant term.
(2) A GLM combining two groups of subjects (cortical versus
subcortical patients or patients versus volunteers) for either para-
digm (paretic, non-paretic). For comparisons between patients and
volunteers, left- and right-sided movement trials were averaged in
voluenteers after flipping of the data about the mid-sagittal line
moving limb of the brain was always contralateral to the moving
limb. Independent variables of these GLMs were movement and
constant. Appropriate contrasts were formed to extract activation
ltisubject GLM, random effects, threshold P < 0.05 (corrected)]. In the upper
osed. The extent of the maximum lesion of patients with cortical stroke is
l stroke patients (top row), activation is found adjacent to the stroke cavity
ns on the medial surface of the hemispheres. Patients with subcortical stroke
arrow), ipsilateral cerebellum (open yellow arrow), mesial hemisphere, and
of patients demonstrate contralateral motor cortex and ipsilateral cerebellar
isylvian areas. Healthy controls show during elbow movement (bottom row)
ement were flipped about the mid-sagittal line), contralateral motor cortex,
Table 2
Average activation per group and task
Side Anatomical/functional Brodmann Talairach coordinate
region areax y z
Cortical patients-paretic
Contralateral cingulate gyrusa 31 17 �31 41
Contralateral fusiform gyrus 37 32 �47 �8
Ipsilateral S1 2 �37 �27 31
Ipsilateral paracentral lobule 5 �19 �43 50
Ipsilateral posterior cingulate 29 0 �38 18
Ipsilateral cingulate gyrus 24 �18 �16 38
Ipsilateral cingulate gyrus 31 �10 �10 43
Subcortical patients-paretic
Contralateral S1 3 47 �21 55
Contralateral M1 4 34 �21 56
Contralateral SMA 6 3 �21 55
Contralateral insula 13 36 �35 19
Contralateral cingulate gyrus 23 6 �23 31
Contralateral S2 40 35 �34 55
Contralateral precentral gyrus 44 46 2 8
Bilateral paracentral lobule 31 F6 �23 45
Bilateral post. cingulate 29 F2 �39 19
Bilateral medial geniculate
body
thalamus F18 �23 �2
Ipsilateral precentral gyrus 6 �53 �3 8
Ipsilateral insula 13 �48 �34 19
Ipsilateral sup. temporal gyrus 22 48 �26 3
Bilateral quadrangular lobe cerebellum F10 �55 �7
Bilateral simple lobule cerebellum F18 �69 �17
culmen of vermis cerebellum 0 �56 �13
Cortical patients-nonparetic
Contralateral M1 4 �28 �24 51
Contralateral S1 2 �34 �33 40
Contralateral S2 40 �34 �38 42
Contralateral sup. temporal gyrus 42 �58 �24 17
Ipsilateral quadranguar lobe cerebellum 20 �47 �21
Ipsilateral culmen of vermis cerebellum 5 �53 �10
Subcortical patients-nonparetic
Contralateral M1 4 �46 �5 43
Contralateral S1 3 �49 �17 43
Contralateral S1 2 �42 �26 29
Contralateral SMA 6 �2 �19 50
Contralateral postcentral gyrus 5 �29 �42 63
Contralateral insula 13 �37 �23 15
Contralateral anterior cingulate 32 �2 25 �9
Contralateral fusiform gyrus 37 �31 �44 �13
Contralateral middle frontal gyrus 46 �40 33 12
Contralateral inferior frontal gyrus 47 �52 23 �1
Contralateral VL nucleus thalamus �19 �12 5
Ipsilateral M1 4 44 �14 45
Ipsilateral S1 2 38 �37 58
Ipsilateral premotor cortex 6a 41 3 38
Ipsilateral inferior frontal gyrus 9 52 5 30
Ipsilateral middle temporal
gyrus
19 45 �58 15
Ipsilateral inferior temporal
gyrus
20 46 �2 �32
Ipsilateral fusiform gyrus 37 44 �52 �20
Ipsilateral superior temporal
gyrus
39 45 �56 17
Ipsilateral inferior parietal
lobule
40 43 �41 40
Side Anatomical/functional Brodmann Talairach coordinate
region areax y z
Subcortical patients-nonparetic
Ipsilateral superior temporal
gyrus
41 47 �26 13
Ipsilateral inferior frontal
gyrus
46 39 34 14
Ipsilateral VL nucleus thalamus 11 �16 2
Ipsilateral culmen cerebellum 8 �53 �12
Ipsilateral quadrangular
lobule
cerebellum 24 �38 �27
Ipsilateral declive cerebellum 5 �76 �14
Ipsilateral pyramis cerebellum 5 �83 �24
Volunteers—left/right
Contralateral M1 4 29 �27 60
Contralateral S1 3 29 �30 60
Contralateral SMA 6 2 �5 60
Contralateral insula 13 47 �23 24
Ipsilateral S2 40 �53 �31 21
Ipsilateral quadrangular
lobule
cerebellum �18 �46 �20
Contralateral quadrangular
lobule
cerebellum 33 �52 �24
a Activation focus adjacent to average stroke cavity.
Table 2 (continued)
A.R. Luft et al. / NeuroImage 21 (2004) 924–935928
present during both (conjunction map) or either one task (differ-
ence map: group A–group B, t value in map A greater than in map
B, yielding regions in which activation is more likely in group A).
Contrasts were balanced for different sample sizes.
(3) A GLM combining paretic and non-paretic or left- and
right-sided movement paradigms per group of patients (cortical,
subcortical) or volunteers, respectively. Image data of right-sided
movement paradigms were flipped about the mid-sagittal line (left
side of brain = contralateral to moving limb). The independent
variables of this model were movement and constant. Appropriate
contrasts were formed to extract activation present during both
(conjunction map) or either tasks (difference map).
(4) Similar to GLM 1. The contrast across the covariate
movement was then weighted according to the subject’s combined
functional score (see above), thereby extracting voxels with a
linear (direct or indirect) correlation between the functional score
and the parameter estimate for movement (Ward et al., 2003).
Random effects analysis with a uniform probability threshold
of 0.05 (corrected for multiple comparisons across the entire
brain) was applied. For visualization of functional data, T1-
weighted high-resolution image series were averaged per group.
The Talairach Daemon (Lancaster et al., 2000) with a query range
of 5 mm was used to identify cortical, subcortical, and cerebellar
regions.
Results
General variables characterizing subject groups
Subcortical and cortical stroke patients did not differ in age
(P = 0.93), time interval since stroke (P = 0.38), side of stroke
(subcortical: left/right 3/8, cortical: 2/7), deficit at the time of
stroke (P = 0.34), or motor function and strength scores at the
Fig. 3. Overall differences between activation patterns per paradigm and
group are demonstrated by plotting the number of significant voxels in
conjunction (black) and difference (gray) multisubject GLMs. The
conjunction GLM yields voxels, which are likely active in both groups/
paradigms. In the difference GLM, voxels more likely to be active in group
A than in group B (positive values) or vice versa (negative values) are
extracted.
A.R. Luft et al. / NeuroImage 21 (2004) 924–935 929
time of study enrollment (FM: 35.5 F 12.6 vs. 29.8 F 9.6, P = 0.3;
Wolf weight: 1.4 F 1.6 vs. 0.6 F 0.8 P = 0.2; Wolf time: 42.3 F26.9 vs. 60.0 F 31.5, P = 0.2; shoulder strength: 7.5 F 4.3 vs.
Fig. 4. Multisubject conjunction GLM maps showing activation present in both,
paretic limb: (a) for cortical stroke patients only a small area in the contralater
postcentral gyri for patients with subcortical stroke.
8.3 F 5.4, P = 0.7; elbow strength: 9.8 F 6.2 vs. 7.9 F 3.8, P =
0.5). Stroke patient and volunteer groups did not differ in age or
gender distribution.
Activation maps in patients and volunteers (GLM 1)
Average activation maps for each group of patients and controls
are presented in Fig. 2. Activation foci and respective Talairach
coordinates are listed in Table 2.
Paretic limb
Activation of contralateral primary motor cortex (M1, BA 4),
primary somatosensory cortex (S1, BA 3, 1, 2) and supplementary
motor area (SMA) was consistently found in subcortical stroke
patients and controls (Fig. 2). In contrast, patients with cortical
stroke did show an activation focus in the contralateral hemisphere,
medial and ventral to the primary motor cortex (Fig. 2). This
activation mapped to the rim of the stroke cavity (outlined in Fig.
2). Other areas activated in cortical stroke were bilateral mesial
hemisphere regions (cingulate gyrus, paracentral lobule) and ipsi-
lateral S1 (BA 2) and premotor cortex (BA 6a). In subcortical
patients, besides activation in contralateral sensorimotor cortex,
activation was found in bilateral mesial hemispheres, bilateral
insulae, and ipsilateral premotor cortex (Table 2). Control subjects
showed additional activation of the contralateral insula and ipsi-
lateral S2.
The cerebellum was active in controls and subcortical stroke
with a strong ipsilateral dominance (Fig. 2). Foci mainly mapped to
quadrangular and simple lobules and the vermis. No cerebellar
activation was found in cortical stroke patients even after lowering
the probability threshold to 0.0001 (uncorrected).
Non-paretic limb
During non-paretic limb movement, both cortical and subcor-
tical patients showed contralateral M1 and S1 and ipsilateral
cerebellar (quadrangular lobe) activation (Fig. 2). In subcortical
patients, several additional regions were active (Table 2). Among
healthy controls (right and left averaged) and stroke patients moving their
al motor cortex is significant, (b) larger regions are extracted in pre- and
A.R. Luft et al. / NeuroImag930
those were ipsilateral sensorimotor areas, mesial hemisphere, and
regions in the temporal lobe.
Comparison: patients versus healthy volunteers (GLM 2)
Cortical stroke
During paretic limb movement, less voxels were activated in
cortical stroke patients than in controls (number of voxels in the
difference map cortical—control, Fig. 3). The difference GLM
showed that activation of contralateral M1, premotor cortex,
precuneus (BA 7), and bilateral cerebellum was more likely in
controls, whereas activation of the ipsilateral S1 (BA 2) was more
probable in cortical stroke patients than in controls. The conjunc-
tion GLM—areas active in both, volunteers and cortical patients—
extracted voxels in the contralateral S1 (BA 3, Fig. 4a).
Non-paretic movement produced similar activation maps be-
tween cortical stroke patients and controls. No activated voxels
were found in the difference GLM. The conjunction GLM
revealed activation of the ipsilateral cerebellum (quadrangular
lobule).
Fig. 5. Multisubject GLM maps (conjunction or difference as labeled) comparing
data is flipped about the mid-sagittal line). (a) In cortical stroke patients, a small are
map shows cerebellar activation to be more likely during non-paretic movemen
contralateral motor cortex, mesial areas, and cerebellum are found.
Subcortical stroke
Unlike cortical stroke patients, subcortical patients presented
more overall brain activation as compared to volunteers (Fig. 3).
Regions in which activation was more likely than in controls were
contralateral SMA, ipsilateral S1 (BA 3), ipsilateral insula (BA
13), bilateral mesial hemisphere areas (BA 7, 29), bilateral
superior temporal gyrus (BA 41), and bilateral cerebellum (simple
lobule). Conjunction analysis revealed congruent activation of
contralateral M1, S1, S2 (BA 40), SMA, bilateral insula (BA
13), and ipsilateral cerebellum in both subcortical stroke patients
and volunteers (Fig. 4b).
The comparison of non-paretic movement-related brain activa-
tion in subcortical patients and controls revealed differences and
similarities. The difference GLM extracted as many voxels as the
conjunction GLM (Fig. 3). Areas more likely to be active in
patients were all in the ipsilateral hemisphere including ipsilateral
M1, S1, BA 5, 9, 13, 21, 39, 41, and bilateral cerebellum
(quadrangular lobule). The conjunction map showed activation in
contralateral M1, S1, S2 (BA 40), SMA, bilateral insula (BA 13),
and ipsilateral cerebellum (quadrangular lobule).
e 21 (2004) 924–935
paretic and non-paretic movement per patient group (non-paretic movement
a of congruent activation is identified in the precentral gyrus. The difference
t. (b) In subcortical stroke patients, large areas of congruent activation in
Fig. 6. Multisubject GLM maps comparing paretic limb activation patterns in cortical and subcortical stroke patients are shown. Activation of contralateral
cortical stroke.
A.R. Luft et al. / NeuroImage 21 (2004) 924–935 931
Comparison: paretic versus non-paretic movement (GLM 3)
Whereas subcortical stroke patients presented high congruence
between overall brain activation during paretic and non-paretic
movement, little congruence was found in cortical stroke patients
(Fig. 3). In cortical stroke, only the contralateral postcentral gyrus
(BA 2) was active in both paretic and non-paretic movement
(Fig. 5a, left). The difference GLM (non-paretic – paretic)
revealed activation in contralateral BA 9 and 37 and in bilateral
cerebellum (quadrangular and simple lobules, Fig. 5a, right). In
subcortical stroke patients, the conjunction GLM between paretic
and non-paretic movement paradigms showed congruent regions:
contralateral M1, S1, premotor cortex, bilateral S2 (BA 40),
bilateral SMA, bilateral insula (BA 13), ipsilateral cingulate
gyrus (BA 29 and 31), and bilateral cerebellum (simple lobule)
(Fig. 5b, left and right). The difference GLM (non-paretic–
paretic) revealed activation in contralateral SMA, ipsilateral
motor cortex and ipsilateral cerebellum is more likely in subcortical than in
Fig. 7. Multisubject GLMs with the combined functional score as covariate are sh
better motor function are indicated in red for cortical and yellow for subcortical
correlates with the functional score in cortical patients (filled red arrow). In patien
hemisphere regions (cingulum), and in ipsilateral postcentral gyrus (yellow arrow
premotor cortex (BA 6a), and bilateral superior frontal gyri
(BA 8, 10) (data not shown).
Comparison: cortical stroke versus subcortical stroke
(GLM 2)
Paretic arm
There was more overall brain activation in subcortical stroke
than in cortical stroke (Fig. 2). The number of voxels more likely
to be active in subcortical than in cortical stroke was higher than
the number of voxels in the conjunction GLM (Fig. 3). The
difference GLM (subcortical–cortical) extracted voxels in contra-
lateral M1, S1, insula (BA 13), ipsilateral superior temporal gyrus
(BA 22), and bilateral cerebellum (quadrangular and semilunar
lobules) (Fig. 6). The conjunction GLM showed activation in
contralateral S1 (BA 3), ipsilateral S2 (BA 40), and bilateral
cingulate gyri (BA 31) (data not shown).
own. Regions in which activation during paretic movement was related to
stroke patients. Activation in ipsilateral hemisphere S1 (BA 2, red arrow)
ts with subcortical stroke, the GLM extracted voxels in perisylvian, mesial
).
Table 3
Correlation of activation and combined functional score (GLM 4)
Side Anatomical/functional Brodmann Talairach
region areax y z
Cortical patients
Ipsilateral S1 2 �46 �25 45
Subcortical patients
Contralateral paracentral lobule 5 15 �35 49
Contralateral insula 13 44 8 7
Contralateral middle temporal gyrus 22 51 �30 6
Contralateral cingulate gyrus 23 5 �22 30
Contralateral precuneus 31 17 �40 30
Ipsilateral S1 2 �60 �21 28
Ipsilateral premotor cortex 6a �37 �5 42
Ipsilateral insula 13 �40 �6 18
A.R. Luft et al. / NeuroImage 21 (2004) 924–935932
Non-paretic arm
For non-paretic limb movement, activation patterns also dif-
fered between cortical and subcortical stroke patients (Fig. 2).
Regions more likely to be activated in subcortical stroke than in
cortical stroke (difference GLM) were contralateral M1, premotor
cortex (BA 6a), S1 (BA 3), and ipsilateral insula (BA 13). The
conjunction map revealed common activation of the ipsilateral
cerebellum (quadrangular lobule) (data not shown).
Correlation brain activation-functional score (GLM 4)
Regions in which activation correlated directly with the com-
bined functional score were identified for both cortical and
subcortical patients (Fig. 7). In cortical stroke patients, the ipsilat-
eral somatosensory cortex (S1, BA 2) was significant. In patients
with subcortical stroke, ipsilateral somatosensory (BA 2) and
premotor areas (BA 6) as well as mesial hemisphere and peri-
sylvian regions were more active in patients with better functional
score (Table 3).
Few regions were identified in which the parameter estimate
was inversely correlated with the combined functional score (more
activation, worse arm function). In cortical stroke patients, these
were ipsilateral precuneus (Talairach coordinates BA 31:�22/�77/
26, 19:�37/�74/36), in subcortical stroke patients, a small area in
the contralateral SMA (BA 6/8: 2/43/39).
Discussion
The main finding of this study is that activation patterns
associated with paretic limb movement differ in patients with
cortical (plus subcortical) versus subcortical lesions despite similar
chronic motor impairment. Secondly, activation patterns associated
with non-paretic limb movement differ from healthy control in
subcortical but not in cortical stroke patients. Subcortical stroke
patients demonstrate homology between activation patterns of
paretic and non-paretic movement.
Activation associated with paretic limb movement
In subcortical stroke, the activation pattern during paretic
movement was characterized by involvement of contralateral
motor cortex and ipsilateral cerebellum, representing standard, that
is, physiological motor circuitry, which was also found to be active
in controls. Additionally, activation was observed in ipsilateral pre-
and postcentral areas, bilateral perisylvian, and mesial regions.
Activation levels in the latter areas and in ipsilateral postcentral
gyrus were related to motor functional scores. Our data do not
confirm for chronic stroke patients, what has been reported in acute
stroke by Small et al. (2002), that is, that ipsilateral cerebellar
recruitment predicts good arm function (nine subcortical and five
cortical strokes). However, our analysis may be insensitive to
interrelation between activation and arm function because our
sample did not include subjects with complete motor recovery.
Therefore, the conclusion seems unjustified that contralateral
motor cortex and cerebellum are irrelevant to motor control of
paretic limb movement in subcortical stroke.
In a longitudinal study, Feydy et al. (2002) found widespread
activation in bilateral sensorimotor cortex (SMC), frontal and
mesial regions during the acute stage after stroke. With progressive
recovery, activation in ipsilateral, frontal, and mesial regions
disappeared while the contralateral SMC focus remained (progres-
sive focusing). In other subjects, widespread activation persisted.
Even, other subjects showed only contralateral SMC activation
already in the acute stage (initial focusing). There was no clear
association with the success of recovery. But, patients with M1
lesion were more likely to show persistence of widespread activa-
tion. Our data demonstrate a different pattern in stroke patients
with chronic impairment: widespread activation was a hallmark of
subcortical stroke whereas cortical stroke showed activation in few
areas, among those bilateral mesial regions and ipsilateral S1. This
discrepancy to the data of Feydy et al. suggests that activation
patterns may change between 6 months (last mapping in the study
by Feydy et al.) and 4.4 years (mean stroke-study interval in our
study). But, in view of limited sample sizes of both studies,
incomparabilities between the subject samples or fMRI paradigms
(hand or elbow or shoulder movements were used by Feydy et al.)
may simply account for divergent results.
Widespread activation in subcortical stroke, similar to our data,
was previously described in patients with striatocapsular infarction
and complete motor recovery (Weiller et al., 1992). In this study,
activation during finger-to-thumb opposition movement was found
in ipsilateral premotor, perisylvian (insula, BA 40) and mesial
regions (cingulate). Our data show similar activation patterns for
chronically impaired subjects performing elbow movement. Due to
these similarities between fully recovered and chronically impaired
subjects, one may be tempted to assume that brain activation is
unrelated to motor function. But this view may be too simplistic.
Other and in part unknown factors influence brain activation
during paretic movement as well as recovery itself. Our data
suggest that activation of regions, such as ipsilateral postcentral
gyrus, perisylvian, or mesial regions, is related to better motor
function; likely, more such areas exist (Small et al., 2002) and
might have been identified here if the sample included fully
recovered subjects.
Apart ipsilateral postcentral gyrus recruitment, cortical stroke
patients presented different patterns of activation than subcortical
stroke. No activation was found in standard corticocerebellar
circuits (contralateral SMC, ipsilateral cerebellum). It seems
expected that patients with motor cortex lesions cannot activate
the motor cortex secondary to critical tissue loss. However,
activation in cortex adjacent to the stroke cavity has been
reported (Cramer et al., 1997). Our data clearly support this
observation (Fig. 2). These adjacent regions represent cortical
areas which, in the intact brain, belong to premotor, frontal, or
A.R. Luft et al. / NeuroImage 21 (2004) 924–935 933
postcentral (somatosensory) cortices. Which regions and to what
extent they are recruited in the individual patient, may depend on
lesion distribution. Movement-related activation of cortex adja-
cent to the stroke cavity may reflect the brain’s attempt to
compensate for lost cortical function. But, anatomically, these
areas have different projections as compared to M1, especially
regarding their corticocerebellar projection. Such projections may
or may not build (or become unmasked) during recovery. The
absence of cerebellar activation in our data suggests that these
projections are not established in chronically impaired cortical
stroke patients.
Despite inoperative physiological corticocerebellar circuits,
functional recovery occurred during the post-stroke period. Other
brain circuits than the standard, which is active in subcortical
stroke, may mediate motor function in cortical stroke patients.
Our candidate region is the ipsilateral postcentral gyrus. Interest-
ingly, cortical as well as (to a lesser degree, Fig. 2) subcortical
stroke patients showed activation in ipsilateral postcentral gyrus
which correlated with motor function. Several studies have
reported activation in the intact hemisphere ipsilateral to the
paretic limb (Cao et al., 1998; Chollet et al., 1991; Cramer et
al., 1997; Green et al., 1999; Honda et al., 1997; Marshall et al.,
2000; Nelles et al., 1999). Recruitment of ipsilateral hemisphere
regions may represent the brain’s attempt to compensate for lost
circuitry via transcallosal reorganization. However, the impor-
tance of ipsilateral recruitment during recovery from hemiparesis
is disputed (Herholz and Heiss, 2000). The presence of MEP
responses in the paretic limb upon stimulation of the intact
hemisphere correlated with poor motor function in several studies
(Hallett, 2001; Turton et al., 1996). Ipsilateral activation may be a
transient phenomenon being high early after stroke and then
decaying with progressive spontaneous recovery (Cuadrado et
al., 1999; Feydy et al., 2002; Marshall et al., 2000). Effective
rehabilitative therapy in chronically impaired stroke patients can
shift motor cortex activation from the ipsilateral to the contralat-
eral side (Carey et al., 2002). Our sample of chronically impaired
patients can be considered as poor recoverers. This group may
have ipsilateral recruitment because recovery remained incom-
plete. That ipsilateral activation is directly related to motor
function suggests that it is required to control the limb while it
is hemiparetic. In patients with full recovery, it may have
disappeared because contralateral areas have taken over. Our data
suggest that this phenomenon of reorganization may be a com-
mon feature of cortical and subcortical stroke.
Activation associated with non-paretic limb movement
Activation patterns associated with non-paretic limb movement
differed from healthy control only in subcortical stroke. In these
patients, activation of paretic and non-paretic movement was
similar, whereas in cortical stroke it is not (Fig. 2). Regions
differentially activated in subcortical non-paretic versus healthy
control movement were mainly in the ipsilateral hemisphere (S1,
M1, and perisylvian regions) and in bilateral cerebellum. Differ-
ences between activation patterns associated with paretic and
nonparetic movement have been described previously. Cramer et
al. (1997) observed decreased activation in sensorimotor cortex
during non-paretic hand movement. Changes in motor cortex
somatotopy that parallel spontaneous recovery have been identified
by transcranial magnetic stimulation of the intact hemisphere
(Traversa et al., 1997). Weiller et al. (1992) found increased
activation in the contralateral hemisphere (insula, prefrontal-BA
46, premotor cortices, BA 40, and anterior cingulate) in patients
with striatocapsular infarction moving their intact limb. Our data
identify the ipsilateral (lesioned) hemisphere as being distinctly
activated by non-paretic limb movement.
Several hypotheses are being proposed to explain alterations
in non-paretic activation patterns. Our data do not support the
notion that recruitment of intact hemisphere networks for paretic
limb movement affects non-paretic limb activation patterns:
cortical patients had stronger intact hemisphere recruitment dur-
ing paretic limb movement but no detectable difference to
controls when moving the non-paretic side. An alternative expla-
nation is that altered activation patterns reflect (minor) functional
impairments of the non-paretic limb (Colebatch and Gandevia,
1989; Jones et al., 1989). These impairments are thought to be
the consequence of transcallosal metabolic depression (Hoedt-
Rasmussen and Skinhoj, 1964). A third interpretation builds on
behavioral modifications adopted by disabled patients, for exam-
ple, the overuse of the intact limb. Accordingly, structural
plasticity can be observed in contralateral (intact) hemisphere of
rodents after unilateral cortical lesion (Jones and Schallert, 1994).
Why in subcortical stroke patients, mechanisms leading to reor-
ganization of the activation associated with non-paretic movement
are more prominant than in stroke patients, cannot be resolved
here. However, it should be mentioned in this context that Feydy
et al. (2002) did not find any differences in activation patterns of
non-paretic and healthy control movement.
Methodological issues and limitations
When comparing the present data to other studies, one has to
consider the movement paradigm implemented here. Most studies
investigate distal limb movements (Cramer et al., 1997; Feydy et
al., 2002; Small et al., 2002). Patients with chronic hemiparesis,
like those studied here, often have more difficulty with distal than
with proximal (elbow) movement. Elbow movement also allows
for a more precise definition of mechanical parameters of move-
ment such as plane, freedom, and angle of motion. Our proprietary
scaffold used inside the scanner was designed to keep these
parameters constant and comparable across subjects. Only the
movement range was slightly adjusted for some patients, who
could not move the elbow without shoulder/trunk coactivation.
Range was considered the parameter least likely to affect brain
activation as opposed to movement rate or complexity (Rao et al.,
1996; Schlaug et al., 1996; Wexler et al., 1997).
Comparing distal and proximal arm movement, fMRI studies in
healthy volunteers showed differential activation of primary motor
cortex (Lotze et al., 2000; Rao et al., 1995) and cerebellum (Grodd
et al., 2001). In a preliminary study, we compared the elbow
paradigm employed here to conventional finger tapping (Luft et
al., 2002). Elbow movement produced activation patterns across
motor regions, which were similar to those of finger movements.
Both movement types differed largely from knee movement.
Finger movement, however, showed a somewhat stronger activa-
tion of bilateral premotor cortex and contralateral M1, consistent
with the notion that distal movements underlie closer pyramidal
control. Differences related to motor cortex somatotopy were
confirmed.
Another limitation of the present study is that patient groups
included left and right hemispheric strokes. Sample sizes were not
sufficient to allow further stratification according to lesion side.
A.R. Luft et al. / NeuroImage 21 (2004) 924–935934
Therefore, potential differences between paretic or non-paretic
left- versus right-sided movement were not addressed in this
study.
Brain lesions were classified into only two categories. Most
likely, further differences can be identified with respect to the
spatial extent of the cortical (which functional regions are involved
in addition to sensorimotor cortex?) and subcortical lesion (which
nuclei or brainstem regions are involved?). Whether a stroke
involves only cortex or cortex plus subcortical areas is also likely
important. We computed a group GLM (fixed effects) including
patients with brainstem (n = 2) versus deep brain lesions (n = 9, see
Table 1). This GLM did not reveal any differences; therefore,
brainstem stroke was grouped together with deep brain lesions.
However, this analysis may be questioned due to the large
difference in sample size. More subjects are necessary to determine
the effects of lesion location on fMRI patterns associated with
paretic movement in more detail.
Conclusion
The present finding that paretic as well as non-paretic move-
ment-related brain activation differs between patients with cortical
and subcortical stroke, reflects the relationships between reorgani-
zation mechanisms and lesion location. Despite obvious differ-
ences, some phenomena occurring during reorganization
(ipsilateral postcentral recruitment) may coexist in cortical and
subcortical strokes. Nevertheless, stratification of patients accord-
ing to lesion location should be considered for studies investigating
recovery mechanisms. Specifically, selection and evaluation of
rehabilitative interventions may be best performed in location-
specific groups of stroke patients.
Acknowledgments
We thank Andrew Goldberg, Jim Boyd, Christina Stephenson
and Jill England for their support. Thanks also to Dr. John Sorkin
for his expert help with statistical analyses. We thank the F.M.
Kirby Center for Functional Brain Imaging, Kennedy Krieger
Institute (Baltimore, MD, USA), and its staff, especially Terry
Brawner, Dr. James Pekar, and Dr. Peter van Zijl. Dr. Luft is
supported by grants from Deutsche Forschungsgemeinschaft (Lu
748/2, 748/3). Funding for this study was obtained from the
National Institutes of Health (P60AG 12583 NIA University of
Maryland Claude D. Pepper Older Americans Independence
Center H133G010111) the Baltimore Department of Veterans
Affairs Geriatrics Research, Education and Clinical Center
(GRECC), Dr. Hanley is supported by 1RO1 NS 24282-08 and
the France-Merrick Foundation, the Johns Hopkins GCRC (grant
NCRR #MO1-00052), the National Center for Research Resour-
ces, MO1 RR-00052, and the Eleanor Naylor Dana Charitable
Trust.
References
Byrnes, M.L., Thickbroom, G.W., Phillips, B.A., Mastaglia, F.L., 2001.
Long-term changes in motor cortical organisation after recovery from
subcortical stroke. Brain Res. 889, 278–287.
Cao, Y., D’Olhaberriague, L., Vikingstad, E.M., Levine, S.R., Welch,
K.M., 1998. Pilot study of functional MRI to assess cerebral activation
of motor function after poststroke hemiparesis. Stroke 29, 112–122.
Carey, J.R., Kimberley, T.J., Lewis, S.M., Auerbach, E.J., Dorsey, L.,
Rundquist, P., Ugurbil, K., 2002. Analysis of fMRI and finger tracking
training in subjects with chronic stroke. Brain 125, 773–788.
Chen, C.L., Tang, F.T., Chen, H.C., Chung, C.Y., Wong, M.K., 2000. Brain
lesion size and location: effects on motor recovery and functional out-
come in stroke patients. Arch. Phys. Med. Rehabil. 81, 447–452.
Chollet, F., DiPiero, V., Wise, R.J., Brooks, D.J., Dolan, R.J., Fracko-
wiak, R.S., 1991. The functional anatomy of motor recovery after
stroke in humans: a study with positron emission tomography. Ann.
Neurol. 29, 63–71.
Colebatch, J.G., Gandevia, S.C., 1989. The distribution of muscular weak-
ness in upper motor neuron lesions affecting the arm. Brain 112 (Pt 3),
749–763.
Cramer, S.C., Nelles, G., Benson, R.R., Kaplan, J.D., Parker, R.A., Kwong,
K.K., Kennedy, D.N., Finklestein, S.P., Rosen, B.R., 1997. A functional
MRI study of subjects recovered from hemiparetic stroke. Stroke 28,
2518–2527.
Cuadrado, M.L., Egido, J.A., Gonzalez-Gutierrez, J.L., Varela-De-Seijas,
E., 1999. Bihemispheric contribution to motor recovery after stroke: a
longitudinal study with transcranial Doppler ultrasonography. Cerebro-
vasc. Dis. 9, 337–344.
Feydy, A., Carlier, R., Roby-Brami, A., Bussel, B., Cazalis, F., Pierot, L.,
Burnod, Y., Maier, M.A., 2002. Longitudinal study of motor recovery
after stroke: recruitment and focusing of brain activation. Stroke 33,
1610–1617.
Fries, W., Danek, A., Scheidtmann, K., Hamburger, C., 1993. Motor re-
covery following capsular stroke. Role of descending pathways from
multiple motor areas. Brain 116 (Pt 2), 369–382.
Fugl-Meyer, A.R., Jaasko, L., Leyman, I., Olsson, S., Steglind, S., 1975.
The post-stroke hemiplegic patient. 1. A method for evaluation of phys-
ical performance. Scand. J. Rehabil. Med. 7, 13–31.
Green, J.B., Bialy, Y., Sora, E., Ricamato, A., 1999. High-resolution EEG
in poststroke hemiparesis can identify ipsilateral generators during mo-
tor tasks. Stroke 30, 2659–2665.
Grodd, W., Hulsmann, E., Lotze, M., Wildgruber, D., Erb, M., 2001. Sen-
sorimotor mapping of the human cerebellum: fMRI evidence of soma-
totopic organization. Hum. Brain Mapp. 13, 55–73.
Hallett, M., 2001. Plasticity of the human motor cortex and recovery from
stroke. Brain Res. Brain Res. Rev. 36, 169–174.
Herholz, K., Heiss, W.D., 2000. Functional imaging correlates of recovery
after stroke in humans. J. Cereb. 20, 1619–1631.
Hoedt-Rasmussen, K., Skinhoj, E., 1964. Transneural depression of the
cerebral hemispheric metabolism in man. Acta Neurol. Scand. 40,
41–46.
Honda, M., Nagamine, T., Fukuyama, H., Yonekura, Y., Kimura, J., Shi-
basaki, H., 1997. Movement-related cortical potentials and regional
cerebral blood flow change in patients with stroke after motor recovery.
J. Neurol. Sci. 146, 117–126.
Jones, T.A., Schallert, T., 1994. Use-dependent growth of pyramidal neu-
rons after neocortical damage. J. Neurosci. 14, 2140–2152.
Jones, R.D., Donaldson, I.M., Parkin, P.J., 1989. Impairment and recovery
of ipsilateral sensory-motor function following unilateral cerebral in-
farction. Brain 112 (Pt 1), 113–132.
Lancaster, J.L., Woldorff, M.G., Parsons, L.M., Liotti, M., Freitas, C.S.,
Rainey, L., Kochunov, P.V., Nickerson, D., Mikiten, S.A., Fox, P.T.,
2000. Automated Talairach atlas labels for functional brain mapping.
Hum. Brain Mapp. 10, 120–131.
Liepert, J., Bauder, H., Miltner, W.H., Taub, E., Weiller, C., 2000. Treat-
ment-induced cortical reorganization after stroke in humans. Stroke 31,
1210–1216.
Lotze, M., Erb, M., Flor, H., Huelsmann, E., Godde, B., Grodd, W., 2000.
fMRI evaluation of somatotopic representation in human primary motor
cortex. NeuroImage 11, 473–481.
Luft, A.R., Smith, G.V., Forrester, L., Whitall, J., Macko, R.F., Goldberg,
A.P., Hanley, D.F., 2002. Comparing brain activation associated with
A.R. Luft et al. / NeuroImage 21 (2004) 924–935 935
isolated upper and lower limb movement across corresponding joints.
Hum. Brain Mapp. 17, 131–140.
Marshall, R.S., Perera, G.M., Lazar, R.M., Krakauer, J.W., Constantine,
R.C., DeLaPaz, R.L., 2000. Evolution of cortical activation during
recovery from corticospinal tract infarction. Stroke 31, 656–661.
Nelles, G., Spiekermann, G., Jueptner, M., Leonhardt, G., Muller, S., Ger-
hard, H., Diener, H.C., 1999. Reorganization of sensory and motor
systems in hemiplegic stroke patients. A positron emission tomography
study. Stroke 30, 1510–1516.
Nelles, G., Jentzen, W., Jueptner, M., Muller, S., Diener, H.C., 2001. Arm
training induced brain plasticity in stroke studied with serial positron
emission tomography. NeuroImage 13, 1146–1154.
Rao, S.M., Binder, J.R., Hammeke, T.A., Bandettini, P.A., Bobholz, J.A.,
Frost, J.A., Myklebust, B.M., Jacobson, R.D., Hyde, J.S., 1995. Soma-
totopic mapping of the human primary motor cortex with functional
magnetic resonance imaging. Neurology 45, 919–924.
Rao, S.M., Bandettini, P.A., Binder, J.R., Bobholz, J.A., Hammeke, T.A.,
Stein, E.A., Hyde, J.S., 1996. Relationship between finger movement
rate and functional magnetic resonance signal change in human primary
motor cortex. J. Cereb. Blood Flow Metab. 16, 1250–1254.
Rijntjes, M., Weiller, C., 2002. Recovery of motor and language abilities
after stroke: the contribution of functional imaging. Prog. Neurobiol.
66, 109–122.
Rossini, P.M., Caltagirone, C., Castriota-Scanderbeg, A., Cicinelli, P., Del
Gratta, C., Demartin, M., Pizzella, V., Traversa, R., Romani, G.L.,
1998. Hand motor cortical area reorganization in stroke: a study with
fMRI, MEG and TCS maps. NeuroReport 9, 2141–2146.
Schlaug, G., Sanes, J.N., Thangaraj, V., Darby, D.G., Jancke, L., Edel-
man, R.R., Warach, S., 1996. Cerebral activation covaries with move-
ment rate. NeuroReport 7, 879–883.
Seitz, R.J., Hoflich, P., Binkofski, F., Tellmann, L., Herzog, H., Freund,
H.J., 1998. Role of the premotor cortex in recovery from middle cere-
bral artery infarction. Arch. Neurol. 55, 1081–1088.
Shelton, F.N., Reding, M.J., 2001. Effect of lesion location on upper limb
motor recovery after stroke. Stroke 32, 107–112.
Small, S.L., Hlustik, P., Noll, D.C., Genovese, C., Solodkin, A., 2002.
Cerebellar hemispheric activation ipsilateral to the paretic hand
correlates with functional recovery after stroke. Brain 125,
1544–1557.
Traversa, R., Cicinelli, P., Bassi, A., Rossini, P.M., Bernardi, G., 1997.
Mapping of motor cortical reorganization after stroke. A brain stimula-
tion study with focal magnetic pulses. Stroke 28, 110–117.
Turton, A., Wroe, S., Trepte, N., Fraser, C., Lemon, R.N., 1996. Contrala-
teral and ipsilateral EMG responses to transcranial magnetic stimulation
during recovery of arm and hand function after stroke. Electroencepha-
logr. Clin. Neurophysiol. 101, 316–328.
Ward, N.S., Brown, M.M., Thompson, A.J., Frackowiak, R.S., 2003. Neu-
ral correlates of outcome after stroke: a cross-sectional fMRI study.
Brain 126, 1430–1448.
Weiller, C., Chollet, F., Friston, K.J., Wise, R.J., Frackowiak, R.S., 1992.
Functional reorganization of the brain in recovery from striatocapsular
infarction in man. Ann. Neurol. 31, 463–472.
Weiller, C., Ramsay, S.C., Wise, R.J., Friston, K.J., Frackowiak, R.S.,
1993. Individual patterns of functional reorganization in the human
cerebral cortex after capsular infarction. Ann. Neurol. 33, 181–189.
Wexler, B.E., Fulbright, R.K., Lacadie, C.M., Skudlarski, P., Kelz, M.B.,
Constable, R.T., Gore, J.C., 1997. An fMRI study of the human cortical
motor system response to increasing functional demands. Magn. Reson.
Imaging 15, 385–396.
Wolf, S.L., Catlin, P.A., Ellis, M., Archer, A.L., Morgan, B., Piacentino, A.,
2001. Assessing wolf motor function test as outcome measure for re-
search in patients after stroke. Stroke 32, 1635–1639.