STROKE: Functional Magnetic Resonance
Title: FUNCTIONAL MAGNETIC RESONANCE IMAGING OF THE HUMAN MOTOR CORTEX BEFORE AND AFMR WHOLEHAND AFFERENT ELECTRICAL STIMULATION
S. Golaszewski1, Ch. Kremser1 M. Wagner1, S. Felber1,2 F. Aichnerl,2 and M. M. Dimitrijevic3
From the Departments of 1Magnetic Resonance and 2Neurology, University
of Innsbruck, Austria and the 3Department Of Physical Medicine and Rehabilitation,
Baylor College of Medicine, Houston, Texas, USA
ABSTRACT. Electrical stimulation of the whole hand using a meshglove has
been shown to improve volitional movement of the hand and arm, and decrease
muscle hypertonia after hemispherical stroke in patients who have reached
a recovery plateau. The goal of this study was to investigate the effect
of stimulation of the nerve afferents of the hand on brain cortical activity
elicited by wholehand subthreshold stimulation for sensation in humans
with intact nervous systems. Brain cortical activity in 6 healthy subjects
(3045 years) was studied using blood oxygenation leveldependent functional
Magnetic Resonance Imaging during a test motor task, fingertothumb tapping
and after 20 minutes of meshglove stimulation of the resting hand prior
to performance of an identical motor task, to test the changes in the conditioned
motor task established after 20 minutes of meshglove stimulation. Fifteen
contiguous echoplanar sequences parallel to the bicommissural plane were
acquired for functional magnetic resonance. Postprocessing of image data
included correction of motion artefacts and calculation of correlation
coefficients between the signal intensity of pixels during rest and finger
tapping and a rectangular reference wave function. The functional Magnetic
Resonance Imaging examinations revealed a signal increase in the primary
and secondary motor and somatosensory areas when comparing the number of
activated pixels during test and conditioned motor tasks. Our preliminary
study indicated that change occurred in a definite pattern in the region
of the regional cerebral blood flow of the brain cortex after meshglove
wholehand stimulation at the subthreshold level for sensation. We assumed
that this increase in regional cerebral blood flow also reflected augmented
We have reported previously that impaired movement of the hand and arm and altered muscle tone of the affected side after hemispherical stroke lesions can be ameliorated by using a wire meshglove to stimulate the afferents of the hand (7). The novelty of this method is that the whole hand is the target of stimulation rather than only the cutaneous, cutaneousmotor and motor nerves, as is the case in functional and neuromuscular electrical stimulation (14, 19, 22). Meshglove stimulation generates (tonic) synchronous input to the brain below the conscious sensory threshold, probably due to depolarization of the large afferents of the whole hand. The beneficial effects of meshglove stimulation are the suppression of muscle hypertonia, augmentation of residual volitional movement and increased awareness of the affected hand (710).
On the basis of this finding, we developed the hypothesis that meshglove stimulation involved neurophysiological mechanisms triggered by externally induced kinesthetic input to posterior column nuclei, thalamus and cortical brain structures. We have speculated that the effect is mediated through an extended spinalbrain mechanism rather than being restricted to the segmental spinal response. In order to investigate whether wholehand meshglove stimulation at the subsensory threshold level for conscious sensation can elicit changes in cortical brain activity, we developed the present functional Magnetic Resonance Imaging (fMRI) study of human cortex activity before and after wholehand afferent electrical stimulation in a group of 6 healthy adult subjects.
The aim of this study was to investigate whether changes in regional cerebral blood flow (rCBF) could be demonstrated by fMRI after wholehand afferent electrical stimulation. fMRI based upon the blood oxygenation level dependent (BOLD) effect allows the identification of physiologically activated brain areas by means of local and transient magnetic resonance signal increases (21). The most accepted explanation is that a local decrease in deoxyhemoglobin concentration within the venous microcirculation will result in an increase of the magnetic resonance signal detected during brain activation (24, 25). Therefore, we applied fMRI to study brain cortical activity in healthy volunteers during volitional motor tasks before and after wholehand meshglove stimulation (28, 29).
In this report, we show that wholehand afferent electrical stimulation below the threshold for perception of sensation can significantly increase responsiveness in rCBF within several cortical regions of the contra and ipsilateral hemispheres.
MATERIALS AND METHODS
Our study was carried out in 6 healthy, right-handed mate volunteers (age
3045 years), who signed the written consent form. The study protocol was
approved by the local ethic committee. The subjects were asked to perform
a fingerto-thumbtapping on/off motor paradigm with the hand, which consisted
of consecutive forward and backward tapping of the second to the fifth
finger with the thumb, with a frequency of approximately 3 Hz. Each sequence
of finger tapping began with an auditory signal. The subjects were instructed
to position the tapping hand over the abdomen and keep their eyes closed
during the fMRI recording. Repeated measurements were made for each subject
on two different days. Furthermore a third identical study was performed
in the same subjects, but with sham stimulation, meaning that the subjects
were not aware that electrical stimulation was being applied.
All experiments were performed on a 1.5 Tesla whole body scanner (Magnetom VISION, Siemens, Germany) with an echoplanar capable gradient system (rise time 600 gsec, 25 mT/ms) and a circular polarized head coil (FoV = 250mm). For fMRI, we employed T2* weighted single shot echoplanar sequences (TRrrE/' = 1.64ms/64ms/90', matrix = 64 x 128, voxel dimension = 2.94 x 1.95 x 3 mm). As shown in Fig. 1, we acquired 15 slices parallel to the bicommissural plane (1, 20). Foam padding and a special helmet fixed to the head coil were used to restrict head motion. A series of 60 sequential images was acquired, each consisting of 10 images during rest alternating with 10 images during finger tapping.
We designed this study to investigate the conditioning effect of meshglove
stimulation on brain cortical activity elicited by the simple sequential
movement of a selfpaced tapping motor task (4, 31, 34). We applied meshglove
stimulation for 20 minutes to the relaxed hand, while the subject was removed
from the magnet. Thus, the first experiment involved a test motor task
(TMT) and the second a conditioned motor task (CMT).
The meshglove was connected to a twochannel stimulator (Medtronic Model 3128 Respond 11) with a common anode for output to the mesh glove and a pair of separated surface electrodes as cathodes just above the wrist, over the tendons of the extensors and flexors of the forearm. A train of stimuli of 50 Hz with a pulse width of 0.3 ins was used for subthreshold stimulation. The subthreshold level of stimulation was defined by decreasing the stimulus strength to a level at which the subject reported that the tingling sensation had disappeared. The amplitude for subthreshold stimulus was between 0.85 and 0.95 mA.
To correct involuntary head movements, the fMRI were realigned offline using the Woods et al. algorithm (36). To quantify changes in rCBF during cortical activation, we identified only those pixels with signal changes that corresponded to the temporal pattern of the motor task. This was accomplished by correlating the normalized imaging data from each voxel with a reference waveform (2, 3). In this study, we assumed that functional activity resembled a rectangular waveform. All pixels with a correlation less than r = 0.45 and pixel clusters of less than 4 pixels were excluded from further analysis. For quantitation of pixel intensity we used a twotailed paired Student's ttest (3, 32). The functional images were coloured according to positive values of the crosscorrelation coefficient (i.e. pixels in phase with the reference waveform), and were displayed on a scale from red (minimum) to yellow (maximum). The pixels not reaching the cutoff point were made transparent. These coloured functional images were then superimposed onto the original echoplanar sequence images. Then, the anatomical location of the activated areas was evaluated using coordinates by Talairach & Toumoux (35).
Table I summarizes the data obtained in all 6 volunteers studied for rCBF changes determined by the number of activated pixels. There is a consistent profile of changes between the studied brain cortical areas of the primary motor cortex, primary sensory cortex and cortical areas of higher order within the sensorimotor system (gyrus frontalis superior, gyrus frontalis medius, lobulus parietalis superior, and lobulus parietalis inferior). These findings were reproducible in the same subjects on another recording day (day 2). The repeated recordings revealed a similar magnitude of change between the first test and the conditioned motor task. Ipsilateral hemispheric cortical activity after meshglove wholehand stimulation, as expected, was much less pronounced.
Slices adjacent to the acquired 15 slices (Fig. 1) that passed the motor hand area and showed the main activation centers were evaluated. The numbers of activated pixels within these areas are summarized in Table I. In subject 1 (day 1), fMRI measurements during TMT showed 15 activated pixels within the contralateral gyrus precentralis, but 23 activated pixels within this same area during CMT. Within the ipsilateral hemisphere, we found 5 activated pixels for TMT and 8 for CMT within the gyrus precentralis. During the second examination, TMT and CMT revealed similar results.
Subjects 2, 5 and 6 showed the same contralateral and ipsilateral responses within the gyrus precentralis. Subjects 3 and 4 showed increased activity only within the contralateral gyrus precentralis during CMT. Within the primary somatosensory region in the gyrus postcentralis contralateral to the tapping fingers, we could see that each subject activated pixels during TMT (pixel range: 637) and CMT (pixel range: 945). Subjects 1, 2, 5 and 6 also showed ipsilateral activation at first in CMT during both recordings carried out on different days, whereas subjects 3 and 4 only showed contralateral activation.
Table 1. Activated pixel numbers during the fingertapping motor task, test motor task and after conditioning of the resting left hand before identical fingertapping conditioned motor task with 20 minutes of subthreshold, electrical wholehand meshglove stimulation. Number of activated pixels during the first day (day 1) and subsequent day (day 2) of the recording sessions from the cortical areas of the gyrus precentralis (GPrC) gyrus postcentralis (GPC) lobulus parietalis superior (LPs), lobulus parietalis inferior (LPi), gyrusfrontalis superior (GFs) and gyrusfrontalis medius (GFm)
Subjects 1-6 showed increased contralateral activation within the lobulus parietalis superior during CMT, and ipsilateral activation increased from no activity to 5 and 17 pixels. However, there was no activity in the lobulus parietalis inferior within the contra and ipsilateral hemispheres in response to TMT, but CMT measurements showed rCBF activity in both hemispheres in nearly all cases.
Within the gyrus frontalis superior, we observed contralateral TMT and CMT activation when subjects 1, 2, 4, 5 and 6 were examined on two different days. On the second day of examination, subject 2 showed both ipsilateral TMT and CMT activation. In subject 3, then was a response only during CMT on the second day Subject 6 showed ipsilateral activation to CMT in both recordings carried out on different days. The gyrus frontalis medius showed contra and ipsilateral response to TMT and CMT only in subject 4 (day 1), while contralateral activation to CMT was seen constantly in all subjects, and in all subjects there was an increase in the maximum intensity of the activated pixels after CMT.
In order to simplify the illustration of data presented in Table 1, we calculated the mean value of all recorded activated pixels of the contralateral (right) hemisphere within the cortical regions studied for the 12 measurements in all 6 subjects (Fig. 2). There was a significant change (p < 0.01) in rCBF between TMT and CMT in gyrus precentralis activity (Fig. 2).
When the same experimental procedure was performed in the same subjects, but with sham meshglove stimulation, there was no increase in rCBF activity of CMT. The results are summarized in Table II. Before and after sham stimulation, there was bilateral activation within the GPrC in subjects 1, 2, 5 and 6, and only contralateral. GPrC activation in subjects 3 and 4 during the two recordings carried out on different days. The lobulus parietalis inferior pattern of activation during sham stimulation was similar to the previously described gyrus precentralis pattern of activation. Activation within the lobulus parietalis superior was seen only contralaterally during sham stimulation.
The same was true for the gyrus frontalis superior The lobulus parietalis inferior and the gyrus frontalis medius showed no activation during sham measurements. There was no statistically significant increase in the number of pixels activated after sham stimulation (Fig. 3).
Fig. 1. Slice orientation parallel to bicommissural plane. The block of 15 slices covers the whole motor cortex. Slices 3 and 4 (from the top) pass through the motor hand area, which is the main site of activity during the finger‑tapping motor task with the contralateral (left) hand.
Figs. 4 and 5 illustrate functional image findings during TMT and applied mesh‑glove stimulation before CMT (Fig. 4) and applied sham stimulation prior to CMT (Fig, 5) by means of the coloured functional images superimposed on the original echo‑planar sequence images. Fig. 4 is composed of functional images from left to right of subject 1 (day 1), subject 3 (day 1) and subject 4 (day 1) during TMT (upper row) and CMT (lower row). Fig. 5 consists of functional images during sham stimulation before CMT in subject 4 (day 2), subject 5 (day 1) and subject 6 (day 1). The sequence, as in Fig. 4, is from left to right; TMT is illustrated in the upper row and CMT in the lower row.
The pixels in phase with the reference brain waveform are displayed on a scale from red (minimum) to yellow (maximum), and the pixels not reaching the cut‑off point are transparent. By comparing these two figures of functional image findings, we see an obvious difference in the rCBF changes after 20 minutes of subthreshold stimulation for sensory perception, and no changes in rCBF when sham mesh‑glove stimulation was applied for 20minutes prior to CMT.
Fig. 2. Mean of activated pixels of the right hemisphere in all 6 subjects studied during test motor task JMT) of the left hand before mesh‑glove stimulation, and during conditioned motor task (CMT) after mesh‑glove stimulation. A statistically significant increase of activation (p < 0.01) after mesh‑glove stimulation is found for the gyrus pre‑ and postcentralis (GPrC and GPoQ, superior and inferior parietal lobules (LPs and LPi) and superior and medial frontal gyrus (GFs and GFm).
During self‑paced simple finger movements, we found an increase in the mean of activated pixels and maximum number of pixels over the contralateral hemisphere within the gyrus precentralis, gyrus postcentralis, lobulus parietalis superior and gyrus frontalis superior, as well as some increase in cortical brain activity in the ipsilateral brain hemisphere within the gyrus postcentralis and lobulus parietalis superior. These findings were expected and had been reported by other investigators who studied the activity of human cortical motor areas during self-paced finger movements (6,23). When we applied 20minutes of conditioning stimulation to the afferents of the whole hand, we found an increase in cortical activity induced by CMT when compared to that elicited by TMT without preceding mesh‑glove stimulation. Thus, we concluded that mesh‑glove whole‑hand stimulation augments cortical brain activity elicited by a motor task in the specific contralateral and ipsilateral motor and sensory areas of the frontal and parietal lobes. Findings that such additional excitatory effect was absent when the sham paradigm was applied further confirmed the validity of the results presented.
In addition, we concluded that whole‑hand mesh-glove stimulation below the threshold for sensation is an active system, which depolarizes a certain population of kinesthetic afferents with cortical projections. There are definite experimental findings to confirm that muscle afferents of groups Ia, Ib, and II have short‑latency projections to the contralateral somatosensory cortex, particularly areas 3a and 4 (motor cortex) (13). The presence of cortical projections of large afferents was also confirmed in a PET study in which the fingers of healthy adult humans were vibrated so that it was possible to activate bilaterally the contralateral SI, the retroinsular cortex parietal operculum and SII (30).
By means of fMRI, we have demonstrated that afferent electrical whole-hand stimulation with a mesh glove leads to an increase in the responsiveness of the somatosensory system and motor cortices during a volitional motor task. We still need to determine the nature of the relationship between increased rCBF and neuronal cortical activity. fMRI and PET scan studies of the rCBF have been carried out to identify the cortical areas activated during electrical stimulation of the median nerve at the wrist (16). It was found that median nerve stimulation from 4020 Hz evoked a single focus of activation in the primary somatosensory area. This study indirectly supports our finding, although stimulus strength was above the sensory threshold and applied over the median nerve, in contrast to our whole-hand stimulation with stimulus below the sensory threshold.
Analysis of the results (Figs. 2 and 3) of fMRI measurements of mean activated pixels, and the maximum number of pixels, of the right and left hemispheres during TMT and CMT after stimulating the left hand showed that both parietal inferior lobules were profoundly activated during CMT. This ipsilateral increase of activity of the inferior parietal lobule was the result of additional kinesthetic input, since during TMT we did not record any inferior parietal lobule activity either ipsior contralaterally.
hand can be a rich source of kinesthetic input. The large afferents
from the hand's intrinsic muscles have high‑density muscle spindles
(18), a large number of joint receptors with corresponding large afferents;
and Golgi tendon organs, with a portion of the tendons within the hand,
but belonging to the forearm muscles (5). Further neurophysiological
research is necessary to determine the kind and size of the population
of hand afferents depolarized under the experimental conditions and
parameters of mesh‑glove stimulation (11). However, regardless of
which hand afferents such stimulation depolarizes, we can be certain
that there are afferents which are involved in proprioception, "which
refers to the sensing of the body's own movement" (27).
In this fMRI study, we described a definite pattern of changes in rCBF of the brain cortex after mesh‑glove stimulation. We expected that an increase in rCBF resulting from such a definite and consistent pattern of whole‑hand stimulation reflected an increase in the actual level of neuronal brain activity.
This externally induced input to the brain came from the resting hand, which was stimulated for a 20‑minute period, and the effect of this input was shown in the post-stimulation period during simple motor tasks. Besides the increase in cerebral brain activity in the pre‑ and post‑central gyrus, it is also important to notice the appearance of cortical activity of the inferior parietal lobule within both hemispheres. The inferior parietal lobule is the site of convergence of kinesthetic input containing premotor and visual information. During reaching and grasping with the hand, the parietal association cortex is involved in a series of sensorimotor transformations to convert the signal to the target location on the retina into a pattern of peripheral motor output signals to muscles, in order to move the hand to the target (15, 33). The results reported in this study encouraged us to expand our understanding about the modification of motor control of wrist extension elicited by mesh‑glove electrical afferent stimulation in stroke patients (10, 19, 22).
In future studies, it will be necessary to prove our findings in a larger population of subjects. Our current image processing method, which perhaps is not the most optimal, is based on magnetic resonance signal changes which are detected and correlated with a rectangular waveform. Our finding that particular cortical areas are not activated during the motor task does not exclude the possibility that functional activity occurs within such regions below threshold analysis. We applied the same threshold for all subjects in each recording session, but the question remained whether we should choose a different threshold for each subject and each experimental condition. Moreover, we did not attempt to control the level of alertness and attention, or carry out any specific preparatory procedure before starting the recordings.
Fig. 3. Mean of activated pixels of the right hemisphere in all 6 subjects during test motor task (TMT) of the left hand before mesh‑glove stimulation, and during conditioned motor task (CMT) after sham mesh‑glove stimulation. There is no statistically significant increase (p > 0.02) of activation for the gyrus pre‑ and post-centralis (GPrC and GPoQ, superior parietal lobule (LPs) and superior frontal gyrus (GFs).
Fig. 4. Functional images from left to right for 3 of the 6 subjects studied. The upper row (a) shows fMRI images during TMT and the lower row (b) fMRI images after CMT.
The regional areas of activation during TMT were constant: the contralateral
primary motor and somato-sensory cortices in all subjects were activated
in each of the fMRI experiments. The resultant activation maps are
consistent with those obtained from electrophysiological (26) and PET
(12) studies. Activation within the premotor, motor cortex and somatosensory
association areas occurred, but was less constant at a threshold level
of analysis with a positive correlation coefficient of r = 0.45.
We also examined the functional imaging data with lower threshold levels
of activity (r = 0.38 and 0.35), but it was very difficult to differentiate
elicited activity from background signal noise, since the changes were
less than 2% and below the analysis threshold of our study.
The anatomical location of the activated regions had an accuracy of approximately I mm, and was based on the brain atlas of Talairach & Tournoux (35). Intra‑ and inter-individual reproducibility was very good. With the echo‑planar imaging multi‑slice technique, we were able to cover the whole motor cortex to sample spatial activation data. We found a large activation pattern within two adjacent parallel slices with a distance of 3mm passing through the motor hand area, as determined by the electrophysiological mapping of the motor cortex of Penfield & Boldrey (26), fMRI (17) and positron emission tomography studies (12, 23).
In conclusion, our findings encouraged us to expand this unique study of human brain activity under the influence of selective kinesthetic input by means of noninvasive methods such as fMRI. The results of this study support our working hypothesis that whole‑hand continuous electrical stimulation at the subthreshold level for conscious sensation involves neuro-physiological mechanisms, which are activated by externally controlled kinesthetic input.
This work was supported by the Biomed2 Project PL 950870 of the European Community, and by a grant from the Kent Waldrep National Paralysis Foundation, Dallas, Texas, USA. The authors express their gratitude to Professor Milan R. Dimitrijevic for his continuous support and encouragement during the development of this study.
Accepted August 27, 1998
Address for offprints:
Dipl. Ing. Dr. Stefan Golaszewski
Department of Magnetic Resonance
University of Innsbruck
Download PDF files about Stroke Research:
NEW!! Clinical Neurophysiology