Article Text
PDF
Abstract
Background and purpose This study evaluated the safety and feasibility of targeted epidural cortical stimulation delivered concurrently with intensive speech–language therapy for treatment of chronic non-fluent aphasia.
Methods Eight stroke survivors with non-fluent aphasia received intensive behavioural therapy for 3 h daily for 6 weeks using a combination of articulation drills, oral reading and conversational practice. Four of these participants (investigational participants) also underwent functional MRI guided surgical implantation of an epidural stimulation device which was activated only during therapy sessions. Behavioural data were collected before treatment, immediately after treatment and at 6 and 12 weeks following termination of therapy. Imaging data were collected before and after treatment.
Results Investigational participants showed a mean Aphasia Quotient change of 8.0 points immediately post-therapy and at the 6 week follow-up, and 12.3 points at 12 weeks. The control group had changes of 4.6, 5.5 and 3.6 points, respectively. Similar changes were noted on subjective caregiver ratings. Functional imaging suggested increased consolidation of activity in interventional participants.
Conclusions Behavioural speech–language therapy improves non-fluent aphasia, independent of cortical stimulation. However, epidural stimulation of the ipsilesional premotor cortex may augment this effect, with the largest effects after completion of therapy. The neural mechanisms underlying these effects are manifested in the brain by decreases in the volume of activity globally and in particular regions. Although the number of participants enrolled in this trial precludes definitive conclusions, targeted epidural cortical stimulation appears safe and may be a feasible adjunctive treatment for non-fluent aphasia, particularly when the aphasia is more severe.
Statistics from Altmetric.com
Introduction
Over one million people in the USA have aphasia following stroke, and this has a major impact on social, vocational and recreational activities. Aphasia therapy has demonstrable value to patients,1 despite the lack of the strongest medical evidence of efficacy.2
Current clinical approaches are behavioural, and typically not based on biological rationales. However, recent research has provided insight into biological factors that could impact on recovery.3 The basis of these interventions is the long-lasting functional and structural plasticity of both intact perilesional tissue and areas remote from the injury. In general, cortical representations are highly dynamic and change in response to administration of pharmacological agents,4 5 electrical stimulation6 7 and behaviour.8 9 Importantly, the efficacy of biological interventions seems to depend on the presence of concomitant behavioural training,4 and its promise appears to be as an adjunct to such training.3 5
In animal models of stroke, low levels of cortical stimulation in combination with rehabilitative training can both increase the area of cortical movement representations and significantly enhance motor recovery. Furthermore, in human stroke subjects with hemiparesis, recent efforts to perform non-invasive cortical stimulation with transcranial direct cortical stimulation (tDCS)10 11 or transcranial magnetic stimulation (TMS) in concert with behavioural interventions12 have also been promising. In addition, epidural cortical stimulation to the primary motor cortex in conjunction with hand rehabilitation appears safe and possibly beneficial in upper limb motor function.13
In aphasia, non-invasive cortical stimulation with repetitive TMS or tDCS has been shown in preliminary studies to improve naming in select patients, even without concomitant behavioural stimulation.14 15 However, neither technique is capable of providing high frequency stimulation to the cerebral cortex, which is required for excitatory effects. Repetitive TMS has been applied at relatively low frequency to the contralesional cortex, with transcallosal effects on the injured hemisphere. tDCS approaches have used either ipsilesional excitatory stimulation (anodal) or contralesional inhibitory stimulation (cathodal) but with very low levels of spatial specificity and a narrow population of target neurons. Epidural cortical stimulation, although a more invasive approach, permits high frequency ipsilesional stimulation of high spatial specificity to targeted neuronal populations. Such stimulation results in a robust facilitatory effect on specific regions of interest, an effect that has been shown in animal model systems to induce plastic neuronal changes.8 16 17 At a cellular level, such plasticity depends on concomitant sensory input,18 19 and thus the present study, unlike the transcranial stimulation studies, uses concomitant behavioural training.
We hypothesised that for individuals with non-fluent aphasia, targeted epidural cortical stimulation to the left ventral premotor cortex would enhance the therapeutic effects of behavioural speech–language therapy. The left ventral premotor cortex was selected because of its known involvement in articulatory planning.20 21 Moreover, recent evidence suggests that mirror neurons in this area respond to the observation of mouth actions and may play a role in early language processing.22 In this article, we report the results of a first study to assess the safety and feasibility of such a therapeutic programme.
Methods
This study was approved by the US Food and Drug Administration under an Investigational Device Exemption. Ethics and administrative approval was also obtained from the institutional review boards of Northwestern University and the University of Chicago.
Four investigational and four control subjects participated. They all had a single ischaemic infarction in the left carotid artery distribution at least 12 months prior to enrolment and were classified clinically as having a non-fluent aphasia with slow agrammatic output and apraxia of speech. They were all premorbidly right handed, spoke English as their primary language and had at least a 12th grade education. Individuals with seizure disorders, haemorrhage or depression (by Center for Epidemiologic Studies Depression Scale23) were excluded, as were those who failed to activate the left lateral premotor cortex during functional MRI (fMRI) to several speech related tasks (see below).
The study was a single blind, randomised controlled trial. Since the recruitment process was spread over a period of time and the sample size was small, the first four participants who met the inclusion criteria were randomly assigned to investigational or control groups (ie, participants 001, 002, 005, 008); the next four eligible participants were matched to one of the already randomised participants by aphasia severity using the Aphasia Quotient (AQ) of the Western Aphasia Battery (WAB)24 and assigned to the opposite group. Table 1 shows the primary characteristics of the enrolled participants. Note that participants are ordered in matched pairs according to the severity of aphasia, as determined by the AQ of the WAB.
Participant characteristics
In addition to the WAB, participants were evaluated with the National Institutes of Health Stroke Scale,25 the Box and Block Test26 and selected subtests of the Behavioural Inattention Test.27 Family members completed the Communicative Effectiveness Index (CETI).28
Participants also underwent task dependent BOLD fMRI at 3 T, using T2* gradient echo spiral acquisition with a standard head coil.29 Tasks performed included: (a) oral imitation of a woman in a video producing single syllables, ‘ta’, ‘pa’, ‘tha’ and ‘ka’; (b) observation of the video production of the same syllables; and (c) oral reading of sentences. A volumetric T1 weighted scan (120 axial slices, 1.5×0.938×0.938 mm resolution) was acquired and averaged to provide high resolution images on which to identify anatomical landmarks and onto which functional activation maps could be superimposed.
For the functional scans, 29 images covering the whole brain were collected every 1.5 s in the axial plane (TR=1.5; echo time=24; flip angle=71). Functional resolution was 3.75×3.75×5.00 mm. Detailed fMRI image processing methods are included in the appendix.
By using a standard parcellation scheme,30 we computed the mean per cent signal change and total volume of activation in the premotor cortex (including the pars opercularis of the inferior frontal gyrus), superior temporal gyrus and occipito-temporal cortex, regions thought to be particularly important in speech production, language comprehension and reading, respectively.
Initial screening of participants and placement of the epidural grid were based on the intersection of activation (common areas of activation) between the imitation and observation tasks (motor resonance), and between the imitation and oral reading tasks (articulation). Participants with activation in the ventral portion of the precentral gyrus or sulcus (surface anatomy corresponding to Brodmann Area 6lv) in either of these two intersection maps were considered eligible.
Participants randomised to surgical implantation of the cortical stimulation device were admitted to the neurosurgical service of the University of Chicago Hospitals. In the operating room, a craniotomy flap was created using a neuro-navigational technique and sited over a region of the left premotor cortex determined by fMRI. An investigational epidural 2×3 grid electrode array, 2.6×2.7 cm in total area (Northstar Neuroscience Inc, Seattle, Washington, USA) was implanted on the dura overlying the ventral precentral gyrus at the site closest to the fMRI activation site. The craniotomy flap was replaced and the electrode lead tunnelled to a subclavicular site and connected to an investigational implanted neurostimulator (Northstar Neuroscience). At the end of therapy, a second surgical procedure was performed to remove the implant.
Both investigational and control participants underwent the identical intensive daily outpatient therapy for 3 h/day, 5 days a week, for 6 weeks. Therapy focused on language production and consisted of: (a) apraxia drills (30 min) that emphasised repetitive articulation of syllables and words; (b) confrontation naming of black and white line drawings (30 min) with hierarchical semantic and phonological cuing provided by a trained speech–language pathologist; (c) unison reading aloud of sentences (60 min); and (d) conversational practice (60 min). Investigational participants received bipolar cortical stimulation at 50 Hz during each therapy session. Stimulation intensity was set at 6.5 mA for all subjects except subject 002 for whom it was set at 4.75 mA (ie, 50% of the level that disrupted language function or induced movement).
The primary endpoint of the study was safety, measured by adverse events or neurological decline. Additional outcome measures assessed aspects of therapeutic efficacy and neurobiological change. The main therapeutic outcome measure was the absolute change on the WAB-AQ from before to after treatment, with a 5 point improvement considered success (this value exceeds the SE of the test). Testing was performed by a speech–language pathologist blind to the participant's group assignment, and with all participants wearing complete head coverings to obscure possible evidence of recent surgery. WAB testing was also repeated 6 weeks and 12 weeks following the end of treatment. Another therapeutic outcome measure was the caregiver rating on the CETI.
We quantified the differences between pre- and post-treatment fMRI scans to characterise the physiology of the stroke recovery process. For this analysis, we focused on the activation changes for the syllable imitation task in: (a) the whole brain; (b) each hemisphere; (c) the left ventral premotor cortex (including pars opercularis of the inferior frontal gyrus); and (d) the left superior temporal gyrus.
Results
Safety
Both the surgery and therapy appear safe. There were no occurrences of wound infection, and postoperative pain reported within the first 48 h was well tolerated. No adverse events affected the course of therapy or had a long term impact on patient well being. No seizures occurred. One participant complained of tingling around the implanted neurostimulator. Treatment was suspended for 1 day and symptoms spontaneously remitted. Neurological function assessed with the National Institutes of Health Stroke Scale, the Box and Blocks Test and subtests of the Behavioural Inattention Test remained stable in all participants. Stability on these tests also serves to rule out non-specific effects of both the speech–language therapy and the cortical stimulation.
Language outcomes
On the primary outcome measure, investigational participants showed a mean WAB-AQ change from baseline of 8.0 points at both post-therapy and 6 week follow-up endpoints, and 12.3 points at the 12 week follow-up (figure 1). The control group had a change of 4.6 points post-therapy, 5.5 points at the 6 week follow-up and 3.6 points at 12 weeks. There were no data for one control participant at the 12 week follow-up. Since this was an initial feasibility study with a small number of samples, there was insufficient power to detect significant differences at any time point.
Mean change in Aphasia Quotient of the Western Aphasia Battery (WAB-AQ) score for the investigational (blue) and control (red) groups, showing a group difference, particularly at the 12 week maintenance time point. Note that the control moderate–severe participant was lost to follow-up at 12 weeks.
In the matched patient pairs, three of the four investigational participants showed a pre- to post- treatment WAB-AQ change that was greater than that of the matched control participant. Investigational participants from most impaired to least impaired had increases of 15.0, 6.45, 7.0 and 3.45 points; matched control participants had changes of 2.65, 11.9, 3.95 and −0.15 points. At the 12 week follow-up, the most severe participant in the investigational group had an increase of 20.3 points. This compares with a change of only −0.5 for the most severe participant in the control group. Although neither the investigational nor control participants with mild–moderate aphasia achieved a 5 point change on the WAB-AQ, the investigational participant had greater changes at all time points. The paired results are shown in figure 2. The pretreatment, post-treatment, 6 and 12 week follow-up scores in the spontaneous speech, auditory comprehension, repetition and naming subscales of the WAB-AQ are included in table 2.
Paired results on the primary behavioural outcome measure, the change in Aphasia Quotient of the Western Aphasia Battery immediately post therapy and at 6 and 12 weeks after the end of therapy.
Aphasia Quotient and subscale scores obtained on the Western Aphasia Battery at each test time
The CETI was used to assess subjectively the utility of communication. The mean change in caregiver ratings mirrors the trends seen with the WAB-AQ, with communication changes post-treatment and during the maintenance period being greater in the investigational than the control group. Investigational group changes were 17.1, 33 and 30.8 points at post-treatment, 6 weeks and 12 weeks, respectively. Respective control group changes were 10.9, 16.3 and 19.6 points.
Neurobiological change
All eight participants underwent anatomic and functional MRI prior to therapy but the control participant with a moderately severe aphasia did not complete the session and did not return for imaging after completion of therapy. Thus we have complete data on three matched pairs of participants: mild–moderate, moderate and severe.
Figure 3 shows the structural imaging data. Although there was a consistent pattern of somewhat larger lesion volumes in the control group than in the investigational group, this was not significant (p=0.26). Furthermore, there was no difference in lesion volume in the inferior frontal pars opercularis/ventral premotor region of interest (the motor portion of ‘Broca's area’ and adjacent premotor cortex) between groups (p=0.90) but there was a weak trend (p=0.16) towards increased lesion volume in the superior temporal gyrus for the control group.
Cortical surface representation of the brain lesions for all matched pairs of participants.
The functional imaging results of interest are the changes in functional brain activation from pre- to post-therapy, and their relationship to behaviour change. To examine the brain–behaviour correlation, we grouped investigational and control participants, and correlated change in overall brain activation on the syllable imitation task to change on the WAB-AQ from pre-therapy to the final post-test (12 weeks after the end of therapy). This analysis demonstrated a strong inverse correlation (R2=0.6159). Decreases in whole brain activation thus correlate with increases in positive change on the primary outcome measure.
In whole brain activation, the ipsilesional left hemisphere showed a decrease from pre- to post-therapy on all functional tasks, and this decreased more in the investigational group than in the control group. Activation changes in the contralesional right hemisphere were much more task and region dependent.
The different participant pairs did not contribute identically to these values. In particular, the mild–moderate participants tended to show activation increases, and the moderate and severe pairs tended to show activation decreases. These changes differed by brain region and by functional task. Figure 4 shows these data for the syllable imitation task. In addition to showing hemispheric data separately, we also chose two regions of interest based on traditional models of language performance and on the stimulation site. These regions were the lateral premotor region (also including the pars opercularis of the inferior frontal gyrus), important for speech production and the site of stimulation (on the left), and the superior temporal gyrus and sulcus, most relevant for comprehension. In the mild–moderate group, activation increased in both regions of interest bilaterally. In the moderate and severe groups, hemispheric activation decreased for the intervention group and increased for the control group. This was also true for the temporal region but was inconsistent in the lateral premotor region, which included the stimulation site.
Functional imaging data for the three matched pairs of participants, mild–moderate (top), moderate (middle) and severe (bottom), while performing the audiovisual syllable imitation task. Data show per cent changes in volume of activation (axis on the left) between the two scans, before therapy to after therapy. Note that in the two more severely impaired pairs, the control participants tended to have increased whole brain activation over the course of therapy but the intervention participants had decreased activation. On the functional MRI scans, orange denotes increased activation following therapy while blue denotes decreased activation following therapy. LPMC, lateral premotor cortex; STG/STC, superior temporal gyrus/sulcus.
Discussion
Clinical treatments for aphasia are behavioural and aim to re-educate the damaged speech and language system. Although studies of biological interventions are increasing, this is the first study to use a surgical intervention for treatment of aphasia from ischaemic stroke.
The invasive nature of the present study demanded an examination of safety. Safety results are promising, with no notable adverse events. This safety profile is better than a previous study of cortical electrical stimulation for treatment of hemiparesis after stroke,31 in which two infectious complications occurred. The authors attribute this complication to the use of externalised leads, unlike the internalised leads used here.
Although the study was not of sufficient power to permit definitive statements about efficacy, some notions emerge from comparison of each pair of investigational and control participants, matched on aphasia severity.
All participants received intensive speech–language therapy, and thus we expected language improvement for all participants, especially considering the benefits of high intensity therapy.32 We also anticipated cortical electrical stimulation to enhance this effect.
Indeed, we found suggestive group level evidence supporting this hypothesis, with more marked responses in some pairs than others. The most marked change occurred in the participant with the most severe aphasia. The investigational patient showed much more improvement than the control patient, and continued to make gains following termination of treatment, with the gap between the participants widening at the 12 week follow-up. This investigational patient was older than the control patient and was more chronic (40.8 months vs 13.1 months) but had a slightly smaller stroke (∼15%), which involved more of the posterior temporal and parietal regions.
Investigational participants with moderate and moderate–severe aphasia both improved more than 5 points on the WAB-AQ, and more than their matched control cases. Neither of the mild–moderate participants reached a 5 point change on the WAB-AQ.33 However, both were already at a relatively high level, decreasing the sensitivity of the measure for indicating improvement. Furthermore, the investigational participant with mild–moderate deficits made more gains than the control participant.
These data suggest that the therapy might be particularly helpful for more severely impaired individuals with aphasia. While it cannot be claimed that these results will generalise to other participants of comparable severity, they warrant further investigation, at least in participants with more severe aphasias.
The functional imaging data also show particular responsiveness to cortical stimulation in the more severe participants. The two more severely impaired participant pairs who underwent imaging showed decreased brain activation over the course of therapy. Such decreasing activation in fMRI studies of recovery have generally been associated with better behavioural recovery, thought to be due to more efficient processing from better circuit reorganisation. This general notion was confirmed in these data by the correlation between the change in activation volume and behavioural scores, across all participants (independent of group).
These outcome results must be interpreted cautiously in view of the small number of participants. In addition, although participants in the investigational and control groups were matched on severity of aphasia, they differed on other characteristics that potentially may impact on therapy outcome, including aphasia chronicity, age, lesion location and lesion size. In most patient pairs, the overall per cent of damage to the left hemisphere was less in the investigational group. Examination of lesion volume in specific regions of interest also showed differences but no specific trends could be found for one or other of the patient groups. Although global differences in lesion volume have not been associated with differences in recovery,34 there does seem to be a pattern in regional injury, both in language35 and motor system.36
In language, damage to ‘Wernicke's area’ or the left posterior superior temporal lobe and inferior parietal lobe has been associated with overall worse prognosis.35 Among our participants, half had damage to parts of this area, and they were equally divided between intervention and control groups. However, lesion volumes did differ (but below a level of statistical significance) between the two groups, with larger temporoparietal (Wernicke's region) injury in the control group. Future studies with larger enrolments will better sort out the importance of this factor.
The imaging findings in this study suggest that better aphasia recovery is associated with a decrease in total brain activation during language tasks. This is consistent with findings related to brain activation patterns in long term learning, but not short term learning,37 in which activation appears to expand in focal regions (eg, primary motor cortex in a motor learning task) prior to decreasing.38 Learning over a longer term, such as that occurred in the present study, is typically associated with decreased brain activation in many regions, including primary regions, but also in those associated with executive or cognitive control functions (eg, prefrontal regions) and with emotion and stress (eg, limbic regions).39 The presumed mechanism for this reduction in brain activation is decreased reliance on conscious mechanisms, decreased frustration and effort, and increased automatisation of the skill.40
In the present study, participants receiving epidural cortical stimulation were more likely to have decreases in functional activation throughout the brain than those who did not receive such stimulation. Furthermore, the moderate and severe participants were more likely to have decreases in regions critical to language performance. Taken together, these data suggest that cortical epidural stimulation promotes a type of neural change that is most consistent with long term learning and reorganisation of neural circuits than with less robust, shorter term changes. These speculations must be confirmed by larger studies, including more participants, additional behavioural and imaging measures, and a sham stimulation arm. While placebo effects cannot be ruled out, results of this preliminary study indicate that excitatory ipsilesional high frequency epidural cortical stimulation is a potentially safe and feasible adjunctive intervention for individuals with chronic non-fluent aphasia from a single left hemisphere ischaemic infarction that spares the ventral premotor cortex.
Acknowledgments
The authors acknowledge the support of Northstar Neuroscience, who funded this study. The authors note that they have no other financial relationship with Northstar Neuroscience, and have not received any personal compensation from the company. Additional support for preliminary studies and specialised analyses has come from Mr William Rosing, whose support is greatly appreciated. The authors would also like to acknowledge the technical assistance of Edie Babbitt, Robert Fowler, Rosalind Hurwitz, Jaime Lee, Nameeta Lobo and Roma Siugzdaite. Special thanks go to Drs Richard Harvey and Elliot Roth for helping to facilitate the study administratively and clinically. Finally, we acknowledge Dr Doug Sheffield, Director of New Technologies at Northstar Neuroscience during the execution of this study, for his supportive sponsorship.
Appendix Detailed fMRI methods
Participants underwent task dependent BOLD fMRI at 3 T using T2* gradient echo spiral acquisition with a standard head coil.1 Tasks performed included: (a) oral imitation of a woman in a video producing single syllables, ‘ta’, ‘pa’, ‘tha’ and ‘ka’; (b) observation of the video production of the same syllables; and (c) oral reading of sentences. A volumetric T1 weighted scan (120 axial slices, 1.5×0.938×0.938 mm resolution) was acquired and averaged to provide high resolution images on which to identify anatomical landmarks and onto which functional activation maps could be superimposed. For the functional scans, 29 images covering the whole brain were collected, one every 1.5 s in the axial plane (TR=1.5; echo time=24; flip angle=71). Functional resolution was 3.75×3.75×5.00 mm.
Images were spatially registered in three-dimensional space by Fourier transformation of each of the time points and corrected for head movement, using the AFNI software package.2 Data were then smoothed with a Gaussian 6 mm FWHM filter to decrease spatial noise. The haemodynamic response function for each condition was established via regression for the 15 s following the stimulus presentation on a voxel-wise basis. There were separate regressors for each of the three experimental conditions. Additional regressors were the mean, linear and quadratic trend components, as well as the six motion parameters in each of the functional runs. A linear least squares model was used to establish a fit to each time point of the haemodynamic response function for each of the three conditions. The response in each voxel was scaled to the mean of the voxel's signal during the study for each participant.
To establish an objective level of ‘activation’, a Monte Carlo Simulation was run to establish an individual voxel significance level and cluster size threshold.3 This procedure established that using a cluster size of 3 voxels and a cluster connection radius of 5.2, an individual voxel p value of 3e-05 would establish the desired whole brain α (p<=0.05). To create functional images, these parameters were used to form data sets containing significant data satisfying the cluster criteria for each of the two scans (before and after) for each of the subjects. Areas that did not have sufficient signal to noise, as quantified by a map of the mean of the blurred time series divided by the SD of the blurred time series,4 were not included in the analysis.
Next, we used the FreeSurfer software package to create surface representations of each participant's anatomy by inflating each hemisphere of the anatomical volumes to a surface representation and aligning it to a template of average curvature.5,6 SUMA was then used to import the surface representations and project the functional data from the three-dimensional volumes onto the two-dimensional surfaces.7 This procedure enables more accurate reflection of the individual data at the group level.8 Statistical analyses were conducted on the surface data using the AFNI/SUMA software package.
In order to use the FreeSurfer software on these images from patients with stroke, we first used a procedure we call ‘virtual brain transplantation’,9 which we have successfully used previously on patients with aphasia and cortical strokes.10 Using the AFNI software, a mask of the lesion was drawn. Next the brain was flipped left to right. Then this flipped version was used to ‘fill in’ the lesion mask, thus ‘filling in’ the lesion with ‘transplanted virtual tissue’ from the non-lesioned hemisphere. This permitted the brain inflation algorithm to work with a minimum of anatomical error. Finally, the contrast between the edges of this transplanted tissue and the rest of the hemisphere was reduced, effectively ‘suturing’ the two tissues together, by using image morphing software.
References—Detailed Methodology
Noll DC, Cohen JD, Meyer CH, et al Spiral K-space MRI of cortical activation. J Magn Reson Imaging 1995;5:49–56.
Cox RW. AFNI: software for analysis and visualisation of functional magnetic resonance neuroimages. Comput Biomed Res 1996;29:162–73.
Forman SD, Cohen JD, Fitzgerald M, et al Improved assessment of significant activation in functional magnetic resonance imaging (fMRI): use of a cluster-size threshold. Magn Reson Med 1995;33:636–47.
Parrish TB, Gitelman DR, LaBar KS, et al Impact of signal- to-noise on functional MRI. Magn Reson Med 2000;44:925–32.
Dale AM, Fischl B, Sereno MI. Cortical surface-based analysis. I. Segmentation and surface reconstruction. Neuroimage 1999;9:179–94.
Fischl B, Sereno MI, Dale AM. Cortical surface-based analysis. II: Inflation, flattening, and a surface-based coordinate system. Neuroimage 1999;9:195–207.
Saad ZS, Reynolds RC, Argall B, et al SUMA: An interface for surface-based intra- and inter-subject analysis with AFNI. International Symposium on Biomedical Imaging: Macro to Nano; 15–18 April 2004, Arlington, VA, USA: IEEE, 2004:1510–13.
Argall BD, Saad ZS, Beauchamp MS. Simplified intersubject averaging on the cortical surface using SUMA. Hum Brain Mapp 2006;27:14–27.
Wier HY, Hasson U, Skipper JI, et al Virtual brain transplantation: an approach for accurate registration and parcellation of brain-injured patients. In: Corbetta M, ed. Twelfth annual meeting on human brain mapping, Florence, Italy, 2006.
Cherney LR, Small SL. Task-dependent changes in brain activation following therapy for nonfluent aphasia: Discussion of two individual cases. J Int Neuropsychol Soc 2006;12:828–42.
References
Footnotes
Funding Northstar Neuroscience Corporation, Seattle, Washington, USA, funded the study but the authors have no financial interest in the company and serve in no consultative capacities with this company.
Competing interests None.
Ethics approval This study was conducted with the approval of the US Food and Drug Administration under an Investigational Device Exemption and by the institutional review boards of both Northwestern University and the University of Chicago.
Provenance and peer review Not commissioned; externally peer reviewed.