Brain activation patterns underlying upper limb bilateral motor coordination in unilateral cerebral palsy: an fNIRS study

Abstract

AIM

To explore cortical activation during bimanual tasks and functional correlates in unilateral cerebral palsy (CP).

METHOD

This cross-sectional study included eight participants with unilateral CP (six females, two males; mean age [SD] 20y 10mo [5y 10mo], 13y 8mo–31y 6mo) in Manual Ability Classification System levels II to III and nine age-matched participants with typical development (seven females, two males; mean age [SD] 17y 8mo [5y 7mo], 9y 4mo–24y 2mo). They performed bimanual symmetric squeezing (BSS) and bimanual asymmetric squeezing (BAS) tasks at 1Hz, and a pouring task with dominant hand (DPour) and a pouring task with non-dominant hand (NDPour) at 0.67Hz, all while a custom array of functional near-infrared spectroscopy (fNIRS) optodes were placed over their sensorimotor area. Mixed-effects were used to contrast groups, tasks, and hemispheres (corrected p-values [q] reported). Analysis of variance and t-tests compared performance measures across groups and tasks.

RESULTS

Participants with unilateral CP showed greater activation in both hemispheres during BAS (non-lesioned: q<0.001; lesioned: q<0.001), and in the lesioned hemisphere during BSS (q<0.001), DPour (q=0.02), and NDPour (q=0.02) than those with typical development. The lesioned hemisphere in unilateral CP showed more activity than the non-lesioned one (BSS: q=0.01; BAS: q=0.009; NDPour: q=0.04). During BAS, higher cortical activity correlated with more synchronous arm activation (r=0.79; p=0.02); activity lateralized towards the non-lesioned hemisphere correlated with better Pediatric Evaluation of Disability Inventory computer adaptive test scores (r=0.81; p=0.03).

INTERPRETATION

Results suggest abnormally increased sensorimotor cortical activity in unilateral CP, with implications to be investigated.

Impaired manual function is one of the major causes of activity limitation for individuals with unilateral cerebral palsy (CP). While unimanual deficits can be compensated for by preferentially using the non- or less-involved hand,¹ many activities of daily living require the cooperative use of both hands in either a symmetrical (e.g. carrying a tray) or an asymmetrical fashion (e.g. opening a jar). Asymmetrical tasks are especially challenging, as they require selective activation of homologous muscles.²

Studies using various imaging techniques have revealed some mechanisms underlying unimanual impairments in unilateral CP and highlighted the neuroplastic potential in children with this condition.³ Technological advances have recently enabled the use of functional near-infrared spectroscopy (fNIRS) to study sensorimotor brain activity.^4-6 fNIRS non-invasively measures changes in the concentrations of hemoglobin species associated with neuronal activation providing a unique opportunity to study brain activity during functional tasks in populations with movement disorders who may not be able to remain sufficiently still for testing. fNIRS does not require participants to be confined and is relatively robust to movement artifacts.

In children with unilateral CP, fNIRS has reliably shown predominant activity in the non-lesioned cortex during unimanual tasks using either hand,^4,5 and was sensitive to changes in the prefrontal cortex activation after bimanual training, which purportedly indicated improved motor planning.⁷ Other studies have reported increased sensorimotor cortical activity during unilateral and bilateral lower-limb tasks^8,9 that correlated with increased muscle activity, reduced selective control, and impaired mobility in bilateral CP.⁹ So far, functional brain imaging techniques, including fNIRS, have not investigated how sensorimotor cortical activity relates to performance during functionally relevant bimanual tasks. Addressing this gap can help advance neurorehabilitation strategies.

The goals of this study were to: (1) compare patterns of sensorimotor cortical activity during bimanual tasks at varying levels of complexity in individuals with unilateral CP and typically developing controls and (2) explore correlates between brain activation and measures of performance and function in unilateral CP. The group with unilateral CP was expected to show predominant fNIRS activity in the non-lesioned hemisphere, while the typically developing group would show bilateral activation in all tasks. We further hypothesized that asymmetrical bilateral tasks would demonstrate larger group differences in brain activation measured with fNIRS and in task performance, as the coordination demands are higher than in the other tasks. Task performance and hand function were monitored with electromyography (EMG) or motion analysis, and validated clinical scales, as potential correlates with cortical activation.

METHOD

Participants

Ten participants with unilateral CP were enrolled in this cross-sectional case–control study. Two had very large lesions on magnetic resonance imaging (MRI) such that there was little remaining brain tissue within the coverage of the fNIRS probe on the lesioned side and were subsequently eliminated from the analyses (Fig. S1, online supporting information). Therefore, the final analyses included eight participants with unilateral CP (six females, two males; mean age [SD] 20y 10mo [5y 10mo], 13y 8mo–31y 6mo), and nine age-matched participants (p=0.43) with typical development (seven females, two males; 17y 8mo [5y 7mo], 9y 4mo–24y 2mo). Table 1 shows the characteristics of the participants.

Table 1:

Participant demographics

Participant	GMFCS	MACS	Age (y:mo)	Handedness	Etiology	Dystonia score (HAT)
1	I	II	25:2	R	Presumed prenatal stroke	1a
2	II	II	20:11	L	Stroke in infancy	1a,b
3	II	II	21:9	R	Right MCA stroke at age 4y	1a,b
4	II	II	20:4	L	Presumed prenatal stroke	1a,b
5	II	II	14:3	R	Perinatal/neonatal right MCA stroke	1a,b
6	I	III	18:10	L	Prenatal left MCA stroke	1a,b
7	I	II	13:8	L	Prenatal left MCA stroke	1a,b
8	II	III	31:6	L	Neonatal insult-subdural bleed, hydrocephalus	1a,b
Unilateral CP			20:10±5:0
Typically developing	NA	NA	17:8±5:11	9R	NA	NA

^aIncreased movement/posture in upper limb with movement of another body part (HAT item 2).

^bIncreased movement/posture in lower limb with movement of another body part (HAT item 2). Participants 1, 3, and 5 had right-sided hemiplegia; participants 2, 4, 6, 7, and 8 had left-sided hemiplegia. Participant 3 had a later injury and did not formally fit the criterion for CP but was not excluded owing to clinical presentation being similar to other participants and for power reasons. GMFCS, Gross Motor Function Classification System; MACS, Manual Ability Classification System; HAT, Hypertonia Assessment Tool; R, right; L, left; MCA, middle cerebral artery; CP, cerebral palsy; NA, non-applicable.

The inclusion criteria were: age 5 years or older; either having typical development or having a diagnosis of unilateral CP; being able to understand and follow simple directions; being able to provide informed consent (adults) or verbal/written assent (children); and agreement not to drink caffeine or alcohol for 24 hours before assessments because both can modify brain activity. Exclusion criteria were any injury or health condition other than unilateral CP affecting the ability to concentrate or move a body part repetitively; a seizure in the previous year; currently taking seizure medication or muscle tone medication; surgery within the past year or botulinum neurotoxin A injections within the past 4 months. Contraindications for MRI included inability to lie still for several minutes, discomfort being in small places, metal implants, and pregnancy.

Procedures

All tests occurred at the National Institutes of Health Clinical Center. Participants were seated on a plinth with a table in front of them (Fig. 1a). To monitor task performance and relate it to central mechanisms, reflective markers were attached to anatomical landmarks on their upper limbs as defined by the American Society of Biomechanics for upper extremity motion analysis,¹⁰ and EMG surface electrodes were positioned over the muscle bellies of the flexor carpi radialis (FCR) according to SENIAM (surface EMG for non-invasive assessment of muscles) guidelines.

Figure 1:

(a) Squeeze tasks setup. (b) Source-detector layout, with light sources in red and detectors in black numbers. Channels within the red lines represent the hand areas of the brain (C3 and C4) and were included in the region of interest analysis. The figure also shows the approximate locations of the sensory areas (in blue, Brodmann areas 1, 2, 3), the primary motor area (in green, Brodmann area 4), and the premotor area (in orange, Brodmann area 6).

The CW6 fNIRS system uses 690nm and 830nm wavelengths emitted from source optodes at 50Hz. The custom-made fNIRS source-detector arrangement included eight sources and 16 detectors (a total of 42 channels) with interoptode distances from 2.5 to 3.3cm and two short-separation channels (1cm) designed to control for superficial (scalp) hemodynamic changes. The probe was positioned on the basis of anatomical landmarks using the international 10/20 system, covering the C3 and C4 sensorimotor cortical areas (Fig. 1b). Additional information is provided in Appendix S1 (online supporting information).

Four bimanual tasks were chosen that represented different types of collaboration between hands used in daily activities and performed in a randomized sequence: (1) bimanual symmetric squeeze (BSS) task, namely in-phase squeezing; (2) bimanual asymmetric squeeze (BAS) task, namely anti-phase squeezing. A soft squeeze toy was used in squeeze tasks. Participants were instructed to use a moderate amount of force; and (3) bimanual pouring tasks, where one hand stabilized a cylinder in a fixed position while the other arm simulated pouring using a second cylinder. This pouring task was performed: (1) with the dominant hand (DPour) and (2) with the non-dominant hand (NDPour) as the pouring hand. Each task started with a 30 second rest and then was executed repetitively at 1Hz for 15 seconds, repeating this cycle (rest-task) for eight blocks with subsequent rest periods of variable duration (20–30s) to reduce anticipatory activation.¹¹ Visual and auditory cues paced the participants and prompted them to start and stop.¹² The pouring task was differently paced at 0.67Hz so that participants did not have to move so fast as to compromise accuracy. Approximately 80 (pouring) and 120 (squeezing) repetitions of each task were obtained. Participants had no previous knowledge of task sequence.

Clinical assessments

Participants with unilateral CP were assessed using the Gross Motor Function Classification System (GMFCS),¹³ Manual Ability Classification System,¹⁴ Edinburgh Handedness Inventory,¹⁵ and Hypertonia Assessment Tool.¹⁶ The Hypertonia Assessment Tool’s spasticity and dystonia items were tested on right and left arms to assess for the presence of involuntary movements, postures, or resistance. The ABILHAND-Kids¹⁷ and the Pediatric Evaluation of Disability Inventory computer adaptive test (PEDI-CAT)¹⁸ questionnaires were completed by parents/guardians or by participants older than 18 years to explore correlations of brain activity with daily functioning. The PEDI-CAT (n=3) and ABILHAND-Kids (n=5) were completed for some individuals who exceeded the validated age; however, the items are not expected to have age-related differences between a 20-year-old and other young adults. T2-weighted MRI were obtained for all participants to verify whether the sensorimotor brain regions were within the probe limits.

Data analysis

fNIRS data processing and statistical analysis

fNIRS signals from all channels were analysed using the NIRS toolbox (https://bitbucket.org/huppertt/nirs-toolbox/src/default/).¹⁹ Processing included conversion of changes in optical density to changes in hemoglobin concentrations using the modified Beer–Lambert law. We chose to report total hemoglobin values, as they represent the sum of oxygenated and deoxygenated hemoglobin, incorporating both aspects of neurovascular coupling.²⁰ Statistical analysis used a first-level general linear regression model with correction for autoregressive errors and robust (outlier) statistics.¹⁹ Estimates of the average amplitudes of changes in total hemoglobin for each NIRS source-detector pair were obtained from the regression coefficients (β) and their statistics from a Student’s t-test on the coefficients.¹⁹ Effective degrees of freedom in the regression model were estimated from a Welch–Satterthwaite correction. For second-level (group) statistical analysis, a linear mixed-effects model was used. Group (unilateral CP, typical development) and task condition (BAS, BSS, DPour, NDPour) were entered as dependent variables in the model. Participant was used as a random effect. The first-level noise covariance estimates were used to perform weighted/whitened regression of the linear mixed model.²¹ In brief, an iterative autoregressive pre-whitening and robust (Huber bisquare weighting) statistic regression²² was used as implemented in the NIRS toolbox.²¹

The Benjamini–Hochberg procedure was used to correct for multiple comparisons in the presentation of multiple task contrasts and brain regions (the corrected p-value is denoted as a q-value).

Because initial data analysis showed that right- and left-handed participants had comparable task performance, channels were flipped across the midline in the left-handed participants to allow for comparisons between designated dominant (non-lesioned for unilateral CP) and non-dominant (lesioned for unilateral CP) hemispheres. Therefore, the dominant or non-lesioned hemisphere (contralateral to the dominant hand) was represented as being on the left side for all participants. For comparisons of cortical activity in the hand area between and within groups, regions of interest (ROIs) were defined by the channels over the 10 to 20 positions of C3 and C4, which correspond to the left- and right-hand areas respectively (Fig. 1b). ROI statistics accounted for spatially covariant noise between measurement channels from the second-level statistical model.²¹

The short-separation channels signal quality was not adequate for use as regressors in the model; however, to monitor the global impact of scalp blood flow, a comparison between short and long separation channels was completed (Appendix S1).

Brain activity and task performance measures

Collecting performance measures is recommended during functional brain imaging,²³ as a way of monitoring the actual task performance. This is especially relevant in individuals with childhood-onset movement disorders, who may show altered motor behavior.

For squeeze tasks, bilateral raw surface EMG data (Delsys Trigno, Natick, MA, USA) from the FCR (which is significantly recruited in a spherical grasp squeeze task)²⁴ were collected and then baseline corrected, rectified, and lowpass filtered at 4Hz. Coactivation of FCR across arms was quantified using cross correlation of the EMG signals during movement blocks. These values indicated the synchrony between arms over time during bimanual tasks, thus reflecting selective control.²

For the DPour and NDPour tasks, three-dimensional motion of markers placed dorsally on the hands was captured (Vicon Motion Systems, Oxford, UK) to monitor the average path length (defined as total hand displacement) and inform about eventual abnormalities in movement control that could potentially affect brain activity measures. It was calculated for each pouring block in the moving and holding hands and then divided by the number of pours in a block for normalization.

To explore correlates of brain activity with performance measures, positive β-values across channels located in the ROIs were summed to provide a measurement of the amplitude of cortical activation.⁹ Laterality indices (LI) were calculated using average t-values in the lesioned or non-dominant (ND) and non-lesioned or dominant (D) hemispheres, according to the following formula: LI=(D−ND)/(D+ND).⁵ Values for lateral indices ranged from +1 (all motor cortical activation occurring in the dominant (non-lesioned) hemisphere), to −1 (all motor cortical activation occurring in the non-dominant [lesioned] hemisphere).

Statistical analysis

Task performance variables were normally distributed. Repeated measures analysis of variance was used to test the effects of group and task on FCR correlation coefficients during BAS and BSS tasks. Unpaired t-tests were used to compare path length values (pouring tasks) between groups. Amplitudes of cortical activation and lateral indices were correlated with clinical measures (ABILHAND-Kids and PEDI-CAT scores) and kinematic and EMG measures of task performance using Pearson’s r. One participant missed the PEDI-CAT assessment and was not included in the correlation tests using this scale.

All statistical tests performed outside of the NIRS toolbox used SPSS 18.0 (SPSS Inc., Chicago, USA) and p<0.05 was the threshold for statistical significance. For tests done using NIRS toolbox, a corrected p-value, q<0.05, was set.

RESULTS

We first present the results of the group-level mixed-effects analysis, then the results of the post hoc ROI-based analyses including group, task, and hemisphere contrasts. Finally, we give the exploratory analysis of interactions between brain activity and task performance.

Group-level mixed effect model

The modeling of the hemodynamic response yielded activation maps (Fig. 2) that showed differences between groups in several tasks. The group with typical development had low levels of activity confined to a smaller area of the brain in squeeze tasks and bilateral deactivation in the pouring tasks, while the group with unilateral CP showed more diffuse and higher levels of activation bilaterally, which were more prominent during the squeeze tasks. Means and standard deviation values for task-related changes in total hemoglobin are reported in Appendix S1. Additionally, we present values of changes in oxygenated hemoglobin for comparison with other studies (Appendix S1); however, all of results in the current study are based on total hemoglobin.²⁰

Figure 2:

Group-level analysis of total hemoglobin changes for all tasks; t-statistics maps, with significantly active channels (q<0.05) in solid lines; color bar represents the t-statistic value with red/blue indicating a statistical increase/decrease in total hemoglobin concentration during the task relative to the baseline period. CP, cerebral palsy; S, sources; D, detectors; BSS, bimanual symmetric squeeze; BAS, bimanual asymmetric squeeze; DPour, dominant pouring; NDPour non-dominant pouring.

ROI-based analyses: group contrasts

Contrasts between groups in the ROIs confirmed significantly greater activation in unilateral CP both in C3 and C4 (sensorimotor cortex) areas during BAS and in the lesioned hemisphere during BSS and DPour (Table S1, online supporting information) compared with typical development.

ROI-based analyses: task contrasts

Within-group ROI analyses (Table S1) in the group with typical development showed lower activity during BAS, DPour, and NDPour compared with BSS in C3 and C4. In the unilateral CP group, BAS was the task with the highest activity, being accompanied by significantly higher cortical activity in both hemispheres compared with BSS and DPour. BSS was higher compared with NDPour.

ROI-based analyses: hemisphere contrasts

Differences between C3 and C4 areas were seen in the group with unilateral CP for all tasks except DPour. BSS and BAS showed greater activation in the lesioned hemisphere (Table S1). The NDPour in unilateral CP showed greater activation on the non-lesioned hemisphere. In typical development, only the DPour task showed greater activation in the dominant hemisphere.

Brain activity and performance interactions: exploratory analysis

Analysis of EMG activity showed a main effect of task (F₁=25.799; p<0.001; partial η²=0.665), with higher correlations between dominant and non-dominant FCRs during BSS as expected compared with all other tasks. There was also a significant group×task interaction (F₁=7.298; p=0.018; partial η²=0.360), with the group with unilateral CP showing higher EMG correlation values (decreased ability to perform alternating movements) than those with typical development during BAS. There was no main effect of group (F₁=0.378; p=0.549; partial η²=0.028).

The average path length of the pouring hand in the NDPour was shorter in the group with unilateral CP (t=2.1; p=0.04; 95% confidence interval [CI] 0.1975–0.3425), with no differences in the other comparisons. The holding hand stayed relatively still regardless of the pouring hand in both groups.

Mean values for the task performance variables are shown on Table 2.

Table 2:

Mean values for task performance measures

	Mean (SE) FCR correlation		Mean (SE) path length, m
	BSS	BAS	DPour	NDPour
TD	0.50 (0.11)	−0.14 (0.05)	0.65 (0.09)	0.66 (0.06)
UCP	0.34 (0.08)	0.14 (0.06)	0.67 (0.21)	0.39 (0.08)

Mean (SE) values of flexor carpi radialis (FCR) correlation coefficients during squeeze tasks and path length during pouring tasks. SE, standard error; BSS, bimanual symmetric squeeze; BAS, bimanual asymmetric squeeze; DPour, pouring task with dominant hand; NDPour, pouring task with non-dominant hand; TD, typical development; UCP, unilateral cerebral palsy.

Correlation tests were only significant for BAS. Brain activity levels in the group with unilateral CP were related with more synchronized EMG activity of FCRs (r=0.79; p=0.02; 95% CI 0.03–0.97), with more brain activation being related to greater EMG synchrony. This correlation was not significant in the group with typical development. The laterality index in the group with unilateral CP correlated to PEDI-CAT daily activity scores (r=0.81; p=0.03; 95% CI 0.04–0.97), with a more positive index (activity more lateralized towards the non-lesioned hemisphere) correlating with better function. Individual results regarding clinical scales are available in Table S2 (online supporting information).

DISCUSSION

The results provide novel findings related to the control of bimanual tasks in unilateral CP. As hypothesized, group differences were particularly notable when asymmetric collaboration between hands was required.

Group-level hemodynamic response

Low activation levels were seen in all tasks in the group with typical development, which may be because of the minimal effort the tasks required for unimpaired individuals. This finding is similar to what was previously reported for a simple focal lower-limb task performed by individuals with bilateral CP.⁹

Increased bilateral activation in unilateral CP during movement of the affected hand may be explained by increased cognitive and motor demands of the task on the cortex, and/or reduced selectivity of motor patterns in unilateral CP.^9,25,26 Effort-related findings in CP include excessive grip forces and unintended recruitment of additional muscles that may lead to mirrored movements and other compensatory movement strategies.²⁷

Sensorimotor cortical activity across groups

Between-group comparisons confirmed our initial hypothesis that larger group differences would be found for tasks with higher coordination demands. BAS showed the highest activations with sensorimotor areas bilaterally being significantly higher than typical development, while during BSS and DPour activation was only significantly higher in the lesioned hemisphere compared with typical development. One possible explanation for the greater activation in the non-lesioned hemisphere during BAS compared with typically developing individuals might be the maintenance of ipsilateral projections and potential overlap of cortical representation of the two hands on the non-lesioned hemisphere²⁸ that fails to allow for selective activation of the hands. Diminished sensation in the affected hand, although not tested, might also have contributed to the squeezing difficulty.

It is also possible that the brain injury affected interhemispheric inhibitory processes that are critical for hand selective control. Moving two hands asymmetrically requires complex interhemispheric coordination to activate or inhibit non-homologous muscles.² Higher activation in unilateral CP can therefore be related to the greater effort needed to produce a coordinated asymmetric movement against the tendency to produce more symmetric movements.²⁹ Finally, asymmetric tasks may increase attentional demands, as suggested by an earlier study on complex action planning in children with unilateral CP that demonstrated greater prefrontal cortical activation compared with controls.⁷ Our findings in the sensorimotor cortex may therefore indicate increased need for sensorimotor resources to perform complex motor tasks.

Indeed, task performance measures showed a positive correlation (greater synchrony) between the hands in those with CP during BAS whereas alternating activation of wrist flexors was seen as expected in controls. This is consonance with previous studies showing that the preferred pattern of bimanual coordination is towards synchrony, because interlimb coupling requires less sophisticated levels of neural control.³⁰ Furthermore, higher levels of cortical activation correlated with more synchronous EMG activity, namely with greater brain activity, movements tended to be less discrete. Although our data do not allow us to state that the task performance caused altered brain activity or vice-versa, higher levels of brain activity should be further explored as one potential factor affecting selective hand control.

Also, high levels of brain activity in the NDPour combined with the decreased path length in unilateral CP suggest efforts to compensate for less effective movement trajectories. Results for these tasks shed light on challenges faced by individuals with unilateral CP when attempting to perform functional tasks and motivate future research to measure brain activity during real life situations, which can be done with fNIRS and/or EEG, to provide evidence of neuroplasticity and generalizability after interventions aimed at improving hand function.

Sensorimotor cortical activity across tasks

The comparisons across tasks support the notion that activities requiring both hands to move asynchronously are especially challenging for individuals with unilateral CP. Activation was higher during BAS in the group with unilateral CP compared with all the other tasks, while inter-task differences were less notable in the comparison group. In the group with typical development, asymmetric bilateral tasks were characterized by negative total hemoglobin values, which may reflect more intra- and intercortical inhibition than activation, while the symmetric task showed higher bilateral activation as expected. The pouring tasks had less activation than the squeezing tasks in both groups, perhaps because they require movements and stabilization of proximal joints.

Sensorimotor cortical activity across hemispheres

Comparing activation across hemispheres surprisingly showed that in typical development there were few interhemispheric differences. This is probably due to the low levels of brain activation across tasks that make the detection of any differences difficult. Also surprising were the findings in the group with unilateral CP of significantly greater activation in the lesioned hemisphere for all tasks using the affected extremity. While this initially appears counterintuitive, data suggest that more effective and efficient motor control is characterized by smaller and more focal brain activation patterns.

An activation pattern more lateralized towards the non-lesioned hemisphere during the BAS task may relate to better functional performance, as suggested by the correlation between the laterality index and PEDI-CAT daily activity domain. This finding is supported by a previous study suggesting a central role of the dominant hemisphere in bimanual function;³¹ however, further investigation is needed to understand the functional implications.

The magnitude of activation did not correlate with the functional scales in our study (ABILHAND-Kids and PEDI-CAT). It is possible that activity in brain regions not measured in this study and/or other activation during tasks not studied here may correlate better with functionality. Additional direct measures of hand function (e.g. the Assisting Hand Assessment) not used in this study could also show differences and should be explored in future studies.

Caution is advised when interpreting fNIRS data in participants with cortical injuries that may distort the underlying anatomy. In this study, two participants had to be excluded owing to extensive loss of brain tissue under the areas covered by the probe, which would not have been known without structural images. Coregistration of the probe location to individual MRIs is another potential, albeit challenging, solution for this. Although some analysis packages include coregistration approaches and non-pathological templates are available,³² further development is needed to enable their use in children with brain injuries. The technique also has depth limitations, such as only measuring blood flow in the top layers of the cortex in older children and adults. Furthermore, a limitation in this study was not using scalp blood flow signals as regressors. However, fNIRS capabilities are evolving rapidly and research using this technology has increased exponentially in recent years.

Additional limitations of our study include the sample size that precludes analysis of individual factors such as handedness, injury type, timing and size, and potential remapping after injury. As reduced sample size may have impacted the findings, the results should be viewed as exploratory.

Summary

Collectively these results provide neural correlates of impaired bimanual function in unilateral CP that warrant further confirmation in larger studies as well as studies investigating how to facilitate selective activation to potentially improve bilateral hand function. While some rehabilitation techniques have leveraged bilateral symmetry to facilitate bimanual function in CP,³³ the more pronounced activation associated with greater difficulty in asymmetric tasks suggest that targeted activities to improve these movements may be of even greater benefit. Practice of symmetrical bimanual tasks may interfere with efforts to improve isolated control of the affected hand, whereas more asymmetrical training may facilitate this.

Although participants in this study were older than those typically receiving intensive rehabilitation, our results demonstrate persistent manual impairments and abnormalities in brain activity. As brain stimulation techniques continue to develop as a therapeutic alternative,³⁴ the potential room for improvement in similarly older participants should not be disregarded. Knowing patient- and task-specific patterns of sensorimotor cortical activity will be central for defining treatment parameters and prescribing more effective individualized approaches

CONCLUSIONS

This study revealed unique characteristics of brain activity in a sample with unilateral CP, such as more diffuse and higher levels of activation compared with controls in several functionally relevant tasks. Correlations between neural and behavioral measures for asymmetrical hand tasks suggest some possible targets for intervention and warrant further investigation. Results suggest fNIRS as a useful tool to assess neural correlates of motor dysfunction and to demonstrate neural mechanisms underlying intervention outcomes.

Supplementary Material

Supplementary_Fig1

Figure S1: Magnetic resonance images of 10 participants with unilateral cerebral palsy.

Table S2

Clinical measures.

Table S1

Results of the group-level mixed-effects analysis and contrasts.

Appendix S1

fNIRS methods and additional results.

What this paper adds.

• Cortical activity in manual tasks is described with functional near-infrared spectroscopy in typical and atypical cohorts.

• Activation levels in unilateral cerebral palsy appear to escalate with task difficulty.

• Increased brain activity may be associated with poorer selective manual control.

• Specific patterns of brain activity may be related to impaired bimanual function.

ABBREVIATIONS

BAS

Bimanual asymmetric squeeze

BSS

Bimanual symmetric squeeze

DPour

Pouring task with dominant hand

FCR

Flexor carpi radialis

fNIRS

Functional near-infrared spectroscopy

NDPour

Pouring task with non-dominant hand

PEDI-CAT

Pediatric Evaluation of Disability Inventory computer adaptive test

ROI

Region of interest