2.1.

Participants

A total of 40 participants were consented into the protocol and completed the study, including 14 (five males) with BCP, Gross Motor Function Classification System (GMFCS) levels I to III, 10 (two males) with UCP, and 16 (nine males) with TD. Participants were included if they were at least 5 years old, able to understand and follow simple directions for performing a repetitive task, and agreed to not drink caffeine or alcohol for 24 h before assessments to avoid associated alterations in blood flow dynamics. Those in the UCP group had unilateral injuries confirmed by structural MRI. Exclusion criteria included any health condition or diagnosis other than CP that would affect the ability to maintain attention or move a body part repetitively for short periods of time, surgery within a year, or botulinum toxin injections within 6 months. All participants, or their parents, as applicable, completed informed consent and assent for those $<18\text{years}$ of age.

Standardized clinical assessments were performed including the GMFCS,²³ Manual Ability Classification System (MACS),²⁴ Hypertonia Assessment Tool,²⁵ and Edinburgh Handedness Inventory with an additional question about lower extremity preferences to determine which leg to test. Furthermore, the ABILHAND-Kids,²⁶ ABILOCO-Kids,²⁷ and Pediatric Evaluation of Disability Index–Computer Adaptive Test (PEDI-CAT) questionnaires (version 2.5)²⁸ were completed. Parents or adult participants completed the ABILOCO-Kids and ABILHAND-Kids questionnaires about difficulty performing ambulation and manipulations tasks, respectively. Answers from each question are input to an online algorithm for transformation into logit score ranging from $-6$ to 6 for linear representation of scores. The PEDI-CAT yields summary scores for mobility and daily activity domains based on adaptive administration of questions. Higher scores indicate better function.

2.2.

Setup

Setup for each participant included placement of the fNIRS optodes (CW6, TechEn, Milford, Massachusetts) on the scalp regions that overlay the sensorimotor cortical areas (Fig. 1). The hair was carefully parted to align with optode placement as described previously.²² Optode arrangements differed slightly for UCP and BCP participants, but a common set of channels was identified for all tasks which included eight sources and 14 detectors with a total of 42 channels.

Fig. 1

Optode arrangement placed directly on scalp locations over the sensorimotor cortex, as viewed from above. Red circles denote light sources, gray filled in circles denote light detectors, and the lines between them are the channels that represent activation on the cortical surface between the emitter–detector pairs. Channels that were included in the lateral region of interest are denoted in black, while those considered to be medial region of interest are in white. Nz, nasion; Iz, inion; Al, left auricular; Ar, right auricular.

2.3.

Motor Tasks

The tasks investigated for this cohort were nondominant hand grasp and relaxation around a soft object (“squeeze” task) and nondominant ankle dorsiflexion and relaxation against gravity (“dorsiflexion” task). Participants were seated on a plinth with trunk support and completed eight task blocks of 15 s each, interspersed with a variable rest period of 25 to 35 s. They were cued to move at 1 Hz with audio²⁹ and visual prompts customized for this experiment. Participants were video recorded during the tasks to aid in interpretation of data and were given breaks in between tasks as needed.

2.4.

Data Analysis

Light intensities in the fNIRS signals were analyzed using a set of commands in the NIRS toolbox.³⁰ In brief, the processing steps included converting raw data to optical density, followed by conversion to oxygenated and deoxygenated Hb using the modified Beer–Lambert law. A deconvolution of the impulse response function was visualized using the <MyStats>.HRF function in the toolbox,³¹ shown in Fig. 2 as representative signals. Group-level statistical analysis was completed using a first-level canonical general linear model,^30,32 which was used to estimate the statistical response of each fNIRS light emitter–detector pair (channel) to the functional task. An autoregressive whitening filter was iteratively computed and applied to the data and linear regressor model to reduce the effect of serially correlated noise.³² A robust linear regression using the Huber bisquare weighting was applied to estimate the model. The regression beta coefficients and $t$-statistical effects were estimated for each fNIRS channel and task condition. This procedure was previously shown to correct the high false-discovery rate problems that result from motion and nontask coupled superficial physiological noise in fNIRS analysis.³² For all analyses, the sum of oxygenated and deoxygenated Hb or the total hemoglobin (HbT) was used to consider both aspects of neurovascular coupling and also for its robustness against artifact from blood flow in the superior sagittal sinus which is important when investigating a task such as dorsiflexion.³³ Statistical estimates were Benjamini–Hochberg corrected for comparisons across all fNIRS emitter–detector pairs and task conditions and the false discovery rate (FDR) corrected $p$-values (denoted as the $q$-value) were used.

Fig. 2

Selected channels for representative participants ($n=1$ per plot) in each group and task. Lines represent the deconvolution of the impulse response function subject-level analysis in the NIRS toolbox, where purple solid lines are from the right hemisphere, contralateral to the task, and green dashed lines are channels from the left hemisphere, ipsilateral to the task. More medial channels are represented in the dorsiflexion task, and lateral channels for the squeeze task to reflect the relative location on the motor homunculus for these two body areas. The 15-s task period is represented by the gray bar on each graph.

2.5.

Evaluation of Hemodynamic Response and Task Performance

To assess differences in brain activity between groups (TD, BCP, and UCP), channels were reversed across the midline for the participants whose tested, functionally nondominant limb was on the right. Therefore, brain hemisphere contralateral to activity was always displayed as being on the right side and was the lesioned hemisphere for those with UCP. To address the primary question of differences between types of CP, the NIRS toolbox was used to test the effect of group on recorded activity, and to evaluate contrasts within the subgroups of CP diagnoses (groups, GMFCS levels, and MACS levels). Secondarily, the contrast between squeeze and dorsiflexion tasks was evaluated within each group. A region of interest analysis was used to distinguish the medial and lateral portions of left and right hemispheres (shown in Fig. 1) to describe topography of activation differences associated with upper and lower extremity tasks.

On an individual participant level, the number of fNIRS channels that was significantly active ($q<0.05$) and the sum of beta values for all active channels were determined for each hemisphere and across both hemispheres as an indication of the extent and level of brain activity, respectively.

Video recordings were reviewed for the absence or presence of mirror movements observed on the limb opposite to the intended motion. Participants were considered to demonstrate no mirror movements if none were observed, mild mirror movements if analogous movements were observed on the contralateral side $<50\%$ of the time, and strong mirror movements if analogous movements were observed on the contralateral side more than 50% of the time.

2.6.

Laterality Index to Evaluate the Balance of Activation Across Hemispheres

On an individual participant level, laterality indices (LI) were calculated using the difference in beta values of significantly active channels ($q<0.05$) between hemispheres divided by the sum. This index quantifies the overall balance in activation between contralateral and ipsilateral hemispheres relative to the task being performed and has been widely used in previous studies. A value closer to 1 indicates more strongly contralateral hemisphere activity, while a value closer to $-1$ indicates more strongly ipsilateral hemisphere activity. Values close to 0 indicate that the two hemispheres were activated to a similar degree during a unilateral task (bilateral activation).

2.7.

Relationship to Functional Scales

Because development of handedness is likely to be altered by early brain injury, we evaluated if there was a systematic impact of Edinburgh scores with clinical scales. In addition, we performed a range of correlation analyses to determine if relationships existed between fNIRS-derived metrics and clinical examination metrics in participants with CP. Variables derived from the fNIRS analysis (LI, number of active channels on either or both hemispheres and sum of beta values in individual or both hemispheres) were correlated using a Pearson’s correlation with ABILHAND and PEDI-CAT daily activity domain for the squeeze task, and with the ABILOCO and PEDI-CAT mobility domain for the dorsiflexion task. A Spearman correlation was performed relating the Selective Control Assessment of the Lower Extremity (SCALE) and GMFCS to the dorsiflexion task fNIRS metrics and the MACS to the hand squeeze task fNIRS metrics.

2.8.

Statistical Analyses and Comparisons Between Tasks

The first set of comparisons between tasks was completed using a contrast in the NIRS toolbox between hand and ankle as described above, which allowed for a region-of-interest-based evaluation of cortical activity differences between a hand and ankle task. In addition, a separate one-way analysis of variance was used to evaluate the effect of group (three levels: TD, UCP, and BCP) on each of the following dependent variables: number of active fNIRS channels, sum of beta values for active fNIRS channels, and LI. Significant main effects were explored further using posthoc tests with Bonferroni corrections.

For all statistical tests performed outside of the NIRS toolbox using SPSS (version 26), $p<0.05$ was considered significant. For any tests done using NIRS toolbox, an FDR corrected $p$-value, $q<0.05$, was used as a threshold for significance.

3.1.

Participant Characteristics and Clinical Features

Participant demographics are summarized in Table 1. Age did not differ significantly across groups. All participants with BCP had spasticity (HAT item 3) in at least one LE (including the side tested), with five having spasticity in the UE as well. There was spasticity present in both lower and upper extremities of the more impaired side for all with UCP. There were no significant differences in clinical exam scores, with the exception of the BCP group having lower PEDICAT mobility scores. Because light transmission can be decreased by darker hair color, summary participant characteristics³⁴ are also reported in Table 1.

Table 1

Participant characteristics.

3.2.

Evaluation of Hemodynamic Response in both Sensorimotor Cortices

The primary aim of this study was to compare and contrast cortical activation patterns in children with unilateral and BCP, with both compared to those in children with TD, in an upper and lower extremity tasks. Data are further presented across functional classification levels within the CP subtypes, using the MACS for the squeezing task and the GMFCS for the dorsiflexion task.

For the nondominant hand squeezing task, there was globally more activity in the groups with CP, compared to the group with TD [Fig. 3(a)]. The TD group demonstrated few significantly active channels in group analysis, including two ipsilateral channels that demonstrated a significant reduction in cortical activity relative to rest (negative $t$-statistic). Although not significant, there were positive $t$-statistics (indicating cortical activation) on the contralateral hemisphere. Both the UCP and BCP groups had more significantly active channels in both hemispheres than TD, which were all indicative of cortical activity (positive $t$-statistics). In the contrast analysis, the UCP group had the highest level of activity, with significant differences found between all groups ($\mathrm{UCP}>\mathrm{BCP}$ $q=0.049$; $\mathrm{BCP}>\mathrm{TD}$ $q<0.001$; and $\mathrm{UCP}>\mathrm{TD}$ $q=0.001$). Within the contrasts for TD and MACS levels, the group with the greatest impairments had the highest levels of cortical activity [Fig. 3(b)]. In this analysis, all contrasts were significant except for the comparison between MACS I and II ($\mathrm{III}>\mathrm{I}$ $q<0.001$; $\mathrm{I}>\mathrm{TD}$ $q<0.001$; $\mathrm{III}>\mathrm{II}$ $q<0.001$; $\mathrm{II}>\mathrm{TD}$ $q<0.001$; $\mathrm{III}>\mathrm{TD}$ $q<0.001$; and $\mathrm{I}=\mathrm{II}$ $q=0.84$).

Fig. 3

HbT activation maps by group and task. Significantly active channels ($q<0.05$) are in solid lines, and the colorbar represents the value of the $t$-statistic ($t$-stat). Positive values (red colors) denote increases in HbT relative to baseline, and negative values (blue colors) indicate that signal was greater during baseline than task. The right hemisphere is contralateral to the task limb, and the orientation of the probe is centered around Cz with left and right hemispheres denoted by “L” and “R” in the typically developing (TD) map in (a).

There was also globally more activity in the CP groups for the dorsiflexion task, compared to the TD group [Fig. 3(c)]. Again, there were no significantly active channels in the TD group, but positive $t$-statistic values were identified relatively close to midline on the contralateral hemisphere. Both CP groups had positive $t$-statistics in both hemispheres, but the only significantly active channels were found in the BCP group. In the contrast analysis, the highest activity was recorded in the BCP group ($\mathrm{BCP}>\mathrm{UCP}$ $q<0.001$; $\mathrm{BCP}>\mathrm{TD}$ $q<0.001$; and $\mathrm{UCP}>\mathrm{TD}$ $q=0.001$). Within the contrasts for TD and GMFCS levels [Fig. 3(d)], all contrasts were significant ($q<0.001$) except for the comparison between GMFCS I and II ($\mathrm{III}>\mathrm{I}$, $\mathrm{I}>\mathrm{TD}$, $\mathrm{III}>\mathrm{II}$, $\mathrm{II}>\mathrm{TD}$, $\mathrm{III}>\mathrm{TD}$, and $\mathrm{I}=\mathrm{II}$).

When contrasting the upper and lower extremity tasks (Fig. 4), there were almost no significantly different channels in any group. In general, there are more positive than negative $t$-statistics, which indicate greater activity during the upper extremity squeeze task than the lower extremity dorsiflexion task in all groups. In the TD group, this difference was found to be significant when the channels on the lateral portion of the contralateral hemisphere (defined in Fig. 1) were grouped in a region of interest ($q<0.001$).

Fig. 4

HbT contrast between tasks. Significantly active channels ($q<0.05$) are shown in solid lines, and the colorbar represents the value of the $t$-statistic ($t$-stat). Positive values (red) indicate higher activity level for the squeeze task, while negative values (blue) indicate higher activity level during the dorsiflexion task. Four regions of interest were evaluated (left lateral, left medial, right medial, and right lateral), and significantly different regions are labeled with horizontal bars. *$q<0.05$, **$q<0.001$.

In the UCP group, there were higher levels of activity during squeeze compared to dorsiflexion across of the ipsilateral hemisphere (lateral ipsilateral $q=0.01$ and medial ipsilateral $q<0.001$), and in the medial aspect of the contralateral hemisphere ($q<0.001$). In contrast, the lateral portion of the contralateral hemisphere was not differently active between tasks ($q=0.89$).

There were no statistically significant differences between squeezing tasks and dorsiflexion tasks in the BCP group when considering any channel individually or portions of either hemisphere as a group (all regions $q=0.91$).

3.3.

Evaluation of Group Differences in Which Hemisphere Had Greater Brain Activation

We anticipated that the group with UCP would demonstrate more ipsilateral brain activity especially when using their nondominant upper limb than those with BCP or with TD. For the LI, there was a nonsignificant trend toward an effect of group ($F=2.842$ and $p=0.076$) in the squeezing task (Fig. 5), where both TD and BCP demonstrated a contralateral response on average, while the UCP group brain activation was shifted toward the ipsilateral hemisphere. There was no effect of group ($F=0.812$ and $p=0.452$) in the dorsiflexion task where all groups tended toward a bilateral or slightly ipsilateral response, with the UCP group being the most ipsilateral on average, and the BCP group value between the UCP and TD values.

Fig. 5

Average of the laterality index (LI) across participants, with standard error bars. A positive LI value indicates more activity on the contralateral hemisphere to the task, while a negative LI indicates higher ipsilateral activity with values centered around 0 indicating more bilateral activation patterns. For squeezing, there was a trend of an effect of group ($F=2.842$, $p=0.076$). For dorsiflexion, there was no effect of group ($F=0.812$, $p=0.452$). TD, typically developing; UCP, unilateral cerebral palsy, and BCP, bilateral cerebral palsy.

Another way to assess hemispheric differences was to compare brain activation outcomes in the hemisphere ipsilateral versus contralateral to the limb performing the task. There were differences found in the number of active channels and beta values from the GLM in that the UCP group had the highest values in the squeeze task, while the BCP group was highest in the dorsiflexion task as stated previously. The sum of significant beta values was used as a measure of intensity of activity, and there was a main effect of group for both the ipsilateral ($F=4.423$; $p=0.023$; $\mathrm{UCP}>\mathrm{TD}$ $p=0.051$; and $\mathrm{UCP}>\mathrm{BCP}$ $p=0.053$) and contralateral ($F=5.031$; $p=0.014$; $\mathrm{UCP}>\mathrm{TD}$ $p=0.018$; and $\mathrm{UCP}>\mathrm{BCP}$ $p=0.051$) hemispheres during the squeeze task. The extent of activity is evaluated by the total number of active channels in each hemisphere. Both CP groups were higher on average than TD. There was a main effect of group for the ipsilateral hemisphere during the dorsiflexion task ($F=4.051$, $p=0.026$; $\mathrm{BCP}>\mathrm{TD}$, $p=0.022$).

3.4.

Relationship to Functional Scales

No correlation was found between fNIRS variables (beta LI, active channels on either or both hemispheres, beta in either or both hemispheres) during the squeeze task with ABILHAND, PEDI-CAT daily activities, or mirror movements. The beta in the right hemisphere during dorsiflexion was inversely related with the ABILOCO ($\mathrm{rho}=-0.467$, $p=0.044$). There were no other correlations found between fNIRS variables (beta LI, active channels on either or both hemispheres, beta in either or both hemispheres) during the dorsiflexion task with ABILOCO, PEDI-CAT mobility score, or mirror movements.

There was, however, a significant relationship between the MACS level and number of active left hemisphere channels ($\mathrm{rho}=0.366$, $p=0.02$), total number of active channels ($\mathrm{rho}=0.352$, $p=0.03$), and beta in the left ($\mathrm{rho}=0.635$, $p<0.001$), right ($\mathrm{rho}=0.459$, $p<0.01$), and both ($\mathrm{rho}=0.567$, $p<0.001$) hemispheres during the squeeze task.

There was also a significant relationship between the GMFCS level and number of active left ($\mathrm{rho}=0.437$, $p<0.01$), right ($\mathrm{rho}=0.432$, $p<0.01$), and both ($\mathrm{rho}=0.467$, $p<0.01$) hemisphere channels, and beta in the left ($\mathrm{rho}=0.559$, $p<0.001$), right ($\mathrm{rho}=0.594$, $p<0.01$), and both ($\mathrm{rho}=0.643$, $p<0.001$) hemispheres during the dorsiflexion task. There was a trend noted for a relationship between SCALE score and beta across both hemispheres ($\mathrm{rho}=-0.397$, $p=0.058$) and beta on the left hemisphere ($\mathrm{rho}=-0.452$, $p=0.060$), indicating that higher activity may be associated with poorer selective motor control in CP.