2.1.

Participants

The Institutional Review Board of the University of Nebraska Medical Center (UNMC) approved the study, and we obtained parental consent and child assent to participate in this investigation. The participating children with HCP were recruited from the physical therapy clinic at UNMC and TD children were recruited through word of mouth. The physical therapists, who had access to the children’s health records, prescreened the children based on the inclusion and exclusion criteria to qualify for the study. The exclusion criteria were the presence of frontal cortical lesions, cognitive impairments, visual deficits, musculoskeletal deformity of the hand and arm, such as any fixed contractures of hand or wrist, which limited the ability to grasp and manipulate objects, manual ability classification level (MACS) level I (handles objects easily and successfully) and V (does not handle objects and has severely limited ability to perform simple actions), and arm weakness due to neurological impairments such as brachial plexus injuries. Twelve children with HCP ($\text{age}=6.8\pm 2.9\text{years}$; $\text{males}=7$) and 15 TD children ($\text{age}=5.8\pm 1.1\text{years}$; $\text{males}=8$) fit the inclusion criteria and participated in this investigation. All children with HCP had a previously defined diagnosis of hemiplegia by a pediatric neurologist. Further details of the participating children are shown in Table 1.

Table 1

Demographic details of the participating TD and children with HCP.

2.2.

Clinical Assessments

As an initial screening, we assessed manual ability of children with HCP using manual ability classification system (MACS).³⁵ A certified assisting hand assessment (AHA) assessor also performed the AHA in children with HCP using a standardized, semistructured, 10 to 15-min play session with toys requiring bimanual use of the arm.^36,37 The session was video-recorded and later scored to determine bimanual coordination and assisting hand function of children with HCP.³⁷ Lastly, we had the children performing the nine-hole peg test (NHPT) and the box and blocks test (BBT) to assess manual dexterity and speed.^38,39 Moreover, the handedness for all children was determined using Edinburgh handedness inventory.⁴⁰

2.3.

Functional Near-Infrared Spectroscopy Experimental Paradigm

The task consisted of a sequential shape-matching task, which was developed for this study. This task consisted of three-complexity levels, such as easy, moderate, and difficult. The easy condition had the same shape types, the moderate condition had two different shape types, and the difficult condition had multiple shape types that were different from each other (Fig. 1). The three-complexity levels of the task were based on the intricacy of the shape identification, accurate selection, manipulation, and the type of grasp required based on the type, size, shape, and orientation of the shape. The children were asked to match the shapes with the corresponding template by selecting an appropriate shape and placing it accurately on a given template. The likelihood that the different complexity levels of the shape-matching task would elicit changes in the HbO concentration within the PFC was initially determined with a pilot study that we conducted with five TD children. The results of this preliminary analysis are shown in Table 2.

Fig. 1

Experimental shape-matching conditions. The shapes template was presented to the child just before beginning the task. (a) Easy: the easy condition had the same shape types, (b) moderate: the moderate condition had two different shape types, and (c) difficult: the difficult condition had multiple shape types that were different from each other. The children were asked to match the maximum number of shapes with the corresponding template for 30 s. The conditions were randomized and each task condition was repeated four times.

Table 2

Pilot data that were performed with a cohort of TD children (N=5; age=5.56±0.63 years; males=4) that were right hand dominant. The children completed the shape-matching task that was used in this investigation. The maximum HbO concentration (μmol) seen during the active period was calculated as a preliminary measure of the neural activity in the prefrontal cortices. Trends in the data suggested that the maximum HbO was highest for difficult condition followed by moderate and easy conditions. R=right arm and L=left arm.

The shape-matching task was performed in a block paradigm, which consisted of a 30-s rest period where the child sat still and a 30-s active period where the child matched the shapes. The children did not have a chance to visualize the shapes during the rest period. To avoid time for prior planning of the task, the assessor presented the template and shapes at time 0. We allowed children to match as many shapes as they could within the 30-s active period. Typically, the children could not match all 12 shapes in this time frame. However, in the rare instance, if the child matched all 12 shapes, the assessor removed the shapes from the template and the child continued matching the shapes on the same template until the 30 s of the active period was over. To avoid anticipation of the respective complexity levels, the conditions were randomized and each task condition was repeated four times. The children performed a total of 12 blocks of the shape-matching task ($\text{three shape complexity conditions}\times \text{four repetitions}$ of each condition) during the entire session with an alternating pattern of the 30 s active and rest periods. The total duration of the data collection was 12 min. Children with HCP performed the task with the affected and the less affected arm, and TD children performed the task with the dominant and the nondominant hand. We chose to evaluate both arms to explore the global nature of cognitive processes required for the movement planning and control and to avoid the arm bias of the hemiplegic hand due to physical restrictions in performing the task in view of impairments in the affected arm.

2.4.

Functional Near-Infrared Spectroscopy Data Acquisition

fNIRS is a neuroimaging technique that utilizes specific wavelengths of infrared light to approximate the absorption characteristics of oxygenated (HbO) and deoxygenated (HbR) hemoglobin within the underlying tissues.⁴¹ Although the infrared light is partially absorbed by nonneural tissue, the relative change in the HbO or HbR is linked with the change in the underlying neural dynamics of the cortices. The fNIRS device consists of a series of photon emitters and detectors. The detectors measure the refracted light, which is used to quantify the amount of HbO and HbR changes in local neural tissues. A greater concentration of HbO corresponds to a heightened amount of activity in the underlying neural tissues.⁴¹ fNIRS is a noninvasive, safe, and portable method for assessing cortical activity. It is robust with respect to body movements and is not performed in a constrained environment like what is seen in a functional magnetic resonance imaging (fMRI) bore. Additionally, fNIRS is quiet (no operating sound) and can be used to assess the ecologically valid motor tasks. Because of all these advantages, fNIRS was used to address the research questions put forth in this investigation.

For this experiment, we used a continuous wave fNIRS system (fNIR Devices LLC, Potomac, Maryland) that utilized two different wavelengths (730 and 850 nm) to measure the concentration of HbO and HbR based on the modified Beer–Lambert law.⁴² The fNIRS system (Fig. 2) was composed of three components: a flexible head piece, which secures the emitters and detectors in a fixed position to allow for fast placement of the sensor pad on the forehead; a control box for hardware; and a computer that runs the data acquisition. The positioning of two light sources and two detectors on the sensor pad yielded a total of four measurement channels. According to 10 to 20 EEG system, the optodes were positioned lateral to the Fpz on the left and right sides of the forehead. The sensors had 2-Hz sampling rate and measured the hemodynamic change in $\sim 5$ to 10 mm of the outermost cortical tissue.^43–45 All optodes were connected to fiber optic cables that allowed the transmission of infrared light to the fNIRS system. We used cognitive optical brain imaging studio software for data acquisition and visualization (fNIR Devices LLC, Potomac, Maryland).

Fig. 2

Depiction of the placement of the fNIRS system on an exemplary child. The system consists of a flexible sensor pad that is positioned on the forehead based on the 10 to 20 EEG system. The flexible sensor pad contains the emitters and photo-detectors. This arrangement results in two measurement channels per hemisphere.

2.5.

Functional Near-Infrared Spectroscopy Data Processing and Analysis

We used the fNIRSoft software (v 3.1; fNIR Devices LLC, Potomac, Maryland) to calculate the optical density of the raw light intensity signals, and the implementation of a modified Beer–Lambert law to determine the changes in the HbO concentration.⁴⁶ Waveforms that were saturated or had motion artifacts were visually identified and excluded from the analysis. A fourth-order digital filter was applied to low- and high-pass filter of the HbO time series with 0.1 and 10 Hz cutoffs, respectively. These filter settings were applied to reduce the potential effects of the physiological noise (e.g., equipment noise, respiration, and cardiac cycle effects) that often accompanies the measured cortical hemodynamics.^47–49 The epochs of each trial were 60 s in duration ($-30$ to $+30\mathrm{s}$), with the presentation of the shape-matching task defined as 0.0 s. The HbO hemodynamic waveforms for each channel were corrected based on the average HbO seen in the baseline period ($-10$ to 0 s), and the four trials performed in each condition were averaged to create the time course of the concentration of the HbO for the right and left PFCs, respectively. Subsequently, average HbO across the active period was calculated as a proxy measure of the neural activity in the PFC. We used the average HbO as a marker for regional brain activation since previous study findings have shown that HbO is more sensitive to neural changes than HbR.⁵⁰

2.6.

Behavioral Data Analysis

The video-recorded behavioral data for the shape-matching task were used for the analysis of the motor task performance. The number of accurately matched shapes was quantified across each trial, and the average performance across the four trials for each condition was used as an outcome variable. We also assessed an average number of errors in matching the shapes across all trials. A wrong match of the shape on the corresponding template, use of the arm other than the testing arm, and inaccurate orientation of the shapes were considered as errors. In addition, we assessed reaction time (RT), which was determined as time to initiate the hand movement after the shape-matching task was presented. RT for the first shape in each trial was assessed, and average RT across all trials was considered for the final analysis. Performance of NHPT and BBT was also video-recorded and scores were calculated by quantifying the time required to complete the NHPT and number of blocks completed during the BBT.

4.1.

Prefrontal Cortical Activation

The arm-specific average HbO concentrations for children with HCP and TD children during the respective conditions are shown in Fig. 3. The overall impression is that the average HbO concentration in the PFC is greater for the children with HCP across the respective conditions. The primary statistical results that are presented in the following sections are shown in Table 3.

Fig. 3

Comparison of arm-specific concentration of oxygenated hemoglobin (HbO) between TD children and children with HCP while performing the easy, moderate, and difficult shape-matching conditions. The data represent the combined activity across the right and left hemispheres of the prefrontal cortices. The concentration of HbO appeared to be highest for the difficult condition followed by moderate and easy condition. Furthermore, it is apparent that the concentration of HbO in the prefrontal cortices was highest for the children with HCP for all of the respective conditions. Error bars are standard error of mean.

Table 3

Main and interaction effects for all outcomes. Only significant results are shown.

There was a significant condition main effect ($P=0.001$; $d=1.31$). Posthoc analyses revealed significant difference in the HbO ($P=0.01$; $d=2.46$) between easy ($0.07\pm 0.01\mu \mathrm{mol}$) and difficult ($0.15\pm 0.01\mu \mathrm{mol}$) conditions. Similarly, there was also a significant difference ($P=0.01$; $d=2.41$) between moderate ($0.10\pm 0.01\mu \mathrm{mol}$) and difficult ($0.15\pm 0.01\mu \mathrm{mol}$) conditions. Altogether, these results indicate that the complexity of the shape-matching task promoted the expected changes in the HbO within the PFC.

We found a significant group main effect ($P=0.001$; $d=3.12$) indicating that the children with HCP ($0.17\pm 0.01\mu \mathrm{mol}$) had greater HbO than the TD children ($0.06\pm 0.006\mu \mathrm{mol}$) overall when completing the respective shape-matching task conditions. There also was a significant group by condition interaction ($P=0.025$; $d=0.76$). Posthoc analyses revealed a significant difference in HbO between the TD children and children with HCP during the easy ($P=0.022$; $d=2.86$), moderate ($P=0.006$; $d=3.92$), and difficult ($P=0.001$; $d=5.69$) task conditions (Fig. 4). Hence, indicating that the children with HCP had a larger concentration of HbO while performing the respective task conditions when the data from the respective arms were combined. Despite this finding, the $\text{group}\times \text{arm}\times \text{condition}$ interaction term was not significant ($P=0.11$).

Fig. 4

Comparison of concentration of the average oxygenated hemoglobin (HbO) between TD children and children with HCP while performing the easy, moderate, and difficult shape-matching conditions. These results are from the combined data from the respective arms and right/left hemispheres of the prefrontal cortices. The concentration of HbO was highest for the difficult condition followed by moderate and easy condition. Furthermore, it is apparent that the concentration of HbO in the prefrontal cortices was highest for the children with HCP for all of the respective conditions. Error bars are standard error of mean. * denotes $P\le 0.05$.

There was also a significant arm main effect ($P=0.001$; $d=1.32$) with the greater HbO when children performed the task with the nondominant/affected arm ($0.13\pm 0.01\mu \mathrm{mol}$) than the dominant/less affected arm ($0.08\pm 0.008\mu \mathrm{mol}$). There was a significant group by arm interaction ($P=0.001$; $d=1.44$). Posthoc analyses revealed a significant difference ($P=0.001$; $d=2.56$) in HbO when the task was performed with the affected arm ($0.22\pm 0.03\mu \mathrm{mol}$) of children with HCP and the nondominant arm of TD children ($0.05\pm 0.008\mu \mathrm{mol}$). Similarly, there was a significant ($P=0.001$; $d=0.67$) difference in HbO concentration when the task was performed with the less affected arm ($0.12\pm 0.01\mu \mathrm{mol}$) of children with HCP and the dominant arm ($0.06\pm 0.009\mu \mathrm{mol}$) of TD children. There was also a significant difference ($P=0.001$; $d=1.62$) in HbO between the affected ($0.22\pm 0.03\mu \mathrm{mol}$) and the less affected ($0.12\pm 0.01\mu \mathrm{mol}$) arm of children with HCP.

There was no significant main effect for the respective sides of the fNIRS probe ($P=0.91$). Hence, indicating that the activity in the respective hemispheres was equivocal.

4.2.

Task Performance

There was a significant group main effect ($P=0.001$; $d=11.1$) for the number of shapes matched, with TD children matching more shapes ($8.02\pm 0.2$ shapes) than the children with HCP ($5.2\pm 0.3$ shapes). There was a significant condition main effect ($P=0.001$; $d=5.3$). Posthoc analyses indicated that children matched a greater number of shapes in easy ($8.13\pm 0.3$ shapes) than moderate ($6.53\pm 0.3$ shapes; $P=0.001$; $d=5.3$) and difficult ($5.10\pm 0.3$ shapes; $P=0.001$; $d=10$) conditions. There was a significant arm main effect ($P=0.001$; $d=3.6$), indicating that the number of shapes matched by the dominant/less affected arm ($7.28\pm 0.3$ shapes) was greater than what was completed by the nondominant/affected arm ($6.21\pm 0.3$ shapes). There also was a significant group by arm interaction ($P=0.004$). Posthoc analyses revealed significant difference ($P=0.001$; $d=9.5$) in the number of shapes matched by the affected arm of children with HCP ($4.1\pm 0.4$ shapes) was less than what was matched by the nondominant arm of TD children ($7.9\pm 0.4$ shapes). Similarly, the number of shapes matched with the less affected arm of children with HCP ($6.3\pm 0.4$ shapes) was significantly less ($P=0.004$; $d=4.5$) than the number completed by the dominant arm of TD children ($8.1\pm 0.4$ shapes). Lastly, for the children with HCP the number of shapes matched by the affected arm ($4.1\pm 0.4$ shapes) was significantly ($P=0.001$; $d=5.5$) less than the number of shapes completed for the less affected arm ($6.3\pm 0.4$ shapes). None of the other interaction terms were significant ($P>0.05$).

4.3.

Reaction Time

There was a significant group main effect for RT ($P=0.001$; $d=6.5$), indicating that overall the TD ($0.9\pm 0.05\mathrm{s}$) had a faster RT than the children with HCP ($2.31\pm 0.3\mathrm{s}$). None of the other main effects or interaction terms were significant ($P>0.05$).

4.4.

Task Errors

There was a significant group main effect ($P=0.001$; $d=8.4$), indicating that the TD children ($1.4\pm 0.2$ errors) had fewer errors during the shape-matching tasks than children with HCP ($4.6\pm 0.5$ errors). There also was a significant arm main effect ($P=0.01$; $d=2.6$), signifying that there were fewer task errors for the dominant/less affected ($2.26\pm 0.39$ errors) and the nondominant/affected arms ($3.44\pm 0.52$ errors). None of the other main effects or interaction terms were significant ($P>0.05$).

4.5.

Nine-Hole Peg Test

There was a significant group main effect ($P=0.001$; $d=6.4$), showing that the TD children ($41.03\pm 1.9\mathrm{s}$) were faster at completing the NHPT than the children with HCP ($96.1\pm 12.0\mathrm{s}$). There also was a significant arm main effect ($P=0.007$; $d=2.5$) indicating that the NHPT was completed faster with the dominant/less affected arm ($53.46\pm 5.28\mathrm{s}$) than the nondominant/affected arm ($78.44\pm 11.64\mathrm{s}$). There was a significant $\text{group}\times \text{arm}$ interaction ($P=0.01$). Posthoc analyses revealed children with HCP were significantly ($P=0.001$; $d=6.2$) slower at the NHPT when they used affected arm ($130.5\pm 19.8\mathrm{s}$) compared with when the TD children used their nondominant arm ($42.6\pm 2.6\mathrm{s}$). Similarly, the children with HCP were significantly slower ($P=0.004$; $d=4.3$) when they used less affected arm ($68.8\pm 9.2\mathrm{s}$) compared with when the TD children used their dominant arm ($39.5\pm 2.9\mathrm{s}$). In addition, the children with HCP performed the NHPT significantly ($P=0.03$; $d=4.0$) slower with their affected arm compared with their less affected arm.

4.6.

Box and Block Test

There was significant group main effect ($P=0.001$; $d=5.9$) indicating that overall the TD children ($33.03\pm 1.3$ blocks) moved more blocks than and children with HCP ($20.5\pm 2.7$ blocks). There was no significant arm main effect ($P=0.11$) or interaction ($P=0.06$).

4.7.

Correlation Analyses

There was a positive correlation between the concentration of HbO and the RT for the easy ($r=0.25$, $P=0.06$), moderate ($r=0.43$, $P=0.04$), and difficult ($r=0.72$, $P=0.01$) complexity levels. Overall, these relationships implied that a greater concentration of HbO in the PFC during the sequential shape-matching task was associated with a slower RT.

There was a positive correlation between the concentration of HbO easy ($r=0.34$, $P=0.014$), moderate ($r=0.39$, $P=0.023$), and difficult ($r=0.43$, $P=0.012$) complexity levels and the performance on the NHPT. Hence, the children that performed slower on the NHPT also tended to have a greater concentration of HbO in the PFC during the sequential shape-matching tasks.

Lastly, there was a moderate negative correlation between the concentration of HbO during the easy ($r=-0.35$, $P=0.010$), moderate ($r=-0.38$; $P=0.005$), and difficult ($r=-0.45$, $P=0.001$; Fig. 5) complexity levels and performance on the BBT. This result suggests that children that moved fewer blocks tended to have greater concentration of HbO in the PFC during the sequential shape-matching tasks.

Fig. 5

Scatter plot of the PFC activation while performing the difficult shape-matching condition and BBT. Abscissa represents number of blocks that children completed and the ordinate represents average oxygenated hemoglobin (HbO) concentration. There was a moderate negative correlation ($r=-0.45$, $P=0.001$), which suggests that children that moved fewer blocks tended to have greater concentration of HbO in the PFC during the difficult shape-matching condition.