Assessment of cortical activity, functional connectivity, and neuroplasticity in cerebral palsy using functional near‐infrared spectroscopy: A scoping review

Abstract

Aim

To map and critically appraise the literature on the feasibility and current use of functional near‐infrared spectroscopy (fNIRS) to assess cortical activity, functional connectivity, and neuroplasticity in individuals with cerebral palsy (CP).

Method

A scoping review methodology was prospectively registered and reported following Preferred Reporting Items for Systematic review and Meta‐Analysis Extension for Scoping Reviews (PRISMA‐ScR) guidelines. A systematic search was conducted in four databases. Empirical studies using fNIRS to assess neural activity, functional connectivity, or neuroplasticity in individuals with CP aged 3 years or older were included.

Results

Sixteen studies met the inclusion criteria. Individuals with CP (age range = 3–43 years; 70% unilateral CP) underwent fNIRS‐based assessment for task‐evoked activity (studies [n] = 15) and/or resting‐state functional connectivity (n = 3). Preliminary observations suggest greater magnitude, extent, and ipsilateral hemispheric lateralization of sensorimotor cortex activity in CP, while magnitude and patterns of prefrontal cortex activity in CP appear dependent on task demands. Normalization of fNIRS‐based activity metrics observed postintervention (n = 3) paralleled improvements in functional outcomes, highlighting their potential as promising biomarkers for functional gains in CP.

Interpretation

This review details the use of fNIRS in CP, highlights research gaps and technical limitations, and offers recommendations to support fNIRS implementation for ecologically valid functional neuroimaging in individuals with CP.

Plain language summary: https://onlinelibrary.wiley.com/doi/10.1111/dmcn.16286

Abbreviations

BCP

bilateral cerebral palsy

CIMT

constraint‐induced movement therapy

fNIRS

functional near‐infrared spectroscopy

MACS

Manual Ability Classification System

MEG

magnetoencephalography

PEDI‐CAT

Pediatric Evaluation of Disability Index – Computer Adaptive Test

PFC

prefrontal cortex

SMC

sensorimotor cortex

UCP

unilateral cerebral palsy

What this paper adds

• Functional near‐infrared spectroscopy (fNIRS) is feasible for assessing task‐evoked brain activity and functional connectivity in cerebral palsy.

• Greater sensorimotor cortex activity scales with lower gross motor and manual abilities.

• Prefrontal cortex activity patterns appear dependent on task complexity and demands.

• Changes in fNIRS‐based metrics may accompany functional improvements following therapeutic intervention.

• Potential for real‐world brain imaging with fNIRS remains largely untapped in cerebral palsy.

A majority of children with cerebral palsy (CP) present abnormal findings on neuroradiological examinations, ¹ with lesions identified through structural magnetic resonance imaging (MRI) often corresponding to patterns and distribution of sensorimotor impairments. For example, focal vascular lesions are commonly observed as perinatal stroke in unilateral CP (UCP), ² while diffuse bilateral white matter injury (e.g. periventricular leukomalacia) is most frequent in bilateral CP (BCP), and grey matter lesions are often associated with dyskinetic CP. ³ However, at least 10% to 15% of children with CP have normal MRI, ² , ⁴ and abnormal findings do not consistently relate to clinical outcomes in CP ⁵ , ⁶ because of the complex interactions between lesion type, timing, location, and extent. ⁷ , ⁸ , ⁹ , ¹⁰ The exaggerated potential for neuroplasticity in the early years of life (Kennard principle), ¹¹ , ¹² individual genetic constitutions ¹³ and epigenetic variance, ¹⁴ and lived experiences further contribute to the heterogeneity in brain structure–function relationships in CP. ¹⁵ , ¹⁶ , ¹⁷

Functional neuroimaging can provide greater insights into neurophysiological dynamics during functional tasks. ¹⁸ , ¹⁹ The most commonly used functional neuroimaging modality in CP is functional MRI, ²⁰ which measures changes in blood oxygenation levels (blood oxygen level dependent response) to capture brain activity during tasks (task‐evoked activation; see, e.g. Phillips et al. ²¹ ), and identify functionally connected brain regions at rest (resting‐state functional connectivity; see, for example, Doucet et al. ²² ). Functional MRI has been used in CP to investigate adaptive plasticity following intervention, ²³ , ²⁴ and resting‐state functional connectivity in the sensorimotor ²⁵ , ²⁶ and language neural networks ²⁷ and their association with clinical outcomes. ²⁸ Alternative modalities, such as electroencephalography (EEG) and magnetoencephalography (MEG), offer superior temporal resolutions compared to MRI and directly capture electrical activity in neuronal populations. Studies using EEG in CP reported reduced activity at specific frequencies (mu‐band ²⁹ ) in the ipsilesional sensorimotor cortex (SMC) during upper limb tasks (reach‐to‐grasp, ³⁰ isometric wrist extension, ³¹ hand squeezing ³² ) and heightened bilateral activity during treadmill walking. ³³ Altered dynamics of the somatosensory, ³⁴ visual, ³⁵ , ³⁶ , ³⁷ and sensorimotor cortices ³⁸ identified through MEG studies in CP shed light on impaired motor planning ³⁹ and execution, with MEG outcomes potentially sensitive to change following intervention. ⁴⁰ , ⁴¹

Despite these promising observations, practical concerns (e.g. scanner constraints, signal noise, and high costs) limit the use of these modalities in children. ⁴² , ⁴³ While efforts to improve scanner‐based experiences have increased acceptance and completion rates, ⁴⁴ , ⁴⁵ concerns persist about the effectiveness of in‐scanner strategies to reduce movement artifacts, ⁴⁶ and of motion correction strategies during data processing. ⁴⁷ These issues are magnified in individuals with CP, who exhibit cognitive, visuospatial, and sensorimotor impairments ⁴⁸ , ⁴⁹ that render them unable or unwilling to remain still, sustain attention, or comply with task instructions during repetitive experimental assessments. ⁵⁰ For example, children with even mild manual impairments struggle to comply with simple hand‐squeezing tasks during functional MRI. ³² Additionally, individuals with ataxic or dyskinetic CP who experience involuntary movements are underrepresented in neuroimaging studies, ⁵¹ resulting in selection bias and limited generalizability of neuroimaging findings. ⁵² These observations highlight the need for child‐friendly, non‐invasive functional neuroimaging tools that can incorporate engaging and ecologically valid methods.

Functional near‐infrared spectroscopy (fNIRS) is a relatively inexpensive, portable, and non‐invasive functional neuroimaging tool with great potential for assessing brain activity in CP. Like functional MRI, fNIRS relies on neurovascular coupling, or increased blood flow (reactive hyperemia) to active neural areas in response to metabolic demand, ⁵³ to provide an indirect measure of neural activity. ⁵⁴ , ⁵⁵ Near‐infrared light of specific wavelengths (optical window, 650–900 nm ⁵⁶ ) is emitted through optodes placed in contact with the scalp, typically through a flexible cap (Figure 1). This light penetrates 1 to 2 cm into the underlying tissue, being reflected, scattered, and absorbed by oxyhemoglobin and deoxyhemoglobin in the outer 5 to 10 mm rim of the cerebral cortex. ⁵⁷ Emergent light is captured by detectors, and attenuation of light intensity is used to quantify relative changes in hemoglobin concentrations through the modified Beer–Lambert law. ⁵⁸ Cortical activity is indirectly reflected by an increase in oxyhemoglobin and a concurrent, smaller decrease in deoxyhemoglobin concentration, with a net increase in total hemoglobin concentration. ⁵⁹

FIGURE 1.

(a) Schematic illustration of the principles of fNIRS measurement. Near‐infrared light passes through scalp connective tissues, undergoing scattering, reflection, and absorption along its probabilistic banana‐shaped path through cortical tissue, before being reflected out to be captured by detectors placed at the scalp. (b) Decreased light intensity is mathematically modelled (modified Beer–Lambert law) using estimates of tissue concentrations (C) and light scattering (G), absorption (ε), and refraction (DPF, differential pathlength factor) to obtain estimates of relative changes in oxyhemoglobin (HbO) and deoxyhemoglobin (HbR) concentration. These changes are surrogate markers of cortical activity, with modern software providing real‐time feedback of cortical activity and functional connectivity at rest or during functional task performance.

Features of fNIRS that make it particularly suited for use in children and adults with neurodevelopmental disorders such as CP are detailed in Table 1. Briefly, fNIRS provides higher spatial resolution than EEG and better temporal resolution than functional MRI, allowing accurate localization and temporal characterization of cortical activity. ⁶⁰ Wireless fNIRS devices enable mobile neuroimaging in natural settings, ⁶¹ facilitating ecologically valid tasks that optimize engagement, ⁶² an especially important consideration in individuals with CP exhibiting cognitive and attention deficits. ⁴⁸ Notably, functional task contexts provoke improved postural control ⁶³ and altered cortical activation patterns in those with CP compared with conventional experimental tasks such as finger tapping. ⁶⁴ Additionally, fNIRS offers greater methodological flexibility, customizable optode template arrangements, ⁶⁵ low cost, and minimal maintenance, making it an ideal neuroimaging tool in low‐resource settings of low‐ and middle‐income countries that show greater prevalence of CP. ⁶⁶

TABLE 1.

Non‐invasive functional neuroimaging modalities.

Modality	Typical resolution (spatial; temporal) a	Movement tolerance; safety; cost	Setting	Additional comments
fMRI	3–12 mm (high); 2–4 s (low)	Low; b high; high	Scanner‐based; loud scans with limited limb motion	Indirect measure of neural activity (BOLD response with ΔHbR only). Whole‐brain and white matter tractography. Can detect subcortical neural activity.
MEG	3–5 mm (high); <1 ms (very high)	Low; high; very high	Scanner‐based via helmet with helium‐encased sensors; silent scan	Direct measure of neural activity (ultra‐minute magnetic fields, 10−15 T). No reference electrode. Whole‐brain and subcortical neural activity detection.
EEG	≥1 cm (low); ≤1 ms (very high)	Low–moderate; high; low	Semi‐mobile via cap‐based using gel adhesive; silent scan	Direct measure of neural activity. Reference electrode needed; cap and gel‐induced discomfort. Whole‐head with high‐density templates. Sensitive to contraction and motion artifact.
fNIRS	0.5–2 cm (moderate); 10 ms–1 s (moderate)	Moderate–high; very high; low	Freely mobile if via wireless cap‐based optodes; silent scan	Indirect measure of neural activity (neurovascular coupling: ΔHbO and ΔHbR). Whole‐head with high‐density templates. Low optical penetration limits detection of activity to cerebral cortex. Discomfort c from cap and optode.

Abbreviations: BOLD, blood oxygen level dependent; EEG, electroencephalography; fMRI, functional magnetic resonance imaging; fNIRS, functional near‐infrared spectroscopy; ΔHbO, change in oxyhemoglobin concentration; ΔHbR, change in deoxyhemoglobin concentration; MEG, magnetoencephalography.

^a Not theoretical limits, but dependent on acquisition and processing protocols.

^b New strategies in development to improve tolerance to movement (e.g. real‐time head motion feedback in cerebral palsy ²¹² ).

^c System‐dependent.

A major drawback of fNIRS is its limited penetration depth, which prevents the assessment of neural activity below the outer 5 to 10 mm of cortical grey matter. This precludes neuroimaging of deep grey‐matter structures (e.g. basal nuclei, thalami) or deeper temporal lobe structures (e.g. hippocampi) that are involved in more extensive lesions in CP. ⁶⁷ Another shortcoming is the susceptibility of fNIRS signal to contamination from task‐induced and/or spontaneous changes in systemic physiology such as fluctuations in heart rate, respiration, blood pressure, or autonomic activity. ⁶⁸ A significant portion of near‐infrared light is absorbed by superficial, extracerebral tissues that are sensitive to non‐neuronal physiology, ⁶⁹ necessitating additional signal processing to account for this contamination. Technical challenges arise with fNIRS data collection in individuals with thick or curly hair that can hinder scalp‐optode contact. Additionally, individuals with darker skin experience reduced penetration depth of near‐infrared light, ⁷⁰ , ⁷¹ potentially introducing systemic biases in fNIRS results. ⁷² The fNIRS community is aware of these limitations, ⁷³ with efforts to address technical shortcoming, ⁷⁴ , ⁷⁵ improve accessibility, and tackle equity challenges in fNIRS research spurring initiatives such as the BRIGHT project, which uses fNIRS to assess neurodevelopment in infants in The Gambia. ⁷⁶

Previous reviews of neuroimaging findings in CP have primarily focused on structural brain lesions ⁷⁷ , ⁷⁸ or connectivity, ⁵¹ with task‐based functional neuroimaging reviews generally limited to conventional scanner‐based modalities such as functional MRI. ¹⁸ , ⁷⁹ While several reviews have detailed the use of fNIRS in typically developing children ⁸⁰ and those with other neurodevelopmental disorders, ⁸¹ , ⁸² a comprehensive review specifically addressing the potential, limitations, and applications of fNIRS in CP is lacking, which may discourage adoption of this modality in research and clinical settings. ⁷³ To address this, this scoping review aims to map and critically appraise the literature on the feasibility and current use of fNIRS to assess cortical activity, functional connectivity, and neuroplasticity in individuals with CP.

METHOD

A scoping review was conducted to synthesize existing knowledge on the use of fNIRS neuroimaging in CP, ⁸³ , ⁸⁴ map current practices, and highlight research gaps. ⁸⁵ In accordance with evidence‐based principles, ⁸⁶ five key steps were included: research question identification, literature identification, study selection, data extraction, and evidence synthesis. Results were reported following Preferred Reporting Items for Systematic review and Meta‐Analysis Extension for Scoping Reviews (PRISMA‐ScR) guidelines. ⁸⁷ The study protocol was registered on the Open Science Framework (https://osf.io/f3u8b).

Research question(s)

How has fNIRS been used to assess cortical activity, functional connectivity, or neuroplasticity in children and adults with CP? What are the methodological characteristics (sample characteristics, experimental protocols, processing pipelines and analyses algorithms) of studies that have used fNIRS in CP? What are the feasibility, potential, and limitations related to the use of fNIRS in CP?

Identification of relevant studies

A systematic search was conducted in PubMed, Web of Science, CINAHL (via EBSCOhost), and PsycINFO (via EBSCOhost) following the JBI format ⁸⁸ (Table 2). Key terms were piloted on PubMed for sensitivity in detecting relevant studies, with the search strategy (Table S1) finalized in consultation with an academic librarian and translated across databases. Hand searching strategies included reviews of reference lists and citation tracking using Google Scholar and Lens. ⁸⁹ No date or language restrictions were applied, and searches were updated on 18th July 2024. In line with previous scoping reviews, ⁹⁰ , ⁹¹ , ⁹² grey literature (Table S2), abstracts, conference proceedings, and opinion pieces were not included, to focus on empirical, peer‐reviewed literature and prevent ‘double‐counting’ of studies.

TABLE 2.

Systematic search components.

Component	Search criteria
Population	Children and adults diagnosed with cerebral palsy (CP) of any etiological origin; includes pre‐, peri‐, or postnatal stroke; all topographical and phenomenological distributions included (unilateral–bilateral; hemi−/diplegia; no restriction on spastic–ataxic–dyskinetic–hypotonic–mixed tone presence)
Concept	Any aspect of brain activity, functional connectivity, or neuroplasticity captured using functional near‐infrared spectroscopy (fNIRS)
Context	All geographical locations; sexes and races/ethnicities included; clinical, academic, and/or research settings accepted; no language restrictions implemented

Study selection

Eligible reports were imported into EndNote (version 20; Clarivate, Philadelphia, PA, USA) and Rayyan. ⁹³ Duplicates were manually removed, and titles and abstracts were independently screened by three reviewers (OAK, SR, KB). Criteria for study inclusion were: (1) empirical human studies using fNIRS to assess brain activity, functional connectivity, or neuroplasticity; (2) studies where more than 50% of the sample comprised children (aged ≥3 years, after complete myelination in infancy ⁹⁴ ) or adults with CP. No demographic or geographical restrictions were imposed. Studies not reporting task‐evoked or resting‐state fNIRS outcomes, or those focusing on neurodevelopmental disorders other than CP, were excluded. Full‐texts were reviewed independently by two reviewers (OAK, SR), with disagreements resolved through discussion, or through consultation with a third reviewer (KB).

Data extraction

A data extraction table was designed before study selection and iteratively refined during full‐text review to ensure completeness of extracted information. ⁹⁰ Two reviewers (OAK, SR) performed data charting with pseudo‐random assignment ensuring studies from the same group were reviewed by one reviewer for consistency. ⁹¹ One reviewer (OAK) validated accuracy of data extraction for all studies.

Evidence synthesis

Data were tabulated and categorized by study background, sample characteristics, experimental procedures, fNIRS parameters, and primary findings. Research trends and clinical implications were highlighted, with recommendations proposed for future fNIRS research in CP.

Critical appraisal of included studies

In line with previous fNIRS work, ⁹⁰ study quality was appraised by three reviewers (OAK, SR, KB) using the adapted 15‐item Downs and Black assessment tool for non‐randomized studies ⁹⁵ (Table S3), with consensus achieved through discussion. Quality was rated as low (<60%), moderate (60–74%), or high (≥75%), on the basis of the proportion of criteria met.

RESULTS

From 112 studies identified through searches, 42 duplicates and 41 irrelevant studies were removed. Full‐text review of 29 studies resulted in the inclusion of 16 studies for analyses. Figure S1 illustrates the study selection process.

Study quality and background

Using the modified Downs and Black checklist (Table S4), study quality scores ranged from 53% to 100%, with an average score of 84%. Three studies ⁹⁸ , ⁹⁹ , ¹⁰⁹ scored 60% to 74% (moderate quality), one study ¹⁰⁰ scored less than 60% (poor quality), and the other 12 studies scored at least 75% (high quality). The lowest scoring item assessed external validity (‘Were participants representative of the population’, 44%), with similarly low scores for questions assessing confounding (‘Were participants recruited from the same population?’, 56%) and internal validity (‘Were participants recruited over the same period?’, 50%).

Study background and sample characteristics

Study settings (Table S5) indicate most studies (n = 13) were conducted in the USA, with two studies in China ¹⁰⁹ , ¹¹⁰ and one in Italy. ¹⁰⁸ All studies were published during the previous 15 years (2010–2024), and were in English. Most studies (n = 13) were observational, with three interventional studies including pre–post assessments ¹⁰⁶ , one including a mid‐intervention assessment, ¹⁰⁸ and another with a 6‐month follow‐up. ⁹⁹ A priori power analyses was reported in only one study, ¹⁰⁷ while two studies ⁹⁶ , ⁹⁹ used post hoc analyses to assess fNIRS's sensitivity at detecting group differences. Attrition rates (10–33%) were reported in six studies. One study excluded 20% of data (2 out of 10 participants) post hoc owing to large lesions observed in the cortical regions of interest on structural MRI. ¹⁰³

Participants' characteristics are also detailed in Table S5. Across 16 studies, 158 individuals with CP were assessed using fNIRS (sample sizes = 2–24), with 70% displaying unilateral affection (i.e. UCP). Participants' ages ranged from 3 to 43 years, with 12 studies recruiting only children, three studies including both children and adults, ¹⁰² , ¹⁰³ , ¹⁰⁴ and one pilot study assessing two adults with CP. ¹⁰⁰ Structural brain imaging confirmed brain injury in seven studies, or was inferred through inclusion/exclusion criteria. ¹⁰¹ , ¹⁰⁵ , ¹⁰⁶ , ¹⁰⁷ Nine studies described brain pathology, with six studies specifying injury type (e.g. perinatal stroke, periventricular leukomalacia) and three studies reported lesion location (e.g. subcortical or cortical lesion). ⁹⁶ , ⁹⁷ , ⁹⁹ Functional classification varied by type of experimental task. Seven of the 11 studies incorporating an upper limb task documented Manual Ability Classification System (MACS) levels. Most participants displayed mild impairment (MACS levels I or II: n = 46, 63%) or moderate impairment (MACS levels III or IV: n = 26, 36%), and one study included a participant with severe impairment (MACS level V). ¹⁰⁶ Gross Motor Function Classification System (GMFCS) levels were consistently reported across studies using a lower limb or whole‐body task, ¹⁰¹ , ¹⁰² , ¹⁰⁴ , ¹⁰⁸ , ¹¹⁰ , ¹¹¹ with most participants displaying mild to moderate impairments (GMFCS levels I–III: n = 71, 87%) and severe impairments less frequently (GMFCS level IV: n = 10; GMFCS level V: n = 1). ¹⁰⁸ , ¹¹⁰ Muscle tone abnormality (primarily spasticity) was reported in eight studies, with two studies reporting limb dystonia. ¹⁰² , ¹⁰³ Mirror movements were documented in four studies, ⁹⁷ , ⁹⁸ , ¹⁰² , ¹⁰⁴ with one study focusing on this phenomenon. ⁹⁸

Experimental procedures

Experimental procedures are summarized in Table S6. Most studies employed motor tasks, except for three studies that used cognitive‐motor dual‐tasks involving varying cognitive demands ¹⁰⁵ , ¹⁰⁶ or concurrent postural challenge. ¹⁰⁷ Upper limb tasks included unimanual protocols such as finger tapping, ⁹⁶ , ⁹⁷ , ⁹⁸ , ⁹⁹ ball grasp‐and‐drop, ¹⁰⁰ and shape‐matching, ¹⁰⁵ , ¹⁰⁶ , ¹⁰⁷ as well as bimanual protocols such as ball grasp‐and‐squeeze and simulated pouring ¹⁰³ and machine‐assisted arm cycling. ¹⁰⁹ One study combined unimanual and bimanual tasks. ¹⁰⁴ Lower limb tasks included functional mobility (treadmill walking, ¹⁰¹ robot‐assisted gait training, ¹⁰⁸ functional strength [progressive lateral step‐up] test ¹¹¹ ), and seated protocols such as passive cycling, ¹¹⁰ active cycling, ¹⁰² and single‐joint movements. ¹⁰² , ¹⁰⁴ Only one study combined upper‐ and lower‐limb tasks. ¹⁰⁴

Major study findings

The major research questions and primary findings of each study are presented in Table S7.

Task‐evoked sensorimotor cortex activity in CP

Early fNIRS studies highlighted altered patterns of local hemodynamic activity in children with UCP, with lower time‐to‐peak activity ⁹⁷ and time‐to‐peak activity/total activation duration ratios ⁹⁶ consistently observed at 2 months and sensitive at distinguishing UCP from age‐matched typically developing children. ⁹⁶ , ⁹⁷ The same research group identified contributions of mirror movements to bilateral SMC activity in UCP during unimanual tasks, ⁹⁸ with methods proposed to isolate these signal contaminants. Age‐related variability in SMC hemispheric laterality was greater in children with UCP, who displayed bilateral SMC activation, unlike contralateral SMC activity displayed by typically developing children older than 7 years. ⁹⁷ Following intensive constraint‐induced movement therapy (CIMT), acute changes in local hemodynamics in children with UCP were sustained at 6‐month follow‐up, while acute change in SMC laterality (increased contralateral SMC activity) was not maintained, and was related to better unimanual, but worse bimanual, function. ⁹⁹ Heightened SMC activity was also observed in adolescents and young adults with UCP during bimanual tasks, ¹⁰³ with asymmetric tasks evoking more exaggerated and lateralized SMC activity that were linked to increased muscle co‐activation and better daily function on the Pediatric Evaluation of Disability Index – Computer Adaptive Test (PEDI‐CAT), respectively.

Few fNIRS studies assessed SMC activity in individuals with BCP displaying bilateral affection. One exploratory study ¹⁰¹ reported greater SMC and superior parietal lobule activity in BCP (n = 4) than controls during treadmill walking. Increased activity in the combined groups was related to greater variability in temporal gait parameters, suggesting heightened neural demands to maintain ongoing gait trajectories. Another study reported greater SMC activity in older adults with BCP during lower limb tasks, ¹⁰² with activity scaling with higher GMFCS levels and greater muscle activation. Increased SMC activity during non‐dominant ankle dorsiflexion was also related to worse selective motor control, and lower walking ability and mobility scores on the ABILOCO and PEDI‐CAT, respectively. The only study comparing SMC activity in UCP and BCP ¹⁰⁴ reported excessive activity in both groups compared with typically developing individuals, with activity scaling with increasing functional impairment (higher MACS, GMFCS levels). Differences between groups with CP were task‐dependent. During non‐dominant hand squeeze, SMC activity was greatest in individuals with UCP and those with more impaired manual abilities (MACS level III), with an positive relationship observed between SMC activity and MACS level. During non‐dominant ankle dorsiflexion, SMC activity was greatest in individuals with BCP and those with more impaired gross motor function (GMFCS level III), with a similar positive association observed between SMC activity and GMFCS level.

Task‐evoked prefrontal cortex activity in CP

The earliest fNIRS investigation of prefrontal cortex (PFC) activity in CP reported similar temporal patterns of hemodynamic activity in two adults with BCP as neurotypical individuals during a ball‐grasp and drop task. ¹⁰⁰ However, PFC laterality differed between the groups, with both individuals with BCP exhibiting ipsilateral PFC dominance, in contrast to the bilateral‐to‐contralateral PFC dominance observed in the neurotypical group. In children with UCP, no hemispheric difference in PFC activity was noted during a shape‐matching task, ¹⁰⁵ although overall PFC activity was greater than controls. Group differences were more pronounced with the more‐affected arm, increasing task difficulty, ¹⁰⁵ and during dual‐task conditions with a dynamic postural challenge of ball‐sitting. ¹⁰⁷ Greater PFC activity during the dual‐task was also linked to greater dual‐task cost in children with UCP. Following intensive CIMT, ¹⁰⁶ reduced PFC activity in children with UCP was comparable to baseline levels in typically developing children, but links between PFC activity and functional improvements were not reported.

Recent fNIRS studies assessing PFC activity during whole‐body tasks such as robot‐assisted gait training ¹⁰⁸ and functional strength (progressive lateral step‐up) test ¹¹¹ revealed task‐dependent patterns. Perpetuini et al. ¹⁰⁸ reported contrasting changes in PFC activity across hemispheres, but no significant changes in SMC activity following robot‐assisted gait training in children with BCP. Cortical activity changes were evident only at the end of the 4‐week intervention, suggesting a dose‐dependent neuroplastic response in children with more severe gross motor impairment. Licea et al. ¹¹¹ observed suppressed PFC activity in children with CP compared with typically developing children across all levels of a progressive lateral step‐up test, even after controlling for task performance differences. However, no significant association between PFC activity and step‐up task performance was observed in the group with CP.

Resting‐state functional connectivity in CP

Three fNIRS studies assessed resting‐state functional connectivity in CP. ⁹⁹ , ¹⁰⁹ , ¹¹⁰ Cao et al. ⁹⁹ noted lower frequency of functional connections in the pre‐, supplementary, and primary motor cortices in children with UCP following intensive CIMT, which was linked to improved functional outcomes. However, changes were observed only in those with mild–moderate manual ability impairment (MACS level II, no change in those classified in MACS level I) and were not sustained at 6 months. More recent studies ¹⁰⁹ , ¹¹⁰ reported similar reductions in intra‐ and interhemispheric functional connectivity in the PFC and SMC of children with CP at rest, during assisted arm‐cycling ¹⁰⁹ and passive leg bicycling. ¹¹⁰ Decreased interhemispheric SMC connectivity in UCP was also strongly related to worse gross motor ability (higher GMFCS level). Zhang et al. ¹⁰⁹ reported greater resting‐state activity in the dominant SMC which remained unchanged during passive‐assisted arm cycling, unlike the increased activity observed in controls. More advanced network analyses by Xie et al. ¹¹⁰ also revealed decreased global and local neural network efficiency in individuals with CP, both at rest and during passive bicycling, alongside decreased strength of motor‐prefrontal connections from the non‐dominant motor cortex.

DISCUSSION

Settings and sample characteristics

Most studies included in this review arose from research conducted over the past 15 years, with nearly all originating from North America. No studies from low‐ or middle‐income countries were identified, despite these regions accounting for up to 98% of the global caseload of CP in children under 5 years ¹¹² with significantly higher CP‐related morbidity. ¹¹³ These constrained study settings reflect equity challenges in fNIRS research ⁷³ , which also presents researchers with opportunities to explore collaborations with not‐for‐profit organizations (e.g. the Bill and Melinda Gates Foundation's Brain Imaging for Global Health [BRIGHT] Project ¹¹⁴ , ¹¹⁵ ) and communities in more diverse settings. ⁷⁶ The low cost and portability of fNIRS make it a particularly promising neuroimaging modality for use in these underserved regions.

Despite small sample sizes (range = 2–24; median = 8), individuals with CP across a wide age range (3–42 years), different topographies (UCP, BCP), and varying levels of functional impairment (GMFCS, MACS levels) were assessed, broadly supporting the feasibility of fNIRS use in CP. Spastic CP was the most frequently reported subtype across studies, with dystonia reported in two studies. ¹⁰² , ¹⁰³ Notably, no study included individuals with the less common dyskinetic or ataxic subtypes of CP, despite their distinct clinical phenotype ¹¹⁶ and evidence of altered neural connectivity. ¹¹⁷ , ¹¹⁸ None of the studies were preregistered, ¹¹⁹ and only one study reported a priori power analyses for sample size estimation, ¹⁰⁷ possibly reflecting the incipient nature of the field. Adequate powering of neuroimaging‐focused studies to detect interventional neuroplasticity in CP is an important issue, with specific sample size recommendations proposed for some MRI‐based outcomes. ¹²⁰ While similar guidelines for fNIRS studies are currently lacking (but see Schroeder et al. ¹¹⁹ for general discussion), future fNIRS research should consider adopting standardized experimental protocols, ¹²¹ consistent preprocessing pipelines, ¹²² and engaging in collaborative initiatives such as the ManyBabies 3 NIRS project ¹²³ to facilitate data pooling across studies.

Additionally, only 10 individuals with significant gross motor impairment (GMFCS levels IV–V; ~6% of total participants with CP) were assessed across two studies, ¹⁰⁸ , ¹¹⁰ despite this group comprising close to 30% of all individuals with CP. ¹²⁴ This underrepresentation highlights ongoing equity challenges in neuroimaging research in CP, restricting generalizability of findings and marginalizing individuals already less likely to access and benefit from evidence‐based interventions. ¹²⁵ A similar issue was observed for upper limb assessments in individuals with UCP, with only four individuals classified at MACS levels IV or V included across two studies, ¹⁰⁵ , ¹⁰⁶ while MACS levels were not reported in three of 11 studies using upper limb tasks. ⁹⁶ , ¹⁰⁰ , ¹⁰⁹ As fNIRS outcomes can vary across functional abilities on both the GMFCS and MACS, ¹⁰⁴ future fNIRS studies in individuals with CP should include these descriptors to improve generalizability of findings and facilitate aggregation of results.

Experimental protocols

Most studies incorporated tasks that were either functional (e.g. shape‐matching, seated cycling, walking) or resembled real‐world behavior (e.g. simulated pouring). Only three studies incorporated whole‐body tasks, ¹⁰¹ , ¹⁰⁸ , ¹¹¹ and upper limb tasks were performed in seated or reclined positions that may not reflect real‐world behavior. Two studies employed experimental protocols with movements that were either fully passive ¹¹⁰ or active‐assisted. ¹⁰⁹ Functional MRI work in UCP demonstrated that passive movements evoke lower cortical activation than self‐generated active movements, ¹²⁶ and passive modalities do not reflect the task‐focused, child‐driven, and activity‐based principles of evidence‐based best practices for CP neurorehabilitation. ¹²⁷ While the great potential of fNIRS to assess neurophysiology during real‐world tasks in unconstrained environments remains largely untapped, ⁶¹ recent efforts to integrate more functional assessment protocols such as robot‐assisted walking ¹⁰⁸ and a progressive lateral step‐up test ¹¹¹ are promising, and may guide future fNIRS research in CP.

Sensorimotor cortex activity

Unilateral CP

Most fNIRS studies in UCP reported exaggerated SMC activity, with potential contributors including mirror movements, ⁹⁸ lower functional abilities (higher MACS levels), ¹⁰⁴ greater task complexity (asymmetric vs. symmetric bimanual or unimanual tasks), ¹⁰⁴ and deficits in upper limb selective voluntary motor control. ¹⁰³ Functional MRI studies in UCP reported similar heightened SMC activity during impaired hand movement, ¹²⁶ , ¹²⁸ including increased bilateral ¹²⁶ and ipsilateral (contralesional) SMC activity, ¹²⁸ with the latter related to residual hand function and strength of mirror movements. ¹²⁸ Mirror movements may reflect altered neurophysiology due to early developmental injury in UCP, ¹²⁹ and may have potential clinical implications such as reduced bimanual function. ¹³⁰ , ¹³¹ Despite early fNIRS work suggesting mirror movements contribute to increased SMC activation in UCP, ⁹⁸ their presence and impact was inconsistently reported. Future studies should report clinically observable mirror movements, with quantitative tools available for more detailed analyses. ¹³²

Studies assessing hemispheric lateralization of SMC activity in UCP reported age‐related variability, ⁹⁷ bilateral activation, ⁹⁹ or trends of ipsilesional SMC dominance during unimanual ¹⁰⁴ and bimanual tasks. ¹⁰³ The functional relevance of SMC lateralization in UCP was less studied, although contralesional SMC dominance during asymmetric bimanual squeezing was related to better daily function, ¹⁰³ while SMC activity during non‐dominant hand squeezing and ankle dorsiflexion was not related to functional outcomes. ¹⁰⁴ Variability in activation patterns reflects methodological differences across studies and aligns with functional MRI literature in UCP that describes three main lateralization patterns: ¹⁸ bilateral dominance in motor functions, ipsilesional dominance in somatosensory functions, and contralesional dominance in language functions. Ipsilesional SMC activation involves the recruitment of preserved perilesional tissue, and generally correlates with better clinical outcomes in UCP ¹³³ and stroke. ¹³⁴

The diverse SMC lateralization patterns reflect the nuanced dynamics of cortical activation in UCP, a complexity further highlighted by intervention studies. The sole fNIRS study reporting SMC changes postintervention found increased contralesional SMC dominance in UCP immediately after CIMT ⁹⁹ that declined at 6 months and was related to worse bimanual function. ⁹⁹ A review of interventional neuroplasticity reported increased ipsilesional SMC activity as the most common change observed postintervention in UCP. ²³ Conversely, a recent functional MRI study reported increased ipsilesional and decreased contralesional SMC activity after hand–arm bimanual intensive therapy including lower extremities (HABIT‐ILE) intervention, with better clinical outcomes related to reduced brain activity. ¹³⁵ Lateralization of SMC activity in UCP reflects adaptive developmental processes after early developmental injury (see Eyre ¹³⁶ and Friel et al. ¹³⁷ for detailed review) that vary by lesion size ¹³⁸ and timing. ¹³⁹ Large lesions may prevent ipsilesional SMC control, rendering contralesional activity the sole contributor to motor function. ⁷⁹ Laterality also varies by side tested and fatigue. ¹⁴⁰ A multimodal neuroimaging study highlighted unique neurophysiological patterns across individuals with UCP, ³² emphasizing significant individual variability and highlighting the challenge of defining clear neural structure–function relationships in this population.

Bilateral CP

While limited, fNIRS studies in BCP offer valuable insights into this group's distinct neurophysiology. Increased SMC and superior parietal lobule activity during treadmill walking in an exploratory study in children with BCP ¹⁰¹ mirrors the excessive motor and parietal cortex activation observed in a larger EEG gait study in UCP. ³³ Similar to fNIRS reports in UCP, ¹⁰⁴ greater SMC activation in BCP was related to lower functional abilities (higher GMFCS levels), ¹⁰⁴ worse selective motor control, and lower PEDI‐CAT mobility scores. ¹⁰² However, unlike UCP, where greater SMC activation is seen during non‐dominant hand squeezing than ankle dorsiflexion, the BCP cohort demonstrated similar SMC activity across these tasks, ¹⁰⁴ potentially indicating more severe upper extremity involvement in this group. ¹⁴¹ This is supported by EEG evidence of exaggerated dominant SMC activation in BCP during reaching ¹⁴² , ¹⁴³ that was also associated with slower, less efficient movements and poor dexterity, suggesting SMC over‐recruitment may reflect excessive cortical resource use during arm movements in BCP. ¹⁴²

Hemispheric lateralization of SMC activity in BCP was task‐dependent, with ipsilateral SMC dominance during non‐dominant ankle dorsiflexion ¹⁰² and contralateral dominance during non‐dominant hip movements, cycling, ¹⁰² and hand squeezing. ¹⁰⁴ This pattern contrasts with the bilateral activation commonly noted in UCP, where underlying brain lesions are often focal, well‐defined, and distinctly impacted by the timing and location of insult, such as in perinatal stroke. ¹⁴⁴ The more widespread bilateral brain injury in BCP results in more severe motor impairments and higher incidences of comorbidities such as intellectual and visual impairments, ¹⁴⁵ that further complicates task performance during neuroimaging studies. Reorganization following bilateral lesions probably follows different developmental patterns than those seen after unilateral injuries, owing to the absence of a relatively intact hemisphere that can act as a scaffold to support adaptive plasticity. ¹⁴⁵ A recent functional MRI study in mildly impaired children with BCP ¹⁴⁶ reported contralateral SMC dominance during non‐dominant ankle dorsiflexion, although a shift to greater ipsilateral dominance and lower activation volume postintervention were independently associated with motor skill gains. These observations highlight the potential for targeted interventions to influence cortical reorganization and improve motor outcomes in BCP. However, the scarcity of functional neuroimaging studies in BCP is a significant barrier to understanding the complexity and variability of neural structure–function relationships in this group. Mobile neuroimaging tools such as fNIRS, with their methodological flexibility and resilience to motion artifacts, are well‐suited to addressing this research gap.

PFC activity

Greater PFC activity in UCP during a shape‐matching task scaled with increased task difficulty, non‐dominant arm use, ¹⁰⁵ and concurrent postural challenge. ¹⁰⁷ After CIMT intervention, PFC activity in CP attenuated to levels comparable to typically developing children, ¹⁰⁶ but no link to improved functional outcomes was reported, generating doubt on whether attenuated PFC activity reflected ‘normalization’ of cortical activity or was the byproduct of generalized motor skill acquisition. The PFC is a prime target for interrogation by fNIRS because of its accessible location underneath the hairless forehead region, allowing for optimal signal acquisition. ¹⁴⁷ Although the PFC is not involved in motor execution, it mediates executive functions such as working memory, sustained attention, and action planning that support and enable motor planning and prediction. ¹⁴⁸ , ¹⁴⁹ Exaggerated PFC activity during motor tasks may represent a resources allocation strategy ¹⁵⁰ for enhanced motor planning in UCP. ¹⁵¹ In contrast, children with BCP with greater motor involvement displayed more variable PFC activity patterns after robot‐assisted gait training, ¹⁰⁸ although the small, heterogenous sample and absence of a comparison group limits definitive conclusions on the functional impact of these changes.

The variability in PFC activity changes following different motor interventions highlights a need for further investigation into the neural correlates of cognitive deficits and their impact on motor outcomes in CP. Half of all children with CP are estimated to display concurrent intellectual disability, ¹⁵² with a recent meta‐analysis identifying moderate–large deficits across all executive function domains in individuals with CP, regardless of gross motor or manual ability (i.e. GMFCS and MACS levels, respectively). ¹⁵³ Given the intimate linkage of cognitive and motor development ¹⁵⁴ and their combined relation to the PFC, ¹⁵⁵ cognitive deficits in CP significantly impact their ability to navigate real‐world tasks that involve dual‐tasking. ¹⁵⁶ , ¹⁵⁷ A recent MEG study in adults with CP ¹⁵⁸ reported weaker PFC oscillatory activity during the encoding phase of working memory was related to worse cognitive outcomes and lower gross motor function (higher GMFCS levels). Despite the potential for fNIRS to assess PFC activity during ecologically valid tasks in real‐world settings, fNIRS studies assessing PFC activity during cognitive tasks or during physical activity are lacking. ¹⁵⁹ Notably, the sole fNIRS study to incorporate a physically demanding task reported suppressed PFC activity in children with CP during a progressive lateral step‐up test ¹¹¹ which was maintained after controlling for their lower performance. However, the lack of significant associations between PFC activity and step‐up performance in CP suggests factors such as lower exercise tolerance or psychological factors such as impaired attention and emotional dysregulation may contribute to the suppressed PFC activity patterns observed in this group. These novel observations emphasize the need for future fNIRS research to explore how PFC activity impacts cognitive–motor interactions in CP.

Functional connectivity and cortical networks

Early fNIRS‐based functional connectivity research ⁹⁹ revealed greater frequency of connections between the supplementary motor, premotor, and primary motor cortices in children with UCP, with most connections observed from the supplementary motor area, although connection strength was not quantified. Conversely, recent fNIRS studies ¹⁰⁹ , ¹¹⁰ reported lower intra‐ and interhemispheric resting‐state functional connectivity in the PFC and SMC of children with UCP, with attenuated connectivity maintained during assisted arm ¹⁰⁹ and passive leg cycling, ¹¹⁰ and related to lower gross motor function. Reduced functional connectivity across the sensory and motor areas was also widely reported in the functional MRI literature in UCP, ⁵¹ although intrahemispheric connectivity in these regions may vary by pattern of corticospinal tract wiring. ¹⁶⁰ Lower efficiency of both global and local cortical networks in UCP reported in a recent fNIRS study ¹¹⁰ mirrored observations from studies using structural MRI ¹⁶¹ and diffuse tensor imaging ¹⁶² reporting similar lowered global efficiency in children with BCP.

Notably, fNIRS functional connectivity research did not include studies of individuals with BCP, in whom functional MRI revealed widely varying patterns of functional connectivity across cortical regions. Significantly widened and increased connections reported with functional MRI in BCP between the somatosensory cortices, ¹⁶³ SMC, and supplementary motor areas ¹⁶⁴ contrasts against decreased connectivity in the bilateral SMC and parietal cortices. ¹¹⁸ , ¹⁶⁵ Functional connectivity analyses have also been used to examine neural networks mediating non‐motor functions, with evidence of altered networks in CP extending to domains of language, ²⁷ visuomotor function, ¹⁶⁶ and cognition. ²⁶ These domains represent rich avenues for future research, with recent development of wearable, high‐density fNIRS devices ¹⁶⁷ setting the stage for fNIRS investigation of these unexplored topics in CP.

fNIRS methodologies

While fNIRS is a promising tool for real‐world neuroimaging, its outcomes are susceptible to subjective decisions by researchers ¹⁶⁸ on issues of signal quality assessment, ¹⁴⁷ , ¹⁶⁹ , ¹⁷⁰ data processing pipelines, ¹⁷¹ , ¹⁷² and methods of statistical analyses. ¹⁷³ Concerns about potential data misreporting have spurred a movement towards greater transparency in decision‐making and reporting within the fNIRS community, with calls for preregistration of study protocols and algorithms for analyses decision‐making ¹¹⁹ to mitigate publication bias and false positive results. ⁶⁹ , ¹⁷⁴ While assessing methodological rigor and modality‐specific challenges of fNIRS exceeds the scope of this review (see Yücel et al. ¹⁷⁵ for comprehensive recommendations), we provide a detailed summary and critical appraisal of fNIRS methods in Table S8 and Appendix S1, respectively. Further, as fNIRS shares neurovascular underpinnings (blood oxygen level dependent response) with functional MRI, readers are referred to the excellent review by Reid et al. ²⁴ for a detailed examination of challenges in interpreting activation changes in functional neuroimaging.

Limitations

This review aimed to inform future research by synthesizing peer‐reviewed empirical literature on the application of fNIRS in CP, and excluded grey literature and non‐scientific articles. While this approach is consistent with other scoping reviews, ⁹² , ¹⁷⁶ , ¹⁷⁷ it may have resulted in the omission of additional experimental settings and observations. However, we provide readers with a list of relevant doctoral dissertations and theses obtained through a systematic search (see Table S2). The inclusion of studies with small sample sizes and methodological limitations (e.g. the pilot study by Chaudhary et al. ¹⁰⁰ on two adults with CP) was necessitated by the nature of scoping reviews. Quality appraisal analyses were performed to address concerns about potential bias, although we acknowledge that caution is warranted when interpreting the study results. Finally, the diversity of fNIRS outcomes across studies prevented aggregation and quantification by meta‐analyses, emphasizing the need for standardized experimental protocols for easier data pooling and comparisons across studies. ⁵¹ Despite these limitations, this review provides a comprehensive overview on the use of fNIRS in CP that can inform future investigations in this field.

Challenges with fNIRS use in CP

Despite its advantages as a user‐friendly, motion‐resistant, and cost‐effective functional neuroimaging modality, fNIRS remains underutilized in CP neuroimaging research, with only 16 studies identified in this review. Alongside the aforementioned limitations such as poor imaging depth, inability to image deeper brain structures, systemic contamination, and data collection challenges with darker skin tones or thick hair, other potential barriers also require recognition. A lack of research confirming fNIRS reliability in CP undermines confidence in its ability to track intervention effects on cortical function. Recruitment and compliance issues are frequent, with early studies that used wired systems reporting high attrition rates (10–33%) and exclusion of data (20% in one study ¹⁰³ ) owing to cortical lesions in brain regions of interest. Practical challenges include prolonged setup times, resistance to fNIRS caps from sensory hypersensitivity, and participants' discomfort, compounded by the lower signal‐to‐noise ratio of fNIRS than functional MRI, necessitating longer and more frequent trials. ⁸⁰ The physiological underpinnings of fNIRS also present challenges in CP. Neurovascular coupling, the basis of fNIRS measurements, may be altered in CP owing to brain injuries such as periventricular leukomalacia or perinatal stroke, ¹⁷⁸ which impair vascular autoregulation ¹⁷⁹ and integrity. ¹⁸⁰ This disrupts hemodynamic signals and complicates interpretation of cortical activity. ¹⁸¹ Additionally, historical reliance on conventional neuroimaging modalities such as functional MRI and EEG, coupled with limited awareness of recent advances in fNIRS technology, has slowed the adoption of fNIRS in CP research.

However, rapid improvements in fNIRS hardware and software offer solutions to many of these challenges. ¹⁸² Recent innovations such as infant‐friendly dual‐tip optodes, ¹⁸³ scalable modules for high‐density whole‐head imaging in infants, ¹⁸⁴ and customizable optode attachments have significantly improved feasibility. Implementing standardized fNIRS methodologies, ¹⁸⁵ analyses pipelines, ¹²² , ¹⁸⁶ and reporting guidelines ¹⁷⁵ can further enhance reliability and reproducibility. Targeted efforts to establish fNIRS reliability in CP and adapt this technology for neurodiverse populations could significantly expand its utility in understanding neurophysiology and intervention effects in CP.

Future directions

The exponential rise in fNIRS studies in neurodevelopmental disorders and neurodivergent populations ¹⁸⁷ , ¹⁸⁸ , ¹⁸⁹ , ¹⁹⁰ , ¹⁹¹ , ¹⁹² reflects the growing recognition of fNIRS as a valuable tool for investigating brain function in neurodivergent populations for whom conventional neuroimaging is challenging. ¹⁹³ Rapid technical, methodological, and technological developments in fNIRS have propelled the field beyond initial feasibility assessments ¹⁹⁴ , ¹⁹⁵ , ¹⁹⁶ towards the identification of diagnostic and prognostic biomarkers (‘fNIRS signatures’ ¹⁹⁷ ) of significant translational value to clinicians and researchers. Future research avenues for exploration include the integration of fNIRS with other neuroimaging modalities such as functional MRI, ¹⁹⁸ EEG, ¹⁹⁹ MEG, ²⁰⁰ , ²⁰¹ and transcranial magnetic stimulation, ²⁰² , ²⁰³ to provide real‐time feedback of functional performance and progress during rehabilitation. ²⁰⁴ This multimodal approach may also uncover underlying pathophysiology ²⁴ , ³² , ²⁰⁵ and enhance confidence in reported results by reducing bias, and minimizing signal noise. ⁷³ Hyperscanning protocols that enable synchronous assessment of multiple interacting individuals in naturalistic settings are particularly well‐suited for fNIRS, and hold promise for studying parent–child interactions. ²⁰⁶ This innovative approach may shed light on the neurophysiological underpinnings of impaired interpersonal social interactions in CP, ²⁰⁷ and promote family‐centered care by facilitating parental involvement in therapy. ²⁰⁸ Additionally, the integration of fNIRS into brain–computer interface applications, ²⁰⁹ either alone or in conjunction with other modalities such as EEG, ¹⁹⁹ , ²¹⁰ may offer tangible benefits to individuals with CP who experience severe motor, language, or cognitive impairments. ²¹¹

The overarching goal of translational fNIRS research is to integrate fNIRS into clinical settings for individualized assessment, prognosis, and monitoring of treatment fidelity and effectiveness in clinical populations of all ages and functional profiles. Achieving these goals in the context of CP requires: (1) continuous advances in fNIRS hardware and software; (2) transparent and comprehensive reporting of experimental methods and results; (3) collaboration between funding agencies, industry, researchers, and community partners; and most importantly, (4) sustained involvement of individuals with lived experience and their families, who stand to benefit most from this research. This review comprehensively summarizes the current state of fNIRS research in CP and represents a critical first step towards realizing these ambitious objectives.

CONCLUSION

This review analyzed 16 studies that confirmed the feasibility and utility of fNIRS for evaluating cortical activity, functional connectivity, and neuroplasticity in individuals with CP, with exaggerated SMC activity observed across motor tasks alongside task‐dependent PFC activity patterns. While fNIRS demonstrated utility in capturing neuroplastic changes postintervention, the lack of reliability data undermines confidence in its application in interventional research. While most studies demonstrated moderate–strong methodological quality, specific challenges to fNIRS use were identified, with recommendations for enhanced rigor and transparency outlined. Leveraging recent technical advancements may help address these challenges and transition fNIRS research beyond feasibility towards integration into clinical practice. By enabling individualized assessments and real‐time monitoring, and facilitating family‐centered care through improved neurophysiological understanding of parent–child interactions, fNIRS offers transformative potential for CP research and rehabilitation that remains largely untapped.

FUNDING INFORMATION

The Eunice Kennedy Shriver National Institute of Child Health and Human Development (RO1 HD090126) and the University of Georgia Athletic Association.

CONFLICT OF INTEREST

The authors have no interests which might be perceived as posing a conflict or bias.

Supporting information

Appendix S1: fNIRS methodology.

Figure S1: PRISMA‐ScR flow diagram.

Table S1: Systematic search strategy used for the PubMed database.

Table S2: Relevant grey literature from ProQuest database search and hand‐searching.

Table S3: Modified Downs and Black checklist for methodological quality assessment.

Table S4: Methodological quality scores based on the modified Downs and Black checklist.

Table S5: Study background and sample characteristics.

Table S6: Experimental procedures.

Table S7: Research questions, sample characteristics, and primary findings

Table S8: Technical specification for fNIRS data collection and processing.

Khan OA, Rahman S, Baduni K, Modlesky CM. Assessment of cortical activity, functional connectivity, and neuroplasticity in cerebral palsy using functional near‐infrared spectroscopy: A scoping review. Dev Med Child Neurol. 2025;67:875–891. 10.1111/dmcn.16238

DATA AVAILABILITY STATEMENT

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.