Introduction

Cerebral palsy encompasses a heterogeneous group of neurodevelopmental conditions that primarily present as disorders of movement and posture, often accompanied by epilepsy, secondary musculoskeletal problems, and impaired sensation and cognition.1 Symptom onset occurs during early childhood, typically before 18 months of age;1 on average, diagnosis is confirmed at 13–19 months.2,3,4 The most common form of the condition is unilateral cerebral palsy (UCP), which impairs the use of one hand and consequently disrupts bimanual co-ordination.

Cerebral palsy, by definition, results from abnormal brain development and/or brain damage that is non-progressive and occurs during very early development. In most cases, the cause is periventricular white matter damage that is presumed to occur during the third trimester of pregnancy, but other abnormalities, such as diffuse grey matter injury, focal infarcts, lesions of the basal ganglia, and/or cerebral malformations, can underlie the condition.5,6,7 Early brain injury that can underlie cerebral palsy can lead to atypical brain development and reorganization, particularly during the first 2 years of life,8 which can complicate the understanding of the condition and the selection of appropriate rehabilitation.

The current definition of cerebral palsy includes the word 'permanent', but notes that “initial disruption to normal brain structure and function ... may be associated with changing or additional manifestations over time.”1 In children with UCP, the type and extent of impairment is primarily determined by the location and size of the brain lesion.9 The ability of patients to gain functionality with therapy might be influenced by comorbidities, such as impaired vision and concentration, learning difficulties, and epilepsy.10

We commence this Review by summarizing current evidence for the efficacy of existing rehabilitative therapies for children with UCP, and by discussing factors that can increase effectiveness, such as the mode, dose, context, relevance and timing of intervention, as well as the child's motivation. We then discuss how future therapies for children with UCP could be informed by research into neuroplasticity in patients with the condition. We present imaging technologies that should be used in future studies to examine neuroplasticity, and the current knowledge that could direct future research that will ultimately shape the future of rehabilitation for children with this lifelong condition.

The appropriate rehabilitation for children with cerebral palsy depends on the motor subtype, the type and extent of brain damage or abnormalities, and other factors, such as age and cognitive ability. Current knowledge is largely limited to children with UCP, so we will concentrate on rehabilitation that targets the principal symptoms of this condition, that is, sensorimotor deficits. The main focus of our discussion is upper limb rehabilitation.

Existing rehabilitative strategies

Rehabilitation for children with UCP should be evidence-based, activity-based (that is, the child performs the activity themselves), relevant to the environment and the child's motivation, and goal-directed. The therapy should be delivered by the parents as a series of challenging but achievable activities, conducted in the appropriate environment and context, and designed to include specific learning goals for the child. The two most extensively investigated contemporary approaches to upper limb therapy in children with UCP are modified constraint-induced movement therapy (mCIMT) and bimanual intensive therapy (BIM; Box 1). Hybrid CIMT, in which the child trains with mCIMT before BIM, is another approach.11,12

mCIMT and BIM are increasingly being delivered to individuals and groups in real-life situations so as to provide context to motor skill learning.12 However, such therapy requires expertise, can be costly, and is not always accessible. For these reasons, the usual upper limb therapy for children with UCP includes occupational therapy together with adjunctive treatments—such as medications to reduce spasticity—that target secondary symptoms. Although occupational therapy can include mCIMT or BIM, the time devoted to these therapies in these contexts is typically a lot less than the time allotted to them in clinical trials. In response to this multifactorial challenge, alternative modes of delivery, such as web-based training and virtual reality platforms, have been developed. Early studies of web-based multimodal training have shown that these approaches can improve motor and planning skills in children with UCP.13,14,15 In addition, robot-assisted virtual reality therapy (Box 1) has been tested in pilot trials16,17,18 and larger studies,19,20 which have provided preliminary evidence that highlights the potential of such therapy for rehabilitation in children with UCP.

The efficacy of the current activity-based therapies, as determined by large systematic reviews and meta-analyses, can be used to gauge the optimal dose, environmental context and intensity of upper limb intervention for children with UCP.12,21 A meta-analysis of 42 randomized controlled trials assessed the efficacy of 14 approaches to upper limb therapy, and found moderate to strong evidence to support the use of intensive activity-based, goal-directed interventions (such as mCIMT and BIM) rather than usual care to improve the quality and efficiency of upper limb movement and to achieve individual goals.12 The conclusions drawn in this systematic review, in combination with knowledge of the factors that impede or enable implementation of therapy, have led to the current consensus on the essential elements of effective upper limb therapy. These elements are intensive structured task repetition, progressive incremental increases in difficulty, and goal-directed approaches that enhance the motivation and engagement of individuals receiving therapy.22

Current activity-based therapies are thought to provide both specific (unimanual or bimanual) and global (occupational performance) improvements in motor function. For example, when mCIMT and BIM, delivered in blocks of 60 h over 10 days, were compared in the randomized INCITE trial (n = 64),21 mCIMT was more effective at increasing the use of the impaired limb, whereas intensive BIM was more effective at improving bimanual co-ordination. This observation confirms that these training techniques have specific effects. Motor-evoked potential recruitment curves, which were measured by transcranial magnetic stimulation (TMS), were also changed more substantially by intensive group-based mCIMT than by BIM (R. N. Boyd et al., unpublished work). Delivery of intensive group-based therapy can, however, be costly, and many therapists have identified barriers to implementation.22

Another randomized trial made a direct comparison between intensive group-based hybrid CIMT over 2 weeks and a comparable amount of individual therapy distributed over 12 weeks.11,23 Improvements in bimanual co-ordination (evaluated with the Assisting Hand Assessment) and occupational performance (assessed with the Canadian Occupational Performance Measure) were similar between the two groups. These results show that with a constant total dose of activity-based therapy, similar results can be achieved despite variations in child:therapist ratios (1:2 or 1:1) and the way in which therapy is delivered (groups or individuals).23 The similar outcomes attained with the use of intensive group-based and individual therapy mean that therapists can choose to deliver upper limb therapy in the way that is best suited to family circumstances and characteristics of the child. In this trial, the willingness of children to independently persevere, solve problems, master challenging tasks, and use their impaired hand for bilateral activities contributed to occupational performance outcomes.24 Family ecology and the therapeutic context were also identified as critical factors that influenced the children's mastery motivation and engagement in therapy.25

The long-term retention of functionality gained from upper limb therapies is not frequently assessed, but some evidence suggests that gains in functionality can last for at least 6–12 months after therapy in children with UCP.26,27,28

The majority of research into the effectiveness of rehabilitation and the mechanisms that underlie responses to therapy has been conducted with school-aged children with UCP, but major brain growth and development occurs in the first 2 years of life. This period could represent a critical window during which rehabilitation might be most effective, but which is missed by modern rehabilitative approaches.29 Empirical evidence for the effectiveness of currently used therapies in infants with UCP is limited, but animal studies have provided evidence for a critical period.30,31,32 For example, in kittens, inactivation of the primary motor cortex in one hemisphere during postnatal weeks 5–7 results in seemingly permanent impairment of contralateral motor skill, and a variety of abnormalities in neural organization.32 Motor training and limb restriction similar to CIMT can reverse these motor deficits when carried out during early development (8–13 weeks of age), but is much less effective when carried out at 20–24 weeks of age.32

Such critical periods of development might primarily reflect time windows during which abnormal neural organization can be prevented, rather than periods during which the brain is simply more able to repair existing damage.33 One pilot study (n = 5) of lower limb rehabilitation has indicated that intense rehabilitation is feasible and effective in children with UCP aged <2 years, and a follow-up clinical trial is in progress.29 Whether this approach is equally feasible and efficacious in upper limb rehabilitation is currently under investigation.34,35

Neuroplasticity-informed rehabilitation

Behavioural and outcomes data have improved intervention strategies for children with UCP by providing insight into the ways in which learning and skill development are influenced by factors such as the dose and intensity of therapy, as well as its relevance to daily life. Although intervention strategies such as mCIMT and bimanual training have been derived from and improved by these data, strategies and improvements based on a mechanistic understanding of the brain remain largely unexplored in children with UCP. Exploitation of this opportunity for the development of rehabilitation strategies requires greater knowledge of the neurobiology of UCP and the neuroplastic potential possessed by children with this condition.

Two complementary arms of research are likely to inform and shape future rehabilitation strategies for people with UCP. The first is the investigation into patterns of atypical brain development. The structural abnormalities seen in the brains of children with cerebral palsy reflect more than focal damage to a single system: early focal brain damage probably influences the development of related systems in utero, influences and modifies critical developmental periods, and has further effects in later life owing to adaptive and impaired behaviours.32,33,36,37 The second arm is the characterization of therapy-induced neuroplasticity; this approach aims to determine how and when brain networks respond to therapy. In combination, these research arms might not only identify networks that are commonly impaired or maladapted in UCP, but also determine how different initial insults lead to differences in the ability of the brain to respond to rehabilitation. For example, future research might reveal that the limited effectiveness of movement-based rehabilitation in some children with UCP is explained by indirect factors, such as attentional deficits or disrupted network dynamics that prevent neuroplastic changes.38,39,40

With these points in mind, one could reasonably surmise that translational neuroscience might suggest not only treatments that improve existing skills, but also pretraining regimens that aim to improve the learning capacity of children and/or to 'reset' maladaptive brain states in the same way that mCIMT aims to 'reset' the learned disuse of the impaired hand. One example of such a pretraining regimen is the use of TMS to enhance the effects of CIMT—an approach that has been used in one small-scale randomized controlled trial.41 In this trial, 19 children aged 8–17 years with UCP received mCIMT alongside either repetitive TMS or sham TMS. Improvements in bimanual co-ordination (evaluated with the Assisting Hand Assessment) were significantly greater in children who received TMS than in children who received sham TMS, although the researchers noted that the results could have been complicated by differing baseline scores between the treatment groups.

In other fields, brain mapping and neurostimulation studies have informed rehabilitation. For example, the use of TMS has revealed that maladaptive neuroplasticity—seen as increased contralesional sensorimotor activity—can undermine recovery from stroke in adults.42 In addition, transcranial direct current stimulation (tDCS) has been used with some success to restore the interhemispheric balance of activation in the sensorimotor system and thereby improve clinical upper limb motor scores43,44,45,46 and acquisition of motor skills.47 These findings led to an investigation in which children with spastic diplegia underwent treadmill training and concurrent anodal tDCS over the primary motor cortex contralateral to the dominant side of the body.48 Some improvements in gait variables and functional mobility were reported, and were maintained for at least 1 month.48

A similar approach is paired associative stimulation, in which peripheral sensory stimulation is paired with TMS of the motor cortex; this approach is thought to induce long-term potentiation in sensorimotor networks.49 This method has been shown to induce changes in motor-evoked potentials (MEPs) in healthy adults 60 min after stimulation49 and in healthy children 75 min after stimulation.50 After a single session of stimulation, these potentials returned to baseline after 24 h in adults.49 The fact that this method was well tolerated in healthy children suggests that it could be suitable for children with UCP.

Measuring neuroplasticity

Although the atypical patterns of brain organization and the potential neuroplasticity in people with UCP are not yet well understood, several findings hint at specific targets for future therapies.

Measurement of brain organization and therapy-driven neuroplasticity can be beneficial in clinical trials beyond simply contributing to an understanding of how and for whom rehabilitation is effective. For example, the measurement of baseline brain organization might enable selection of a more homogeneous cohort of participants, or better matching of participant pairs before randomization. This approach can minimize between-group differences at baseline, thereby boosting statistical power, improving interpretability and increasing clinical significance. Furthermore, the demonstration of neuroplastic changes in response to therapy provides confidence that clinically important changes in functional outcomes are an index of neurological improvement rather than a reflection of changes in task strategy between tests.

Methods for the measurement of neuroplasticity and brain organization that currently show promise are discussed below, together with some of the key insights into neuroplasticity in children with UCP that these methods have provided. Tables 1 and 2 summarize these methods and findings.

Table 1 Characteristics of noninvasive neuroimaging modalities used to measure neuroplasticity
Full size table
Table 2 Key findings in neuroimaging studies of children with UCP
Full size table

TMS and atypical brain organization

TMS is a noninvasive method that depolarizes neurons by means of strong, brief magnetic pulses. When applied over the primary motor cortex, TMS can induce measurable electromyographic responses via the corticospinal tract.51 This method is widely considered to be the gold standard noninvasive technique for mapping the corticospinal pathways, as it provides a direct measure of functional connectivity between the primary sensorimotor cortex (S1–M1) and muscle.52 For the measurement of neuroplasticity in the context of rehabilitation, several additional metrics can be obtained before and after rehabilitation, including motor thresholds, MEP recruitment curves, and transcallosal inhibition.51,53,54,55 The motor threshold is the lowest level of stimulus intensity that produces a reliable MEP in ∼50% of 10-20 consecutive stimuli.51 MEP recruitment curves characterize the amplitude of MEPs across a range of stimulus intensities, relative to the maximum amplitude of compound muscle action potentials obtained with supramaximal stimulation of the median nerve at the wrist.56

TMS provides data with high temporal resolution, but poor spatial resolution.54 The procedure can induce headaches in children, and its use is restricted by some researchers owing to concerns that it might induce seizures in people with epilepsy (a condition commonly seen in children with cerebral palsy57), although reports of seizures following TMS are rare.55 Furthermore, cortical mapping by TMS can be challenging, and the procedure can be uncomfortable for some recipients. The acceptability of using repetitive TMS in children needs to be considered alongside the benefits when planning such studies.58

TMS has yielded an interesting finding with respect to brain organization in UCP: in approximately one-third of children with the condition, the impaired hand is directly controlled by the ipsilateral hemisphere.59,60,61,62 This brain organization is particularly common among patients who exhibit mirror movements.60,61 During the first ∼18 months of typical postnatal development, interhemispheric competition reduces the number of ipsilateral corticomotor connections.8 In children, such as those with UCP, who have early brain lesions, damage to the contralateral corticomotor connections is proposed to prevent this interhemispheric competition, or alter its balance, leaving ipsilateral connections intact (Figure 1).33 Such atypical development is not necessarily mirrored in the somatosensory system, however, and TMS has revealed that this system is often still contralaterally organized in such children (Figure 1).62 Children with ipsilateral motor control of their impaired hand typically have poorer motor control than do children with contralateral motor control. A major cause of motor dysfunction in UCP could, therefore, be impaired transmission of sensory feedback from the impaired hand to the motor cortex that controls that hand.62 In addition to visual hemianopias, such impaired sensorimotor function and mirror movements probably account for much of the poor bimanual co-ordination exhibited by children with UCP.

Figure 1: The influence of periventricular lesions on corticospinal laterality.
figure 1

Images show corticospinal (blue) and thalamocortical (green) connection pathways. a | In typical development, bilateral corticospinal connections develop and are unperturbed during early development. After birth, interhemispheric competition mediated by cross-callosal connections (not shown) results in the removal of connections that project to the ipsilateral side of the body. b | During early development in individuals with cerebral palsy who have small periventricular lesions, the lesions can weaken or eliminate the corticospinal connections, which are consequently unable to compete effectively with the opposite side of the brain during later development. The result is bilateral motor cortex control of the impaired side of the body. c | During early development in individuals with cerebral palsy who have focal or large lesions, the lesions might completely eliminate corticospinal connections in one hemisphere. The result is ipsilateral motor cortex control of the impaired side of the body. If the insult occurs before development of somatosensory connections, the primary somatosensory system can still develop on the ipsilesional side of the brain, causing a discrepancy between the laterality of sensory feedback and motor signalling.

PowerPoint slide

Full size image

Diffusion MRI and thalamocortical pathways

Diffusion MRI measures the directional diffusion of water molecules, thereby enabling inferences to be made about the structural characteristics of brain tissue.63 Diffusion MRI images can be used for tractography, which elucidates white matter pathways and structural connectivity between brain regions (Figure 2). The technique can also provide measures that reveal information about the microstructural integrity and density within these white matter pathways. These measures include the tensor metrics of mean diffusivity, fractional anisotropy and radial diffusivity, as well as measures such as apparent fibre density that rely on more-modern higher-order mathematical models of tissue structure.64

Figure 2: An example of functional-MRI-guided diffusion MRI tractography.
figure 2

This method provides information about the structural connectivity of functionally relevant brain regions, and eliminates the need to register brains to templates, which is often impossible when malformations or pathology are present. Here, a child with unilateral cerebral palsy performed hand-tapping tasks during a functional MRI session. Voxels that were significantly activated (yellow) were used as the seed (starting) region for diffusion tractography. Activated voxels from several slices are overlaid here for illustrative purposes. The bottom images are axial sections at the level of the dashed line in the top images. a | Corticospinal tracts were identified as tracts that passed from the seeding region through the posterior limb of the internal capsule, then to the brainstem inferior to the pons. b | Corticothalamic tracts were identified as tracts that passed from the seeding region through the thalamus, then to the brainstem inferior to the pons.

PowerPoint slide

Full size image

Few studies to date have used diffusion measures to longitudinally assess neuroplasticity in children with UCP. Diffusion MRI has, however, demonstrated that the microstructural integrity of white matter in the corticospinal tract is lower in children with UCP than in children with typical development,65,66 and correlates with performance on a variety of clinical assessments.65,66,67 Tensor measures of white matter microstructural integrity also seem to change in response to motor training in healthy adults.68

In line with evidence from studies that used TMS and suggested that disrupted sensorimotor integration is important in UCP, studies involving diffusion MRI have indicated that damage to the thalamocortical radiations and consequent impairment of sensory feedback is a key factor that underlies impaired motor function in children with cerebral palsy.67,69,70 Diffusion MRI data also suggest that thalamocortical fibres can bypass periventricular injuries as they project to the cortex, possibly explaining how somatosensory networks can develop connections to the contralateral (impaired) hand even when white matter damage has already disrupted the equivalent corticomotor tract (Figure 1).71

EEG, MEG and motor planning

EEG and magnetoencephalography (MEG) are noninvasive methods that measure electrical activity in the brain—primarily in the cortex—via electrodes on the scalp (EEG) or magnetometers above the scalp (MEG). Both methods can be used to elucidate the timing, amplitude and direction of signals from groups of neurons that respond to stimuli, and they provide a millisecond or sub-millisecond temporal resolution.72,73 Although the spatial resolution of MEG (often in the order of single millimetres) is often higher than that of EEG, MEG is only sensitive to signals that travel tangentially to the cortical surface.73,74

One small-scale study that used MEG revealed locations in the brain that responded to tactile stimulation of the first, third and fifth digits; these sites seemed to be further apart in children with spastic motor type cerebral palsy than in children with typical development.75 Another study that used MEG reported abnormal beta-wave activity in children with spastic hemiparesis during the planning stage of a knee extension task.76 This finding suggested that motor planning, rather than muscular recruitment, might have been the limiting factor in the controlled execution of the motor task. If similar impairments are found to occur commonly in UCP, an increased therapeutic focus on motor planning and sensorimotor co-ordination, rather than motor output, might accelerate rehabilitation or improve its effectiveness in people with this condition.

EEG studies have shown that typically developing infants as young as 6–8 months exhibit different neural responses when observing goal-directed and non-goal-directed actions.77,78,79,80 Such responses of the mirror neuron system suggest that novel rehabilitation approaches that involve action observation might be possible in very young children.81,82

Imaging of grey matter and cortical thickness

Intervention-induced grey matter changes can be investigated with whole-brain, region-of-interest or voxel-based morphometry approaches, the results of which complement diffusion MRI measures of white matter tissue integrity. These approaches require a high-resolution structural MRI scan that enables reliable delineation of the boundaries between white matter and grey matter, and identification of the pial surfaces. Increases in grey matter volume or thickness are particularly interesting in relation to neurorehabilitation, as they reflect improvements that are mediated by tissue 'growth', rather than by novel use of pre-existing neural substrates.

Studies have shown that motor training in healthy adults83 and CIMT in adults after stroke84 increase the volume of cortical grey matter in brain regions that are functionally relevant to the training. In children, decreases in cortical thickness have been shown in the first 3 months after traumatic brain injury, after which some areas 'recover' and others continue to thin.85 Another study showed that the cortical thickness of the right precentral gyrus inversely correlates with upper limb motor function in children with congenital left-sided hemiplegia.86 Encouragingly for the therapy of UCP, a pilot study of young children with UCP (n = 10, mean age 3 years 3 months) found that the volume of grey matter in the sensorimotor cortex increased after 3 weeks of CIMT.87

Exactly which cellular processes underlie these grey matter changes is unknown, but a combination of glial, vascular, dendritic, synaptic and axonal changes is probable.88 Neurogenesis might also contribute, but its role is thought to be minor.88,89

fMRI and potential neural recruitment

Functional MRI (fMRI) is used in neurorehabilitation either to identify cortical areas that are active under certain conditions (an approach known as task-based fMRI that is used for cortical mapping) or to identify cortical areas that are likely to be functionally connected (an approach known as resting state fMRI). The most frequently used type of fMRI is blood oxygen level-dependent (BOLD) imaging, which relies on a detectable oversupply of oxygenated blood to localized areas during periods in which their neural activity is increased. Unlike TMS, fMRI offers whole-brain coverage in less than 10 min for each task. The final resolution of fMRI after processing is ∼3–12 mm.90 fMRI cannot, however, provide information on whether activity is excitatory or inhibitory, and its temporal resolution is inferior to that of TMS, EEG and MEG: each frame typically averages ∼2–4 s of brain activity.

To measure neuroplasticity with fMRI, scans should ideally be obtained during a motor or cognitive task, both before and after intervention, for at least 20 people per experimental group.91 Groupwise analyses can then be used to identify voxels with significantly different activity between time points, and thereby determine whether the location or area of activation has changed substantially. Conventional voxel-based group comparisons can be ill-suited to the study of children with cerebral palsy, however, owing to the heterogeneous size and location of brain lesions seen within most cohorts. This issue can be addressed by performing region-of-interest analyses that measure the interhemispheric balance of activation between the sensorimotor cortices before and after therapy in the same child with UCP.92 As fMRI is an indirect measure of neuroplasticity, all such analyses should ideally be supported by other evidence of change, such as TMS measurements, and correlated with functional improvements in a motor task that is not performed in the scanner.

In practice, conducting an fMRI study with good statistical power is a challenge, because it requires a homogeneous cohort of children with UCP who are able to perform tasks consistently in the noisy and constrained environment of the MRI scanner. In our experience, success can be improved by preparation with mock scanners, and by the presence, during experimental scanning, of people who are familiar to the child. The use of cluster-based fMRI analyses to increase the sensitivity of comparisons between pre-therapy and post-therapy conditions should be avoided, as these analyses do not provide voxel counts that are genuinely quantitative of the 'true' activation volume.93,94

Several small-scale studies have reported fMRI-detected changes in response to therapy in children with UCP. For example, in a pilot study of virtual reality therapy, cluster-based S1–M1 voxel counts were increased after therapy in three adolescents aged 13–15 years with UCP, whose Jebsen–Taylor hand function test scores also indicated meaningful improvements in speed and dexterity.95 Another study found a change in the balance of activation between hemispheres in two of four children with UCP after they received CIMT.96 One of these two children, however, showed no clinical improvement, rendering the practical relevance of such fMRI changes unclear. A similar study that included seven children with UCP highlighted the difficulties of scanning in this population: the results were mixed, and the researchers noted that standard analyses were not feasible owing to movement artefacts in the fMRI scans.97

Multimodal studies and CIMT

Multimodal neuroimaging offers unique opportunities to improve our understanding of neuroplasticity by combining complementary information. For example, such an approach could help to determine the dynamics of structure–function relationships (Figure 2). Three related studies of children with UCP used a multimodal approach to investigate brain changes that occur in response to mCIMT.98,99,100 These reports, which incorporated data from two studies with overlapping participants, used fMRI to assess how brain activation changed when a ball-squeezing task was performed with the impaired hand.

After exclusion of scans that were influenced by confounding factors, the first of these studies found no activation changes in two participants, and increased bilateral activation of the hand knob of the sensorimotor cortices in one individual aged 16 years with UCP.98 The second study assessed data from five patients aged 10–20 years with UCP: group analysis of contralateral corticospinal connections showed an increase in cluster-based S1–M1 voxel counts after mCIMT.99 In the third study, an associated analysis showed bilateral activation of S1–M1 in adolescents with corticospinal projections ipsilateral to the impaired hand; the balance of this activation shifted towards the impaired hemisphere after training.100 In the same study, CIMT increased TMS-evoked MEPs in the primary descending corticospinal connection to the impaired hand (but not to the unimpaired, constrained hand) in children with UCP, regardless of whether the impaired hand was controlled by the contralateral or ipsilateral motor cortex. MEG data acquired at the same time showed that somatosensory-evoked magnetic fields marginally increased in both groups after mCIMT.100

In combination, data from these three studies suggest that mCIMT reorganizes the sensorimotor system and/or promotes recruitment of additional grey matter, thereby enabling more-effective processing of sensory inputs and conversion of these inputs into corticospinal signals.

Conclusions

Current rehabilitation for children with UCP focuses on single-mode activity-based therapies that are selected on the basis of behavioural observations. The effectiveness of these therapies relies on the intensity and timing of treatment, the extent to which improvements transfer into the patients' daily lives, and the child's ability to maintain close attention to the training task.

Ongoing research with the aim of improving treatment outcomes is investigating therapy-induced neuroplasticity, and how large-scale atypical brain organization can result from brain injuries sustained during early life. The extent of these phenomena is currently unclear, and their clinical manifestations are poorly understood, although several findings suggest that the impairments observed in many children with UCP result from disrupted integration of sensorimotor information and motor planning. A wide variety of tools, including TMS, EEG, MEG, and diffusion, structural and/or functional MRI, could be used to measure brain structure and function in future clinical trials. Approaches that combine complementary information from more than one of these tools are also likely to be beneficial. Studies that use these tools are likely to improve our understanding of the neurobiology of UCP and the neuroplastic processes that are promoted by effective rehabilitation. This greater understanding could ultimately transform rehabilitation for this condition.

Review criteria

Searches of the authors' personal collections of literature, the PubMed database, Google, and Google Scholar were reviewed for influential and original publications that were relevant to the discussion. Search terms used were “cerebral palsy”, “hemiplegia”, “stroke”, “rehabilitation”, “CIMT”, “BIM”, “robot therapy”, “robot assisted”, “MRI”, “fMRI”, “EEG”, “TMS”, “PET”, “MEG”, “neuroplasticity”, “reorganization” and “tDCS”. References from these articles were inspected for additional material.