Stroke is a heterogeneous disorder wherein survivors experience varied motor recovery trajectories and functional outcomes. There is lack of consensus regarding the positive neural signatures of poststroke functional recovery1 that likely stems not only from the heterogeneity of stroke but also from a limited understanding of functional, task-specific neural activation patterns. In the current issue, Lim and Eng2 provide a proof of concept study in which they employed functional near-infrared spectroscopy (fNIRS) to investigate neural activation pattern differences in sensorimotor cortical regions poststroke. The principal study findings have important potential clinical applications. First, fNIRS-based characterization of changes in blood oxygen levels provides an indirect marker of neural activity that is safe, noninvasive, and portable. Second, using fNIRS, results show similar activation profiles during unimanual and bimanual training that could inform clinical decision-making in stroke rehabilitation. Third, employing fNIRS during dynamic, functionally-relevant tasks holds promise for closing the translational gap for functional imaging from the laboratory to the clinic.

Understanding the key neurobiological factors that promote or impede response to rehabilitation interventions for individual patients is necessary to maximize functional recovery poststroke.1,3 Training-induced neural reorganization, including changes in cortical activation patterns, is thought to contribute to functional recovery.4 Therefore, changes in neural activation patterns associated with response to therapeutic interventions and improvements in motor function are important, and can be captured with noninvasive neuroimaging techniques.5,6

Functional magnetic resonance imaging (fMRI) and electroencephalography (EEG) are examples of research-based neuroimaging techniques that enable the study of movement-related brain activity during task performance. fMRI offers excellent spatial resolution with few known side effects but provides indirect measure of neural activity, has several contraindications, requires a fixed head position, and is expensive. EEG measures neural activity directly, offers high temporal resolution, has fewer contraindications than fMRI, and is inexpensive.6 However, EEG has low spatial resolution and typically requires minimal head movement. Due to these limitations, laboratory-based motor tasks often focus on simple, single joint movements that may not translate to the coordinated, context-dependent, multijoint motor skills required for activities of daily living. fNIRS, like fMRI, is a hemodynamic measure sensitive to changes in regional cerebral blood flow and the resultant regional changes in oxygenation of hemoglobin.5 These changes are then used as a proxy for neural activity based on the assumption that increased activation utilizes energy and therefore requires an influx of oxygen to maintain or replenish energy stores within active regions. Although not insensitive to motion artifact, fNIRS is less restrictive, more portable, and less costly than fMRI and can provide enhanced spatial resolution compared with EEG. Thus, fNIRS could offer a wearable imaging modality in the clinic to index brain activity during functional activities.5

Several studies have investigated advantages of unimanual versus bimanual training to address arm dysfunction poststroke with conflicting results.7 However, the neural correlates of such approaches have not been comprehensively characterized, especially in a functional context. The current findings suggest increased ipsilesional activity correlated with maximum grip strength but not performance on the Box and Blocks test, which raises questions regarding what changes in oxygenation during task performance reflect. As the authors point out, task-related increases in oxygenation in stroke could reflect greater effort, reduced neural efficiency, and/or greater task demand. Interpretation and clinical application of the current findings are influenced by the cross-sectional experimental design, limited sample size, prolonged poststroke duration, and differences in age between groups. However, the current findings largely corroborate previous literature supporting the notion that therapeutic decision-making in stroke rehabilitation should be determined by the needs of the individual patient.

Previous studies have utilized other imaging modalities and shown global, bilateral increases in sensorimotor activation compared with healthy individuals that often correlate with poorer motor performance after stroke.8 Also, normalization of brain activation patterns has been correlated with training-related improvements in paretic arm function.4 The current study employed a cross-sectional experimental design and was performed in the chronic stage of recovery, so statements regarding progression of function cannot be made. However, similarity of findings regarding increased ipsilesional activation in the stroke group support investigating the capacity for fNIRS to evaluate task-related cortical activity patterns longitudinally. These patterns may then provide an objective biological marker of recovery of poststroke function.

Although fNIRS is a potentially promising clinical imaging tool, there are limitations. Most notably, fNIRS relies on the hemodynamic response to provide an indirect assessment of neural activity that is dependent on neurovascular coupling. Given the indirect nature of the fNIRS signal, the "increased activation" in ipsilesional sensorimotor regions, represented by a larger area under the curve following task initiation for the change in oxygenated hemoglobin, may indicate a slower hemodynamic response time, a delayed process of cortical inactivation, or a longer recovery period for active tissues in a lesioned brain. In addition, abnormal neurovascular coupling dynamics could also contribute to observed differences in fMRI and fNIRS studies in stroke. Hemodynamic measures are also unable to differentiate changes in inhibitory and excitatory signaling. Thus, increased oxygenated hemoglobin could result from increased inhibitory signaling that may downregulate activation of associated cortical circuits of interest. fNIRS is also limited in its measurement of deep-lying brain structures because of the lack of the signal detection for subcortical tissue and is subject to potential signal contamination due to blood flow in the scalp.9 Furthermore, systemic changes in mean arterial blood pressure and heart rate have been found to falsely indicate cortical activation using fNIRS.10 To mitigate the poor temporal resolution of imaging techniques dependent on the hemodynamic response, combined multimodal imaging paradigms could be used to maximize the strengths and minimize the shortcomings of fNIRS.6

Despite its shortcomings, fNIRS provides opportunities to address previously unanswered clinical questions. Information gathered with fNIRS or other clinically feasible imaging approaches (eg, EEG) has clear diagnostic, prognostic, and therapeutic potential. Identifying signatures of brain activation associated with abnormal movement patterns could expand the repertoire of evaluation tools available to clinicians.5,6 In the future, one might even envision fNIRS or other brain imaging approaches becoming part of physical therapy training and ultimately implemented as part of the standard of care for rehabilitation of stroke and other neurologic conditions.

REFERENCES