Functional near-infrared spectroscopy in non-invasive neuromodulation

Abstract

Non-invasive cerebral neuromodulation technologies are essential for the reorganization of cerebral neural networks, which have been widely applied in the field of central neurological diseases, such as stroke, Parkinson's disease, and mental disorders. Although significant advances have been made in neuromodulation technologies, the identification of optimal neurostimulation parameters including the cortical target, duration, and inhibition or excitation pattern is still limited due to the lack of guidance for neural circuits. Moreover, the neural mechanism underlying neuromodulation for improved behavioral performance remains poorly understood. Recently, advancements in neuroimaging have provided insight into neuromodulation techniques. Functional near-infrared spectroscopy, as a novel non-invasive optical brain imaging method, can detect brain activity by measuring cerebral hemodynamics with the advantages of portability, high motion tolerance, and anti-electromagnetic interference. Coupling functional near-infrared spectroscopy with neuromodulation technologies offers an opportunity to monitor the cortical response, provide real-time feedback, and establish a closed-loop strategy integrating evaluation, feedback, and intervention for neurostimulation, which provides a theoretical basis for development of individualized precise neurorehabilitation. We aimed to summarize the advantages of functional near-infrared spectroscopy and provide an overview of the current research on functional near-infrared spectroscopy in transcranial magnetic stimulation, transcranial electrical stimulation, neurofeedback, and brain-computer interfaces. Furthermore, the future perspectives and directions for the application of functional near-infrared spectroscopy in neuromodulation are summarized. In conclusion, functional near-infrared spectroscopy combined with neuromodulation may promote the optimization of central neural reorganization to achieve better functional recovery from central nervous system diseases.

Introduction

With the aging population, environmental changes, and other factors, the risk of neurological disorders has become more evident, resulting in an increase in the number of individuals suffering from central nervous system diseases, such as stroke, Parkinson's disease, and mental disorders. The rehabilitation of these dysfunctions is of significant concern to the quality of daily living of patients and the burden on families and countries. Recently, advances in neuromodulation techniques have facilitated the progress of brain science and have become essential for the diagnosis and treatment of neurological diseases.

Neuromodulation technology involves the utilization of physical elements such as light, magnetism, electricity, and ultrasound to modify the signal transmission of the central nervous system and then alter the excitability of the neurons and related neural networks (Krames et al., 2018). Non-invasive cerebral neuromodulation, including transcranial magnetic stimulation (TMS) and transcranial electrical stimulation (TES), has shown the potential to regulate the excitability of related functional areas and provoke reorganization of neural circuits in the brain with a variety of advantages, such as safety, adjustable parameters and targets for neurostimulation, and easy operation (Polanía et al., 2018; Uzair et al., 2022; Song et al., 2023).

Despite advances in cerebral neuromodulation techniques, several challenges remain to be addressed. One of the most critical challenges is the lack of guidance from neural circuits to determine the optimal stimulation parameters, such as the targeted cortical area (Chang et al., 2022). Moreover, the neural mechanism underlying neuromodulation is still not well understood and is currently the main area of research interest. The combination of cerebral neuromodulation technology with non-invasive neuroimaging tools allows for a better understanding of the neuromodulation effects in the brain related to behavioral performance. A set of neuroimaging modalities, including functional magnetic resonance imaging (fMRI), electroencephalogram (EEG), radioactive tracer-based positron emission tomography, magnetoencephalography, and functional near-infrared spectroscopy (fNIRS) imaging are commonly utilized for non-invasive detection of brain activities (Khan et al., 2022). These modalities offer advantages and disadvantages in terms of cost and spatial and temporal resolution. For the application combined with cerebral neuromodulation, several technical challenges of these imaging modalities need to be considered, such as limited mobility, susceptibility to motion artifacts and electric-magnetic field disruption, and confinement to laboratory environments. As a novel brain imaging tool, fNIRS offers advantages such as mobility, portability, impressive interference immunity, suitable electromagnetic compatibility, and cost-effectiveness. Coupling fNIRS with cerebral neuromodulation can be applied to dynamically monitor the neuroplastic responses during neurostimulation in terms of time and space. This makes it possible to evaluate, guide, and provide feedback on the outcomes of neuromodulation interventions in real time, thus advancing the development of precise neuromodulation in neurorehabilitation and neurocognitive augmentation (McKendrick et al., 2015).

This review aims to provide a narrative overview of the application of fNIRS in the field of cerebral neuromodulation, with a focus on TMS, TES, as well as neurofeedback and brain-computer interface (BCI) techniques. The utilization of imaging-coupled neuromodulation may provide information regarding ways to induce plastic changes in the cortex, the duration of its effects, and how it relates to enhanced performance.

Search Strategy

Articles related to fNIRS combined with a neuromodulation protocol were identified based on a search strategy performed in Medline, PubMed, and Scopus using logical combinations of the following key terms: fNIRS & neuromodulation, neurostimulation, TMS, TES, transcranial direct current stimulation (tDCS), transcranial alternating current stimulation (tACS), BCI, and neurofeedback. Additional published articles were selected through a manual search of the references in retrieved articles. These retrieved articles were further screened with restrictions on publication time (up to January 2023) and language in English. Original studies reporting on fNIRS-derived hemoglobin changes associated with neuromodulation were included.

Imaging Principle and Advantages of Functional Near-Infrared Functional Imaging

Biological tissue exhibits high scattering and low absorption in the near-infrared spectrum window (650–950 nm). This near-infrared light is capable of penetrating the tissue to a depth of 20–30 mm, thereby reaching the cerebral cortex of the skull (Chance et al., 1997). Hemoglobin is the most prominent chromophore that is physiologically dependent on absorption in biological tissues. The near-infrared light absorption spectrum demonstrates that oxygenated hemoglobin (HbO₂) and deoxyhemoglobin (HbR) can be distinguished by their light absorption characteristics (Scholkmann et al., 2014). By utilizing the modified Beer-Lambert law, the attenuation of light can be determined from the incident and outgoing light information, thus enabling real-time detection of the hemodynamics of the cerebral cortex (Ekkekakis, 2009).

Cerebral autoregulation ensures that the supply of local cerebral blood flow and oxygen metabolism rate remain in a balanced state, thus maintaining a relatively stable concentration of HbO₂ and HbR in the blood vessels. As shown in Figure 1, upon the disruption of the steady state by stimulation, the dynamic cerebral blood flow and brain metabolic rate increase significantly, and the concentrations of HbO₂ and HbR in the blood vessels are altered accordingly (Scholkmann et al., 2014). The process of oxygen metabolism in neurons involves oxygen being consumed to produce energy, leading to a decrease in HbO₂ concentration and an increase in HbR concentration (Scholkmann et al., 2014). Thus, an increase in neuronal activity is associated with a rise in local cerebral hemodynamic changes. In the brain activated area, the local cerebral blood flow rate is much higher than the local oxygen metabolism rate, resulting in a higher concentration of HbO₂ and a lower concentration of HbR (Scholkmann et al., 2014). The alteration of cerebral hemoglobin concentration is significantly associated with the activation of brain neurons (Liao et al., 2013). Thus, based on this correlation, it can be inferred that the alteration of local cerebral blood oxygen detected by fNIRS can be used to indicate the cerebral neural activity indirectly.

Figure 1.

Signals measured by fNIRS based on the physiological reactions related to neural activity.

The changes in cerebral blood oxygenation are strongly associated with neuronal activity in the brain. Local functional activation leads to a heightened regional metabolism and oxygen requirement of neurons (neurometabolic coupling), resulting in augmented regional cerebral blood flow (neurovascular coupling). The local increased blood flow is accompanied by a concomitant increase in HbO₂ and decrease in HbR, which can be measured with fNIRS. This technique is a non-invasive optical imaging modality that can detect brain functional activity by measuring the cerebral hemodynamics based on blood oxygen level-dependent changes. Created with Microsoft Office (vision 2019). fNIRS: Functional near-infrared spectroscopy; HbO₂: oxygenated hemoglobin; HbR: deoxyhemoglobin.

Studies involving multimodal neuroimaging, including fNIRS, fMRI, and EEG, have substantiated the effectiveness of fNIRS as a non-invasive neuroimaging technique (Sasai et al., 2012; Chen et al., 2015). Over the last twenty years, significant progress has been made in the methodologies and applications of fNIRS imaging technology, boasting high stability, sensitivity, and channel density. fNIRS has the advantage of being capable of simultaneously measuring two hemodynamic parameters, including HbO₂ and HbR, providing a more comprehensive description of brain function, thus increasing the accuracy and reliability of research (Tachtsidis and Scholkmann, 2016). fNIRS offers a compromise between the spatial resolution of fMRI and the temporal resolution of EEG. Additionally, it is suitable for a wide range of individuals, including infants and those with metal implants. Moreover, fNIRS has strong anti-motion interference capabilities, making it possible to monitor the brain dynamics of multiple individuals in a naturalistic setting (McKendrick et al., 2016). Most importantly, by the principle of optical imaging characteristics, the advantages of anti-electromagnetic interference make fNIRS suitable to combine with magnetoelectric neuromodulation technology, thus allowing for the dynamic assessment of brain functional activity during neurostimulation (Liu et al., 2022).

Functional Near-Infrared Functional Imaging in Combination with Neuromodulation Techniques

Combining non-invasive brain stimulation and neuroimaging technologies, it is possible to observe the responses of the local cortex and the related brain areas associated with neurostimulation. Brain functional imaging can be employed simultaneously during the implementation of neuromodulation (online) or following neuromodulation (offline). fNIRS is an economical and scalable technology to couple with non-invasive brain stimulation, thus allowing us to investigate the short and long-term neuroplastic responses to the intervention of neuromodulation (Chang et al., 2022). Figure 2 shows the fNIRS-based neuromodulation framework and its applications in the field of neuroscience.

Figure 2.

fNIRS-based neuromodulation framework and its applications in the field of neuroscience.

Coupling fNIRS with neuromodulation technologies can monitor the corresponding cortical response characteristics, including cortical activation, brain functional connectivity, effective connectivity, and functional network metrics. By inputting the brain functional response in real time to the neuromodulation techniques (i.e., TMS, TES, BCI, and neurofeedback), it is possible to adjust the parameters of neurostimulation, including frequency, time instances, stimulation durations, target, and intensity, forming a closed-loop strategy that integrates evaluation, feedback, and intervention for neurostimulation. The combination with fNIRS provides a theoretical basis for precise and targeted neuromodulation interventions in the field of cognitive augmentation, neuro-rehabilitation, and skill learning. Created with Microsoft Office (vision 2019). BCI: Brain-computer interface; fNIRS: functional near-infrared spectroscopy; ROI: region of interest; TES: transcranial electrical stimulation; TMS: transcranial magnetic stimulation.

fNIRS-based TMS studies

TMS is a non-invasive technique that utilizes electromagnetic pulses to modulate the neurons firing in the brain, thereby enabling the evaluation and regulation of brain function of patients with neurological disorders, including stroke rehabilitation, mental disorders, and many other conditions. The protocols of TMS applied in clinical rehabilitation include single-pulse, repetitive, and theta burst stimulation (TBS). Repeated TMS is capable of either facilitating excitation or inhibition of cortical activity via the mechanisms of long-term potentiation and long-term depression, respectively (Pell et al., 2011). To visualize cortical activations during repeated TMS, it is necessary to measure them at high spatiotemporal resolution while preventing interference with the magnetic properties of the coil. Currently, fNIRS has been adopted to observe the spatiotemporal changes in brain activity during neurostimulation, providing an objective and quantifiable evaluation of transient and prolonged cerebral functional responses to the repeated TMS intervention (Curtin et al., 2019). The most common approach is to place the stimulation coil directly on the fNIRS channel, considering that the magnetic field attenuation due to distance requires a corresponding increase in stimulation intensity. Furthermore, by crafting a coil structure that is tailored for the purpose, it becomes possible to seamlessly integrate it with the fNIRS source-detector optode, effectively eliminating any potential interference. The following outlines the applications of fNIRS in combination with TMS.

Based on the correlation between neuron activation and hemodynamic changes, fNIRS offers a reliable approach to monitor the alterations in cortical excitability induced by repeated TMS (Oliviero et al., 1999; Allen et al., 2007). Utilizing fNIRS, an evaluation of concomitant alterations in cerebral hemoglobin levels can be performed in the stimulation target or its associated distal brain areas. An fNIRS study has shown bilaterally decreased prefrontal oxygenation during an emotional Stroop task following inhibitory continuous TBS stimulation of the left dorsolateral prefrontal cortex (DLPFC) (Tupak et al., 2013). Inhibitory TBS over premotor cortex, primary hand motor area (M1), and primary sensory area induced significantly decreased cortical activity of the contralateral areas, indicated by a decreased HbO₂ concentration (Mochizuki et al., 2007). Furthermore, this can be combined with fNIRS to identify long-term physiological changes following TMS. The concentration of HbO₂ in the unstimulated cortex increases following the application of 1 Hz TMS over the contralateral hemisphere. This increase lasts for a period of 40 minutes after the stimulation (Chiang et al., 2007). Additionally, the application of a single intermittent TBS to the cerebellar vermis in healthy adults results in increased activation in the bilateral supplementary motor area (SMA) during balance-related activities (Tan et al., 2021).

TMS-induced alterations in neural activity depend significantly on the specific stimulation parameters utilized, including intensity, coil direction, targeted cortical region, etc. (Pashut et al., 2011). The combination of TMS and fNIRS can improve our understanding of the effects of different stimulus parameters on brain function, thus providing valuable guidance for parameter optimization. Single-pulse TMS with 90% or 110% motor threshold induced significant oxyhemoglobin increases at the stimulation site (Noguchi et al., 2003). However, significant decreases in both deoxy- and total-hemoglobin were observed in single-pulse TMS at 120% and 140% of the motor thresholds (Mochizuki et al., 2006). These findings might suggest a reduced baseline firing of the corticospinal tract neurons induced by a lasting inhibition provoked by a higher intensity TMS. Moreover, the alterations in the hemoglobin concentration following a single-pulse TMS depend on whether the target muscle was in an active or relaxed condition (Furubayashi et al., 2013). For the coil direction, Thomson et al. employed fNIRS to investigate the impact of TMS coil orientation on prefrontal hemodynamic changes, and observed significant changes in HbO₂ concentration following TMS with a 45° coil, whereas only minor alterations were observed at 135° and 225° (Thomson et al., 2013). Furthermore, the utilization of fNIRS as a target navigator for TBS modulation in individuals with hemiplegia has been proposed. According to the findings of a randomized control study, fNIRS-guided TBS has been demonstrated to be more effective for rehabilitation than motor-evoked potential guided TMS in stroke motor rehabilitation (Chang et al., 2022). This suggests that fNIRS provides the guidance of the target area for TMS, leading to improved neuronal interaction and enhanced neuronal reorganization processes in the targeted neural circuit.

The utilization of fNIRS-based measurement of cerebral hemodynamics during neurostimulation shows promise as a biomarker to determine responsiveness to TMS, as well as evaluating the effectiveness of TMS treatment (Shinba et al., 2018). The increases in frontal hemoglobin concentration during TMS are positively correlated with clinical outcomes of treatment for patients with depression (Shinba et al., 2018). Individuals showing a decreased functional response in the prefrontal cortex during TMS experience less clinical improvement, and patients with schizophrenia exhibit lower activation of the bilateral inferior parietal lobes than healthy controls following high-frequency repeated TMS on the DLPFC. Additionally, it was suggested that the reduced information processing was correlated with the abnormal fronto-parietal connectivity assessed by fNIRS in individuals with schizophrenia (Ćurčić-Blake et al., 2022). Central-peripheral paired magnetic stimulation can enhance the functional network associated with the SMA measured by fNIRS, promoting the reconstruction of neural circuits post-stroke to enhance motor function (Xu et al., 2021). The reconstruction of the descending motor pathway from the SMA to the affected limb assumes a crucial role in the restoration of upper limb motor function in stroke survivors.

Therefore, fNIRS has been shown to be a reliable tool to evaluate the immediate and lasting effects of TMS on the regional cortical activity of the stimulated areas and related cortices, independent of the afferent feedback from the peripheral neuromuscular activity. Despite the challenges and methodological issues associated with the use of concurrent TMS-fNIRS (Parks, 2013), real-time observation of the hemodynamic changes via fNIRS plays a crucial role in neural rehabilitation, helping to optimize intervention parameters, identify intervention responsiveness in individuals, and evaluate rehabilitation efficacy, providing further understanding of how TMS impacts behavior and cognition in a meaningful way.

fNIRS-based TES studies

TES involves the application of a low intensity current to the cortex through a pair of anodal-cathodal (excitation-suppression) electrodes attached to the scalp, thereby altering neural activity and regulating the neural function of the brain. With the advantage of cost-effectiveness, portability, and potential for home-based therapy, TES has gained popularity as a non-invasive neuromodulation technique in the field of several neurological disorders, such as stroke, mental disorders, and traumatic spinal cord injury (Mazzoleni et al., 2019; Kumru et al., 2020; Figeys et al., 2021). The primary patterns of TES consist of tDCS and tACS. A growing body of evidence has indicated the feasibility of fNIRS to describe the cortical activation induced by TES by measuring the intracranial hemodynamic response based on neurovascular coupling (Dutta et al., 2015). Additionally, a recent review has summarized the collective findings of TES combined with fNIRS studies focused on diverse stimulation parameters, such as time instances, stimulation durations, and stimulation targets (Hong et al., 2022). The present discussion focuses on the issues that can be resolved through the application of fNIRS in conjunction with TES.

Studies have demonstrated that fNIRS is an appropriate method to monitor longitudinal changes of cortical hemodynamics associated with behavioral improvement underlying the neural mechanisms of TES (Khan et al., 2013; Verma et al., 2019; Schommartz et al., 2020). Li et al. (2019) reported that a tDCS intervention for 4 weeks on bilateral DLPFC (2 mA, left anode and right cathode) could elicit an increase in HbO₂ levels in the bilateral prefrontal lobes during emotional judgment and working memory tasks; this was not observed in their sham stimulation group. Jones et al. (2015) reported that improved working memory performance is associated with the level of oxygenated blood levels measured by fNIRS following anodal tDCS. Narita et al. (2018) reported that long-term tDCS interventions on the left DLPFC of the anode in schizophrenic patients induces an increase in HbO₂ concentration in the left temporoparietal region, which is associated with a decline in the psychiatric scores. A recent investigation also utilized a combination of tDCS and fNIRS to explore the potential effects of tDCS, and showed that the tDCS-induced increase of the hemodynamic response in the DLPFC appears to be associated with faster decision speed during a sequential decision-making task (Schommartz et al., 2020). Berglund-Barraza et al. (2020) applied fNIRS to monitor potential changes in cortical activation patterns following a phonological working memory training protocol enhanced with an anodal tDCS stimulation to the SMA. Additionally, a follow-up study employed fNIRS to monitor the longitudinal changes in cortical hemodynamics following a 2-week-long tDCS treatment protocol for patients with knee osteoarthritis, and revealed neuromodulatory effects of tDCS on cortical pain processing (Pollonini et al., 2020).

A systems biology approach has been provided to investigate the initial transient hemodynamic response to tDCS based on fNIRS data (Arora et al., 2021). The primary consequence of tDCS to modify the cortical excitability depends on the polarity of the applied current. The results revealed that tDCS could elicit an increase in the concentration of HbO₂ in the cortex, particularly in the area of anodic stimulation, which is in line with the tDCS-fMRI study (Polanía et al., 2011). Specifically, anodal tDCS has the potential to enhance the local cerebral blood flow and activate the neurons and hemodynamic response within the stimulated brain tissue, whereas cathodal current has the opposite potential (Zheng et al., 2011; Dutta, 2015). Moreover, both anodal and cathodal stimulation with tDCS applied to the M1 could elicit a cortical hemodynamic response in the contralateral premotor cortex, SMA, and M1 (Takai et al., 2016). The anodal tDCS to the left sensorimotor cortex (SMC) during a finger sequence task resulted in significantly decreased activation in the bilateral SMCs, as observed in an fNIRS study. This finding suggests that bilateral SMCs may exhibit enhanced efficiency in performing the same motor task owing to improved neuronal transmission (Muthalib et al., 2016). Furthermore, tDCS in combination with fNIRS can be utilized to evaluate the stimulation-related changes in the brain functional network and the results showed that functional connectivity was significantly decreased during tDCS compared with pre- and post-tDCS (Yan et al., 2015).

The combination with fNIRS allows for the identification of the optimal stimulation parameters for TES, such as duration, intensity, frequency, and electrode positioning, all of which can impact the effectiveness of the neuromodulation (Jamil et al., 2017). Studies have applied fNIRS combined with functional network analysis to optimize the parameters of tDCS application (Yan et al., 2015). Ghafoor et al. (2022) used fNIRS and EEG simultaneously to compare the effects of tDCS and tACS on the bilateral DLPFC, and the results showed that the mean hemodynamic responses elicited by tACS were marginally lower than those induced by tDCS. Another study analyzed the effects of different intensities of anodal tDCS on the M1 in patients with Parkinson's disease, and indicated that tDCS over M1 resulted in an enhancement of postural responses with a superior response observed for 2 mA compared to 1 mA (Beretta et al., 2020). Furthermore, an fNIRS study was conducted to assess the cortical activation response in the SMC during training, specifically comparing the effects of a motor task during tDCS (online) versus following (offline) tDCS. The findings suggest that engaging in a motor task concurrently with tDCS would result in a more pronounced activation of the SMC, and that this effect lasted for 30 minutes after stimulation (Besson et al., 2019).

These findings suggest that fNIRS is a safe, promising, and efficient neuroimaging tool to couple with TES for the intervention of neurological disorders in natural settings (Wu et al., 2022). A set of fNIRS probes was designed based on the analysis of the electric field distribution during tDCS (Sharma and Roy Chowdhury, 2018). Appropriate coupling of the fNIRS probe and the stimulation electrode facilitates the monitoring of hemodynamic responses in the stimulated site and related areas. Additionally, the use of fNIRS short-distance channels (less than 10 mm between the source and the detector) is a crucial addition to this configuration, considering the scalp interference caused by heat and erythema under the tDCS electrode pad. However, comfort is a key element to be considered; an excessive amount of equipment on the scalp will inevitably lead to a decrease in the comfort levels of the individuals involved. The development of photoelectric elements and electrode sensor housings which are lightweight, wireless, and user-friendly is conducive to a more comfortable experience for the user.

fNIRS-based neurofeedback and BCI

Modulation of neural signals can also be achieved through neurofeedback training or a BCI, which provides feedback on central nervous information to the subject or creates information communication to control external machines (Bamdad et al., 2015; Sitaram et al., 2017). Neurofeedback and BCI have been widely used in clinical neuroeducation, such as for stroke, attention-deficit hyperactivity disorder, autism, epilepsy, and mood disorders. Neurofeedback enables individuals to self-regulate their cerebral activity in real time by presenting feedback through various sensory forms translated from the origins of neural signals, thereby promoting an improvement in behavioral performance (Sitaram et al., 2017). The fNIRS-based neurofeedback technology is employed to extract functional activity responses related to training, which are then utilized to provide feedback to subjects participating in the intervention process. The neurofeedback mechanism creates a closed-loop approach that allows individuals to self-regulate their cortical activity, thereby promoting their initiative in rehabilitation training (Hosseini et al., 2016).

fNIRS-based neurofeedback training can enhance the plasticity of the cerebral cortex and improve the motor rehabilitation in stroke patients with hemiplegia (Mihara et al., 2013). fNIRS-based neurofeedback can be classified into two categories based on the type of feedback signals: cortical activity neurofeedback and functional network neurofeedback. Cortical activity neurofeedback based on fNIRS has been widely applied in a variety of populations, such as in patients with hyperactivity disorder (Hart et al., 2013; Marx et al., 2014; Blume et al., 2017), schizophrenia, and anxiety disorders (Sobanski and Wagner, 2017), as well as in stroke motor rehabilitation (Mihara et al., 2013) and cognitive enhancement for healthy individuals (Kober et al., 2014; Hosseini et al., 2016). The source of the feedback signal is selected according to the specific function to be modulated. For example, the feedback signal for stroke motor rehabilitation is usually the cortical activity related to the motor area, while the feedback signal for cognitive enhancement is usually the functional area related to the cognitive function. One study indicated that the cortical response in the contralateral motor area can be enhanced through real neurofeedback training, whereas sham neurofeedback leads to a widespread activation of the cerebral cortex (Kober et al., 2014). In particular, fNIRS-based neurofeedback can be combined with motor imagery for stroke rehabilitation and has shown greater effectiveness than motor imagery therapy. Additionally, Sakurada et al. (2022) reported that fNIRS-based neurofeedback training demonstrated superior working memory ability during the target searching task than that of sham neurofeedback. The application of fNIRS-based neurofeedback training has demonstrated a significant improvement in the self-control and executive functions of children with attention-deficit and hyperactivity disorder (Marx et al., 2014; Blume et al., 2017). The functional integration principle highlights the fact that the execution of a certain function requires the involvement of multiple brain regions that collaborate as a functional network within the brain. Consequently, neurofeedback technology based on the brain functional network can better reflect complex information processing. For instance, Xia et al. (2021) applied functional network neurofeedback based on fNIRS to modulate the functional connectivity between frontoparietal areas related to cognitive functions in healthy individuals. The results revealed that the fNIRS-based neurofeedback group had superior cognitive performance compared to the control group.

BCI can control external devices through the acquisition of brain signals directly, independent of the integrity of the peripheral nervous system, making it a promising technology to restore motor functions in patients with disabilities (Dobkin, 2007; Daly and Wolpaw, 2008). fNIRS-based BCI has recently gained popularity and has been proven to be a viable tool in the field of stroke rehabilitation. Studies have applied fNIRS and EEG synchronously to create a motor imagery-based BCI to control functional electrical stimulation and demonstrated the feasibility of fNIRS-BCI, to realize motor training in a closed-loop manner for potentially enhanced outcomes (Wang et al., 2019). Currently, the precision of BCI classification based on motor imagination paradigms is restricted due to the presence of a lot of noise, as well as the overlap of physiological noise and the frequency bands of the task response. A wide variety of methods for feature extraction and classification have been explored to improve the performance of fNIRS-BCIs (Zhang et al., 2017; Shin and Im, 2020). Furthermore, the application of time-resolved fNIRS in BCIs has yielded satisfactory classification accuracy, indicating its potential for clinical use and further performance improvement (Abdalmalak et al., 2020).

Taken together, fNIRS-neurofeedback and BCI have a wide range of applications, particularly in children and older patient populations. Methods applied in studies are still quite heterogeneous and further agreements and standardizations of protocols remain necessary. Together with more rigorous research and reporting practices, further methodological improvements may lead to a more solid understanding and wider application of neurofeedback and BCI based on fNIRS.

Current limitations and future prospects of functional near-infrared spectroscopy in neuromodulation

Notwithstanding the widespread use and various clinical experiments conducted with central neuromodulation technology to ascertain its rehabilitative efficacy, the understanding in terms of promoting neural reorganization for neurological disorders remains limited and inconsistent. Moreover, the efficacy of neuromodulation technology depends on the precise selection of neurostimulation protocols, such as the intensity and target of the stimulation. For instance, a study has speculated that stroke-related motor dysfunction is associated with a hemispheric imbalance, which is evidenced by the excessive inhibition of the contralesional side on the ipsilesional side (Di Pino et al., 2014). This hemispheric imbalance can be ameliorated by modulating the transcranial nerve at a macro level of the nervous system, mainly by activating the affected hemisphere with high-frequency stimulation of TMS or anodal tDCS and inhibiting the excessive activity of the healthy hemisphere with low-frequency stimulation or cathodal tDCS (Hirakawa et al., 2018; McIntyre et al., 2018). However, experiments have revealed that both intervention strategies possess certain inadequacies. The compensatory role of the contralesional hemisphere in stroke rehabilitation remains a topic of significant debate, particularly in cases where patients have suffered severe damage (Morishita and Hummel, 2017). The distinct pathological patterns of the two hemispheric functional networks resulting from stroke may determine the most suitable area for stimulation. Additionally, the majority of the stimulation targets for mental disorders are located in brain areas linked to cognitive functions, such as the prefrontal cortex, which is responsible for executive functions such as decision-making, planning, and problem-solving (Lefaucheur et al., 2014). However, these regions do not exhibit the same measurable physiological response as in M1. In this regard, the accuracy of the target and stimulus intensities cannot be guaranteed. Therefore, the combination of neuroimaging and neurostimulation is essential to establish a theoretical basis to understand the neural reorganization mechanism to provide scientific guidance to select the optimal parameters, and then establish a “closed-loop” regulatory system that integrates assessment, intervention, and feedback.

fNIRS has the potential to be widely applied in the field of neuromodulation, owing to its unique technical advantages. The monitoring of a specific cortical response pattern associated with neuromodulation via fNIRS imaging is of utmost significance in guiding the establishment of stimulation parameters and realizing personalized rehabilitation therapy. Future research should focus on exploring the correlation and causality relationship between the cortical response and the improvement of behavioral performance, thereby providing a deep insight into the underlying neural mechanisms of neuromodulation therapy. Furthermore, fNIRS has the advantage of enabling repeatable measurements over a greater number of individuals. Therefore, future research could benefit from conducting large-sample clinical trials through multi-center studies, which would help solve the problem of low statistical power prevalent in neuroscience research. Additionally, due to the absence of a standardized signal processing procedure for fNIRS data, the analytical method employed in research remains quite heterogeneous, thus influencing the comparability and reproducibility of existing studies. To advance future research, it is crucial to make further agreements and standardize fNIRS signal processing methods that enable the reporting, replication, and interpretation of results. The implementation of online data processing techniques may facilitate the development of a functional imaging-based personalized target localization technology to facilitate precise neural regulation. Moreover, it is also essential to explore the possibility of combining fNIRS with other imaging modalities, such as fMRI and EEG, to establish a multimodal neuroimaging approach and apply it in neuroregulatory technology that can realize more accurate brain functional detection and contribute to the development of more effective rehabilitation intervention programs.

Conclusion

Accurate and objective evaluation of brain functional response is essential to achieve neuromodulation interventions for patients with neurological disorders. fNIRS technology, being non-invasive, eco-efficient, and tolerant to electromagnetic interference, makes it suitable for combination with cerebral neuromodulation, allowing for imaging of brain dynamics during neurostimulation and providing an objective, quantitative assessment of the stimulation-related effect in the application of neurological rehabilitation. This review outlined the potential mechanisms and existing evidence regarding the utilization of fNIRS in conjunction with neuromodulation techniques, with a particular focus on TMS, TES, neurofeedback, and BCI. The integration of fNIRS and neuromodulation technology has the potential to offer valuable insights into the optimization of stimulation parameters, thereby facilitating the advancement of personalized and accurate neurorehabilitation. Going forward, the combination of neuroimaging and neuromodulation technologies will be essential for the diagnosis, functional evaluation, and clinical intervention of central nervous system diseases.

Funding Statement

Funding: This work was supported by the National Natural Science Foundation of China, No. 32271370; National Key Research and Development Project, No. 2020YFC2004200; and Fundamental Research Funds for Central Public Welfare Research Institutes, No. 118009001000160001 (all to ZL).