Introduction
Stroke, which refers to a medical condition in which
insufficient brain blood supply results in cell death, is a major cause
of death and disability worldwide (
1,
2).
Survivors are accompanied with the deterioration or loss of functions,
for example, sensorimotor sequelae including motor weakness and
impairment of voluntary motor control, spasticity, incoordination,
apraxia, sensory loss/numbness, dysphagia, and dysarthria, and stroke
could also lead to various cognitive and psychiatric deficits (
3–
5).
These disfunctions are associated with cortical impairment due to
insufficient blood supply and brain oxygenation. Therefore, monitoring
the changes of brain circulation and oxygenation
could timely reflect
rehabilitation and recovery and the effect of therapy.
Neuroimaging has been shown to be an effective
monitoring and therapeutic tool, evaluating the evolution of neural
activity and stroke rehabilitation and recovery (
6).
Traditional methods such as functional magnetic resonance imaging
(fMRI), positron emission tomography, electroencephalography (EEG) and
magnetoencephalography (MEG)—have provided considerable initial insight
into brain changes during recovery. However, several shortcomings
including a confining monitoring environment, subject head fixation and
high cost have limited applications in tasks which require constant
movement or real-time monitor (
7,
8).
Near-infrared spectroscopy (NIRS), introduced in 1977 by
Jöbsis et al. as a monitoring tool of cerebral and myocardial
oxygenation (
9),
has partially overcome these difficulties. NIRS is a non-invasive
neuroimaging tool that has several potential advantages including
real-time monitor, low price, simplicity, portability, relatively small
equipment, and it's almost completely safe and non-invasive nature (
8).
NIRS can be divided into continuous wave NIRS (CW NIRS),
time domain NIRS (TD NIRS), and frequency domain NIRS (FD NIRS). CW
NIRS emits continuous wave and measures the changes in the intensity of
the light that passed through the tissue, whereas TD NIRS utilizes a
short pulse of laser light and measures the arrival times of photons
emerging from the tissue. FD NIRS records the intensity of the detected
light as well as the phase shift. These signals can then be converted to
the concentration of oxygenated (oxy-Hb) and deoxygenated hemoglobin
(deoxy-Hb). One of the most common used algorithm is Modified
Beer–Lambert Law (MBLL). CW NIRS
could not measure absolute
concentrations of oxy-Hb and deoxy-Hb, because this method assumes a
homogenous tissue which is not true. This does not change the results of
qualitative analysis, but may lead to error in quantitative outcome. TD
NIRS recording the temporal broadening of the pulse as it penetrates
the investigated area allows accurate quantification of the
concentrations and has better spatial resolution (
10).
Neural activity increases oxygen demands, thus increasing cerebral
blood flow due to neurovascular coupling. NIRS
could capture the changes
of oxy-Hb and deoxy-Hb to infer changes in brain activity (
6,
11).
Recent years have witnessed rapid development of the techniques of NIRS
from the single-location measurements to two dimensions and then three
dimensions. One of its most promising application is in brain-computer
interface (BCI) which was firstly introduced by Coyle et al. (
12).
BCI use brain activity to control external devices bypassing the
peripheral nervous system. NIRS presents as a valuable tool in its brain
signal acquisition for its non-invasive nature and real-time
monitoring. However, fNIRS-BCI system is still mainly used in researches
due to slow information transfer rate and low accuracy (
13).
Stroke includes two major types: ischemic, due to lack
of cerebral blood flow, and hemorrhagic, due to bleeding. During
ischemic stroke cerebral blood supply is disrupted by narrowing of
vessels caused by atherosclerosis or embolism. In hemorrhagic stroke
caused by hypertension or ruptured aneurysm blood flow is reduced due to
direct blood loss or vessel compression. In either case, a significant
decline in blood supply
could be observed. The reduction in cerebral
oxygenation results in injury of neurovascular unit, decreased neuronal
activity and accumulation of anaerobic metabolites. NIRS
could monitor
oxygenation signal changes, thus reflecting this pathophysiological
process (
1).
NIRS is well-established as a safe and effective
monitoring tool for stroke recovery, including upper limb, lower limb
recovery, motor learning, cortical function recovery (
8), cerebral hemodynamic changes, cerebral oxygenation (
14), therapy (
15,
16), clinical researches and evaluation of the risk for stroke (
17).
Brain-computer interfaces (BCIs), a newly emerging tool, use brain
activity to control external devices, facilitating paralyzed patients to
interact with the environment. NIRS combined with BCI offers great
potential as a therapeutic tool (
18,
19). In addition, NIRS has used in several clinical studies (
20,
21),
reflecting hemodynamic or oxygenation changes of brain, as well as in
evaluation of the risks for postoperative stroke and muscle metabolism (
22). NIRS has shown to be an effective and promising method, however, its clinical value still remains controversial (
23).
Therefore, to address these discrepancies, we conducted this systematic
review to summarize the existing application of NIRS in stroke
patients.
Muyue Yang
Zhen Yang3, 
Tifei Yuan