Walking is a complex motor behavior with a special relevance in clinical neurology. Many neurological diseases, such as Parkinson’s disease and stroke, are characterized by gait disorders whose neurofunctional correlates are poorly investigated. Indeed, the analysis of real walking with the standard neuroimaging techniques poses strong challenges, and only a few studies on motor imagery or walking observation have been performed so far. Functional near-infrared spectroscopy (fNIRS) is becoming an important research tool to assess functional activity in neurological populations or for special tasks, such as walking, because it allows investigating brain hemodynamic activity in an ecological setting, without strong immobility constraints. A systematic review following PRISMA guidelines was conducted on the fNIRS-based examination of gait disorders. Twelve of the initial yield of 489 articles have been included in this review. The lesson learnt from these studies suggest that oxy-hemoglobin levels within the prefrontal and premotor cortices are more sensitive to compensation strategies reflecting postural control and restoration of gait disorders. Although this field of study is in its relative infancy, the evidence provided encourages the translation of fNIRS in clinical practice, as it offers a unique opportunity to explore in depth the activity of the cortical motor system during real walking in neurological patients. We also discuss to what extent fNIRS may be applied for assessing the effectiveness of rehabilitation programs.
Walking is one of the most fundamental motor functions in humans,1–3 often impaired in some focal neurological conditions (ie, stroke), or neurodegenerative diseases, such as Parkinson’s disease (PD).4 Worldwide almost two thirds of people over 70 years old suffer from gait disorders, and because of the progressively ageing population, an increasing pressure on health care systems is expected in the coming years.5
Although the physiological basis of walking is well understood, pathophysiological mechanisms in neurological patients have been poorly described. This is caused by the difficulty to assess in vivo neuronal processes during overt movements.
During the past 20 years, functional magnetic resonance imaging (fMRI) has been the preferred instrument to investigate mechanisms underlying movement control6 as well as movement disorders.7 fMRI allows measuring the blood oxygenation level-dependent (BOLD) signal that, relying on variations in deoxy-hemoglobin (deoxyHb) concentrations, provides an indirect measure of functional activity of the human brain.8 Patterns of activation/deactivation and connectivity across brain regions can be detected with a very high spatial resolution for both cortical and subcortical structures. This technique, however, is characterized by severe limitations and constraints about motion artifacts and only small movements are allowed inside the scanner. This entails dramatic compromises on the experimental design and on the inclusion/exclusion criteria. Multiple solutions have been attempted to overcome such limitations. For instance, many neuroimaging studies have been performed on the motor imagery,9,10 but imaging can be different from subject to subject,11 and imagined walking and actual walking engage different brain networks.12 Other authors have suggested the application of virtual reality,13 and there have been a few attempts to allow an almost real-walking sequence while scanning with fMRI.14,15Additional opportunities to investigate the mechanisms sustaining walking control include the use of surrogate tasks in the scanner as proxy of walking tasks,16 or to “freeze” brain activations during walking using positron emission tomography (PET) radiotracers, which allow the retrospective identification of activation patterns, albeit with some uncertainties and low spatial and temporal resolution.12
Therefore, until now there has not been an ecological way to noninvasively assess neurophysiological correlates of walking processes in gait disorders.
Functional near-infrared spectroscopy (fNIRS) is becoming an important research tool to assess functional activity in special populations (neurological and psychiatric patients)17 or for special tasks.18–21 fNIRS is a noninvasive optical imaging technique that, similarly to fMRI, measures the hemodynamic response to infer the underlying neural activity. Optical imaging is based on near-infrared (650-1000 nm) light propagation into scattering tissues and its absorption by 2 major chromophores in the brain, oxy-hemoglobin (oxyHb) and deoxyHb, which show specific absorption spectra depending on the wavelength of the photons.22 Typically, an fNIRS apparatus is composed of a light source that is coupled to the participant’s head via either light-emitting diodes (LEDs) or through fiber-optical bundles with a detector that receives the light after it has been scattered through the tissue. A variation of the optical density of the photons measured by detectors depends on the absorption of the biological tissues (Figure 1A). Using more than one wavelength and applying the modified Beer-Lambert law, it is possible to infer on the changes of oxyHb and deoxyHb concentrations.23 fNIRS has a number of definite advantages compared to fMRI, its major competitor: (a) it does not pose immobility constrains,25 (b) is portable,26 (c) allows recording during real walking,27 (d) allows long-lasting recordings, (e) it does not produce any noise, (f) it makes possible the investigation of brain activity during sleep,28 (f) it allows to obtain a richer picture of the neurovascular coupling as it measures changes in both oxyHb and deoxyHb concentration with high temporal resolution (up to milliseconds). High temporal resolution is usually not mandatory for the investigation of the hemodynamic response whose dynamic takes at least 3 to 5 seconds, but it can be useful for the study of transient hemodynamic activity like the initial dip29 or to detect subtle temporal variations in the latency of the hemodynamic response across different experimental conditions.19,21,30 The major drawback of fNIRS in comparison to fMRI is its lower spatial resolution (few centimeters under the skull) and its lack of sensitivity to subcortical regions.18,19 However, this might be considered a minor limitation, as there is a large body of evidence suggesting that (a) cortical mechanisms take place in walking,31 (b) the organization of the motor system is distributed along large brain regions,32and (c) the function of subcortical structures is mirrored in the cerebral cortex.33
Figure 1. Illustration of penetration depth of near-infrared light into the tissue in a probe configuration used to investigate motor performances during walking task (upper row). The picture shows brain reconstruction from a high-resolution anatomical MRI. The spheres placed over the skull correspond to vitamin E capsules employed during the MRI to mark the positions of the optodes and to allow the coregistration of the individual anatomy together with the optode position. In this illustration, only the photons propagation from one source (S) to one detector (D) have been simulated. The yellow-red scale indicates the degree of sensitivity74 for the considered source-detector pair to the head/brain structures. (A, B, and C) Lower row: Examples of fNIRS experimental device used for assessing brain activity during real walking tasks. These fNIRS approaches included either commercial device, such as (A) wireless portable fNIRS system (NIRx; Germany) or support systems for treadmill walking activity with body weight support24 (B) or with free movement range (C).