Portable and Personalized Neurofeedback for Stroke Rehabilitation: Multimodal EEG-fNIRS Neurofeedback

Near infrared spectroscopy (NIRS) is an increasingly popular and flexible technology for studying brain function. fNIRS measures changes in both oxy- and deoxy- hemoglobin and provides information similar to fMRI as it measures blood hemoglobin levels. Compared to fMRI, fNIRS has both advantages and disadvantages. First, concerning the NIRS signal, it offers higher temporal resolution than fMRI, due to a higher sampling rate, as illustrated in figure 6. Second, the practicality of the fNIRS is a major advantage over fMRI: it is easier to use, portable, safe, almost silent, and less expensive. In addition, fNIRS acquisitions are less sensitive to head movements than fMRI. Nevertheless, there are also several disadvantages to this technology. The fNIRS signal has a lower spatial resolution and has a lower signal-to-noise ratio compared to fMRI, depending on the target regions and the scalp-brain distance. Moreover, the hemodynamic response measured by fNIRS is delayed compared to the electrical response measured by EEG. Due to limitations in the power of light emitters, fNIRS cannot be used to measure deep cortical activity. 
 
These findings suggest that NIRS could be an appropriate substitute for fMRI in the context of neurofeedback (NF). Preliminary studies using fNIRS showed promising results for NF especially due to its practical advantages [1], [2]. As explained previously, fNIRS cannot be used to measure deep cortical activity. In our application case, post-stroke motor rehabilitation, this shortcoming should not be a problem, as motor areas are accessible by this modality. Moreover, the portability as well as the reduced sensitivity to head movement are also very interesting for the stroke patients, for which fMRI acquisitions could be more complicated. In [3], the authors showed the efficacy of fNIRS NF for patients with stroke with gait and balance impairments. In addition, a NIRS system can be easily coupled to an EEG cap to provide a portable EEG-fNIRS NF. This is in this perspective that the Empenn team recently acquired a fNIRS system, EEG and MRI compatible, through the Neurinfo platform. 

Some recent studies [4], [5] have compared the use of fNIRS to EEG in the context of NF and have shown that the two modalities, although consistent, have different and complementary properties. This confirms our conclusions on the bimodal NF EEG-fMRI concerning the interest of combining the two measures of cortical activity: the electrophysiological and the hemodynamic responses. To the best of our knowledge, simultaneous EEG-fNIRS NF has never been investigated. As fNIRS shares some properties with fMRI, the integration of this new modality to EEG definitely has the potential to enable a portable and more specific NF training for patients. Our ambition within this research axis, and in particular through the planned post-doctoral fellowship, will thus be to investigate the contribution of this new modality to propose an EEG-fNIRS NF and compare it to the enhanced EEG-NF setting. First, a bimodal EEG-fNIRS NF protocol will be implemented using classical targets (motor areas), evaluated by each of the two modalities. This will be validated on a few healthy subjects. Next, it will be interesting to try to adapt the model developed in an other work package of the PEPERONI project to be able to use this information from fNIRS to help predict fMRI scores in non-MRI sessions, using machine learning methods. Indeed, as mentioned above, the information from fNIRS has similarities with that measured by fMRI. It therefore seems logical to assume that fNIRS will be able to greatly help this prediction. Finally, recent works have focused on the estimation of brain connectivity using fNIRS and has shown a very good concordance with that measured with fMRI [6], [7]. Using fNIRS, the authors of [8] identified patterns of connectivity that were modified during the post-stroke recovery phase (without neurofeedback). As a last step, we can therefore consider incorporating the networks identified in an other WP into our prototype protocol.

[1]    S. H. Kohl, D. M. A. Mehler, M. Lührs, R. T. Thibault, K. Konrad, and B. Sorger, “The Potential of Functional Near-Infrared Spectroscopy-Based Neurofeedback—A Systematic Review and Recommendations for Best Practice,” Front. Neurosci., vol. 14, 2020, doi: 10.3389/fnins.2020.00594.
[2]    S. R. Soekadar, S. H. Kohl, M. Mihara, and A. von Lühmann, “Optical brain imaging and its application to neurofeedback,” NeuroImage Clin., vol. 30, p. 102577, Jan. 2021, doi: 10.1016/j.nicl.2021.102577.
[3]    M. Mihara et al., “Effect of Neurofeedback Facilitation on Poststroke Gait and Balance Recovery: A Randomized Controlled Trial,” Neurology, vol. 96, no. 21, pp. e2587–e2598, May 2021, doi: 10.1212/WNL.0000000000011989.
[4]    V. Kaiser et al., “Cortical effects of user training in a motor imagery based brain–computer interface measured by fNIRS and EEG,” NeuroImage, vol. 85, pp. 432–444, Jan. 2014, doi: 10.1016/j.neuroimage.2013.04.097.
[5]    A.-M. Marx et al., “Near-infrared spectroscopy (NIRS) neurofeedback as a treatment for children with attention deficit hyperactivity disorder (ADHD)—a pilot study,” Front. Hum. Neurosci., vol. 8, Jan. 2015, doi: 10.3389/fnhum.2014.01038.
[6]    L. Duan, Y.-J. Zhang, and C.-Z. Zhu, “Quantitative comparison of resting-state functional connectivity derived from fNIRS and fMRI: A simultaneous recording study,” NeuroImage, vol. 60, no. 4, pp. 2008–2018, May 2012, doi: 10.1016/j.neuroimage.2012.02.014.
[7]    P. Pinti et al., “The present and future use of functional near‐infrared spectroscopy (fNIRS) for cognitive neuroscience,” Ann. N. Y. Acad. Sci., vol. 1464, no. 1, pp. 5–29, Mar. 2020
[8]    K. M. Arun, K. A. Smitha, P. N. Sylaja, and C. Kesavadas, “Identifying Resting-State Functional Connectivity Changes in the Motor Cortex Using fNIRS During Recovery from Stroke,” Brain Topogr., vol. 33, no. 6, pp. 710–719, Nov. 2020, doi: 10.1007/s10548-020-00785-2.