A New Approach to Separate Haemodynamic Signals for Brain-Computer Interface Using Independent Component Analysis and Least Squares

Abstract

Brain-computer interface (BCI) is one technology that allows a user to communicate with external devices through detecting brain activity. As a promising noninvasive technique, functional near-infrared spectroscopy (f NIRS) has recently earned increasing attention in BCI studies. However, in practice f NIRS measurements can suffer from significant physiological interference, for example, arising from cardiac contraction, breathing, and blood pressure fluctuations, thereby severely limiting the utility of the method. Here, we apply the multidistance f NIRS method, with short-distance and long-distance optode pairs, and we propose the combination of independent component analysis (ICA) and least squares (LS) with the f NIRS recordings to reduce the interference. The short-distance f NIRS measurement is treated as the virtual channel and the long-distance f NIRS measurement is treated as the measurement channel. Least squares is used to optimize the reconstruction value for brain activity signal. Monte Carlo simulations of photon propagation through a five-layered slab model of a human adult head were implemented to evaluate our methodology. The results demonstrate that the ICA method can separate the brain signal and interference; the further application of least squares can significantly recover haemodynamic signals contaminated by physiological interference from the f NIRS-evoked brain activity data.

Cortical Signal Analysis and Advances in Functional Near-Infrared Spectroscopy Signal: A Review

“…Morren et al ( 2004 ) analyzed and detected fast neuronal signal with a source-detector separation of 3 cm using ICA technique. Zhang et al ( 2013 ) has employed ICA methodology to explore the existence of particular wave form, modeled as Gamma variants. Similarly, Santosa et al ( 2013 ) has used ICA to extract the pHRF from regression vector including pHRF, a baseline correction and physiological noises.…”

Signal Processing in fNIRS: A Case for the Removal of Systemic Activity for Single Trial Data

“…It typically also comprises physiological noise such as heart beat (1–1.5 Hz), respiration (0.2–0.5 Hz) and low frequency content resulting from blood pressure fluctuations (Mayer waves; 0.1 Hz) (Naseer and Hong, 2015; Kamran et al, 2016). Several methods are available to remove these systemic, non-evoked components (Jang et al, 2009; Zhang et al, 2013; Naseer and Hong, 2015; Kamran et al, 2016). The most widely used method is band-pass filtering with cut-off frequencies of approximately 0.01–0.9 Hz (Naseer and Hong, 2015; Kamran et al, 2016).…”

Evoked Hemodynamic Response Estimation to Auditory Stimulus Using Recursive Least Squares Adaptive Filtering with Multidistance Measurement of Near-Infrared Spectroscopy

“…Here, we use the power spectral density to calculate the CNR before and after filtering. In the in vivo experiments, the brain activity signal and noises cannot be separated completely, and then the energy of signal and noise is obtained by integrating the power spectral density of “signal frequency band” and “noise frequency band” by using the method in the literature [ 13 ]. The CNR is then calculated as the square root of “signal energy” and “noise energy.” For the auditory block-design experiment, each block time is 40 seconds.…”

“…Least mean squares-based adaptive filtering has previously presented the potential to suppress physiological interference either for simulated signal or for real human data [ 9 , 12 ]. Other novel effective methods [ 3 , 13 , 14 ] relying on combining different algorithms were recently used for reducing the physiological interference and improving the signal to noise ratio of the evoked brain activity signal.…”