Noise reduction in fNIRS data using extended Kalman filter combined with short separation measurement

It is challenging to remove the physiological noise that is not evoked by the brain activity in fNIRS signals. We propose a novel method to effectively remove the superficial noise in the hemodynamic signals by combining an extended Kalman filter (EKF) with a short separation measurement based on a nonlinear balloon model. To demonstrate the improved performances of the proposed method over the existing linear Kalman filter (LKF), we use a synthetic hemodynamic signal to compare. As a result, the proposed EKF recovers the modeled hemodynamic responses with lower errors and higher correlation than the LKF.