Characteristics of separation parameter in fNIRS signal separation using hemodynamic modality

Abstract

Functinal Near Infrared Spectroscopy(fNIRS) has been increasingly usedas a non-invasive measurement method of brain function activity.However, changes in blood flow due to body movements and physiologicalfluctuations are easily mixed into the actual measurementsignal. There are several methods for separating mixed signals, suchas bandpass filtering, Independent Component Analysis and a methodusing a Hemodynamic Modality separation model proposed by Yamada etal. In this study, the characteristics of a parameter of thehemodynamic modality separation method were investigated. Wedynamically determined the coefficient by using the mutual informationas an evaluation function in order to confirm whether there was atendency to be different for each measurement channel and section(task section and non-task section). As a result, differences werefound in the coefficients between the channels and sections where thevariation related to the task was confirmed and the other.