A model-free approachto increasing the effect size of FNIRS data

Localizing brain activity in noisy functional near-infrared spectroscopy (fNIRS) data plays an important role when investigating task-related hemodynamics of the neuronal sites. We present a novel method for capturing drifts in the fNIRS data which increases the effect size of interest of the oxygenated (HbO) and deoxygenated (HbR) hemoglobin responses. Using linear least-squares, a consistent hemo-dynamic response function (HRF) of the fNIRS HbO/HbR response is estimated as a first-step that leads to an optimal estimate of the drift based on a wavelet thresholding technique. The de-drifted fNIRS responses are then obtained by removing the estimated drifts from the fNIRS time-series. Its performance is assessed using both simulated data and a real fNIRS data set obtained from a finger tapping task. The application results reveal that the proposed model-free method performs optimal de-drifting and increases the effect size of the fNIRS data.