Signal Extraction of Brain Activity Response by Near Infrared Spectroscopy based on Recursive Least Squares Adaptive Filtering

Abstract

Near infrared spectroscopy (NIRS) is a non-invasive technique having the potential to allow the study of brain functional activation through the monitoring of changes in haemodynamic variables. However, the NIRS data derived in this application are often contaminated by physiological interference arising, for example, from cardiac contraction, breathing, and spontaneous low frequency oscillations. There have therefore been efforts to develop signal processing schemes aimed at improving signal quality. In the present paper, the multidistance NIRS probe configuration has been adopted. The near-distance source-detector pair is used to derive the superficial haemodynamic changes and the fardistance source-detector pair is used to derive deep haemodynamic changes. The recursive least squares algorithm was used to process the NIRS data in order to suppress physiological interference. We also utilized Monte Carlo simulations, based on a five-layer model of the adult human head, to evaluate our methodology. The results suggest that the recursive least squares approach has the potential to reduce physiological interference in NIRS data. Important advantages of this method are that it is adaptive and it can be used for real-time signal extraction of brain activity.KeywordsNear infrared spectroscopyMonte CarloLambert-Beer lawRecursive least squaresMultidistance measurement