Orthogonal functions based estimation of fNIRS signal strength as GLM parameters, in the course of real-time brain activation detection

This paper presents a method to estimate the signal strength of the functional near infrared spectroscopy (fNIRS) as parameters of orthogonalized form of general linear model (GLM). The idea is to convert the basis function (design matrix) or explanatory variables of a GLM into orthogonal functions using the usual Gram-Schmidt orthogo-nalization procedure. The coefficients associated with the orthogonal functions are then easily estimated by applying the orthogonality condition. The original GLM coefficients are computed from these estimates. With this formulation, the estimation can be done recursively, making the approach applicable for real-time estimation. The results verify the effectiveness of the proposed technique by providing the activity levels at the measuring channels simultaneously while performing the motor-task experiment. It is worth mentioning that the technique is also providing the tracking of the motor tasks along with the activity of the channels during the experiment. The proposed method can be utilized for fNIRS based real-time brain imaging technology to cope the delay, inserted by the currently available off-line methods, in obtaining the medical diagnostics. It can further be utilized in applications demanding the real-time biofeedbacks.