Denoising Near Infrared Spectroscopy Signals with Principal Component Analysis Improves the Detection of Metabolic Foci

In this work Principal Component Analysis (PCA) was applied, in order to denoise Near-InfraRed Spectroscopy data recorded from 10 subjects performing visual tasks at different flickering frequencies. Due to the poor signal quality, indeed, the sole application of General Linear Model had not been able to extract any metabolic focus of activation, at the group level. Nevertheless, after pre-processing with PCA, the same dataset could provide stable, bilateral occipital pattern of activation at group level, by means of the same General Linear Model used before. Overall, the employment of Principal Component Analysis during the preprocessing stage provided improved detection of the metabolic foci.