FILTERING OF FUNCTIONAL NEAR INFRARED SPECTROSCOPY SIGNALS BY EIGENVALUE BASED METHODS


Eken A.

22nd IEEE Signal Processing and Communications Applications Conference (SIU), Trabzon, Turkey, 23 - 25 April 2014, pp.373-376 identifier

  • Publication Type: Conference Paper / Full Text
  • City: Trabzon
  • Country: Turkey
  • Page Numbers: pp.373-376
  • Middle East Technical University Affiliated: No

Abstract

Functional Near Infrared Spectroscopy is used in neuroimaging studies to observe the oxyhemoglobin (HB02) and deoxyhemoglobin (HB) changes. Blood oxygen level dependency (BOLD) signal that is collected by using this system shows response in related region in brain against an applied stimulus. Therefore in these signals, signal to noise ratio (SNR) is quite important to decide the behavior of brain in related region In this study, fNIRS data was filtered by using eigenvalue based methods such as Principal Component Analysis (PCA) and Truncated Singular Value Decomposition (tSVD). Using SNR and Autoregressive (AR) power spectrum performance results were compared.