Evaluation of Cyclic and Global Retrospective Corrections of Physiological Signals on Activated fMRI

INTRODUCTION Cardiac and respiratory fluctuations (physiological noise) are known to be confounding signals in the detection of cortical activation by fMRI. The standard correction method RETROICOR [1] has been used to perform a global correction by folding the whole time series (TS) into only one signal cycle and to use this global template to correct each individual cycle. We extended RETROICOR by the cyclic retrospective correction [2]. The cyclic method performs the correction piecewise at each cycle of a physiological signal. RETROICOR uses a Fourier series fit of order 2 and a respiratory phase based on a histogram (φh). In [2], we introduced a respiratory linear phase (φl) and a polynomial fit into the retrospective correction method. The goal of this study was to evaluate and compare different approaches to physiological noise correction on fMRI data acquired at different repetition times (TR).