Functional magnetic resonance imaging study: choice among linear basis set for hemodynamic response functions in acute stroke patients.

Background and Aims: Several studies examining the variability of the blood oxygenation level dependent (BOLD) response have sug- gested that a single Hemodynamic Response Function (HRF) model may not be optimal, since HRF differs from person to person, from day to day, and probably also in different region of the brain. The purpose of the present study is to develop a procedure to optimize the choice of HRF model in acute stroke patients following the stimulation of median nerve and a method to detect the performance of the HRFs in order to improve the identification of activated area. Methods: Seventeen acute stroke patients were examined using func- tional magnetic resonance imaging (fMRI) to investigate responses in the contralateral and ipsilateral primary somatosensory cortex (SI) following electrical stimulation of median nerve. fMRI activation was examined using a 3T MR imager. We analysed the data using an house software implemented in Matlab with all default HRFs of Brain Voyager QX and SPM5 plus other modified HRF that we made on the BOLD signal of patient. Results: A wide activation in contralateral SI was observed in 4 of 17 patients, a small activation in 7 of 17 and no activation in 6 of 17. Conclusions: We propose a mathematical approach in order to compare the performance between HFRs. Using the same HRF for each subject neglects the variability of hemodynamic response and could lead to false negatives and to false positives.