Selection of Optimal Hemodynamic Response Function for FMRI Analysis on Acute Stroke Patients

In this study we developed and tested a procedure designed to choose optimum hemodynamic response function (HRF) models for identifying changes in cerebral hemodynamics after delivery of median-nerve electrical stimulation in patients with acute stroke. Seventeen acute stroke patients underwent functional magnetic resonance imaging (fMRI) to investigate hemodynamic responses in the contralateral and ipsilateral primary somatosensory area (SI) after median-nerve electrical stimulation. fMRI activation images were acquired with a 3 T MR scanner. When the optimal model was applied after electrical stimulation, hemodynamic responses in the contralateral SI cortex differed in the 17 patients and showed wide intersubject variability. We analyzed the fMRI data by a novel software SOHIA (Selection of Optimal HRF and Image Analysis) developed in our laboratory and implemented in Matlab. Our results indicate some heterogeneity in the optimal HRF of our 17 acute stroke patients. Therefore the use of the same model in all subjects is inappropriate, and may results in bias in localization and/or extension of the activation map. Based on Akaike information criterion (AIC) for each subject SOHIA allows us to select the best HRF among available HRFs of Brain Voyager QX (BVQX) and Statistical Parametric Mapping 5 (SPM5) software.