Ins Sousa1,2, Pedro Vilela3, Patricia Figueiredo1
1Institute for Systems & Robotics, Instituto Superior Tcnico, Lisbon, Portugal; 2Healthcare Sector, Siemens, S.A., Lisbon, Portugal; 3Imaging Department, Hospital da Luz, Lisbon, Portugal
An approach for the optimization of a Gamma impulse response function to be used in a GLM analysis ASL perfusion and BOLD data, collected during a Cued Deep Breathing paradigm, is proposed. The optimal parameter values of the Gamma function were determined, for the whole-brain in terms of maximum number of active voxels, and on a voxel-by-voxel basis in terms of the maximum Z score. The results obtained indicate that a voxe-by-voxel optimization may be preferable due to the spatial heterogeneity of the parameters, yielding optimal lag and width maps and contributing to more sensitive and accurate cerebrovascular reactivity measurements.