Variational variable selection to assess experimental condition relevance in event-related fMRI

Brain functional exploration investigates the nature of neural processing following cognitive or sensory stimulation. This goal is not fully accounted for in most functional Magnetic Resonance Imaging (fMRI) analysis which usually assumes that all delivered stimuli possibly generate a BOLD response everywhere in the brain although activation is likely to be in-duced by only some of them in specific brain regions. Gener-ally, criteria are not available to select the relevant conditions or stimulus types (e.g. visual, auditory, etc.) prior to acti-vation detection and the inclusion of irrelevant events may degrade the results, particularly when the Hemodynamic Re-sponse Function (HRF) is jointly estimated. To face this is-sue, we propose an efficient variational procedure that auto-matically selects the conditions according to the brain activity they elicit. It follows an improved activation detection and lo-cal HRF estimation that we illustrate on synthetic and real fMRI data. Index Terms — Irrelevance detection, Stimulus type se-lection, Joint detection-estimation, Bayesian hierarchica