Sparse imaging and continuous event-related fMRI in the visual domain: a systematic comparison.

Abstract

Continuous image acquisition as used in most functional magnetic resonance imaging (fMRI) designs may conflict with specific experimental settings due to attendant, noisy gradient switching. In sparse fMRI, single images are recorded with a delay that allows the registration of the predicted peak of an evoked hemodynamic response (HDR). The aim of this study was to assess validity and sensitivity of single-trial sparse imaging within the visual domain. Thirteen subjects were scanned twice. Either continuous or sparse image acquisition was applied while participants viewed single trains of flashlights. Sparse fMRI results were compared to continuous event-related fMRI results on single- and multisubject level regarding spatial extent, overlap, and intensity of activation. In continuously recorded data, the variability of the HDR peak latency was examined because this measure determined the timing of sparse image acquisition. In sparse fMRI, the sensitivity was analyzed considering different numbers of averaged trials. Sparse imaging detected the core activity revealed using continuous fMRI. The intensity of signal changes detected by continuous or sparse fMRI was comparable. The HDR peak latency was stable across sessions, but intersubject and regional variability might have affected the power of sparse fMRI. In sparse imaging, adding trials resulted in extension of activation and improvement in statistical power. The comparison with established continuous fMRI confirms the validity of sparse imaging. Conventional event-related data acquisition and analysis provided more comprehensive results. However, only sparse fMRI offers the opportunity to apply stimuli and record further biosignals free of scanner-related artifacts during intervals without image acquisition.