Modeling the hemodynamic response in single-trial functional MRI experiments.

Today, most studies of cognitive processes using functional MRI (fMRI) experiments adopt a single-trial design. Highly flexible stimulation paradigms require new statistical models in which not only the activation amount but also the time course of the measured hemodynamic response is analyzed. Most previous approaches have been based on a linear regression context and have introduced hemodynamic model functions to improve the signal detection. In this report a nonlinear regression context is derived, from which shape parameters for the hemodynamic response are obtained per trial and per region of interest. These parameters allow the investigation of stimulus-induced shape variations of the hemodynamic response. By embedding the estimation into a robust statistical framework and rigorously analyzing the spatiotemporal interactions in the fMRI data, it is possible to derive statistically valid descriptions of single hemodynamic responses. The model, estimation algorithm, validation, and an example analysis from a single-trial fMRI study are reported. Magn Reson Med 42:787-797, 1999.