Detection of consistently task-related activations in fMRI data with hybrid independent component analysis

fMRI data are commonly analyzed by testing the time course from each voxel against specific hypothesized waveforms, despite the fact that many compone nts of fMRI signals are difficult to specify explicitly. In contrast, purel y data-driven techniques, by focusing on the intrinsic structure of the dat a, lack a direct means to test hypotheses of interest to the examiner. Betw een these two extremes, there is a role for hybrid methods that use powerfu l data-driven techniques to fully characterize the data, but also use some a priori hypotheses to guide the analysis. Here we describe such a hybrid t echnique, HYBICA, which uses the initial characterization of the fMRI data from Independent Component Analysis and allows the experimenter to sequenti ally combine assumed task-related components so that one can gracefully nav igate from a fully data-derived approach to a fully hypothesis-driven appro ach. We describe the results of testing the method with two artificial and two real data sees. A metric based on the diagnostic Predicted Sum of Squar es statistic was used to select the best number of spatially independent co mponents to combine and utilize in a standard regressional framework. The p roposed metric provided an objective method to determine whether a more dat a-driven or a more hypothesis-driven approach was appropriate, depending on the degree of mismatch between the hypothesized reference function and the features in the data. HYBICA provides a robust way to combine the data-der ived independent components into a data-derived activation waveform and sui table confounds so that standard statistical analysis can be performed. (C) 2000 Academic Press.