Estimation of Baseline Drifts in fMRI

Classification of fMRI time series in a low-dimensional subspace with a spatial prior

We propose a new method for detecting activation in functional magnetic resonance imaging (fMRI) data. We project the fMRI time series on a low-dimensional subspace spanned by wavelet packets in order to create projections that are as non-Gaussian as possible. Our approach achieves two goals: it reduces the dimensionality of the problem by explicitly constructing a sparse approximation to the dataset and it also creates meaningful clusters allowing the separation of the activated regions from the clutter formed by the background time series. We use a mixture of Gaussian densities to model the distribution of the wavelet packet coefficients. We expect activated areas that are connected, and impose a spatial prior in the form of a Markov random field. Our approach was validated with in vivo data and realistic synthetic data, where it outperformed a linear model equipped with the knowledge of the true hemodynamic response.

Comparison of detrending methods for optimal fMRI preprocessing

Because of the inherently low signal to noise ratio (SNR) of fMRI data, removal of low frequency signal intensity drift is an important preprocessing step, particularly in those brain regions that weakly activate. Two known sources of drift are noise from the MR scanner and aliasing of physiological pulsations. However, the amount and direction of drift is difficult to predict, even between neighboring voxels. Further, there is no concensus on an optimal baseline drift removal algorithm. In this paper, five voxel-based detrending techniques were compared to each other and an auto-detrending algorithm, which automatically selected the optimal method for a given voxel time-series. For a significance level of P < 10(-6), linear and quadratic detrending moderately increased the percentage of activated voxels. Cubic detrending decreased activation, while a wavelet approach increased or decreased activation, depending on the dataset. Spline detrending was the best single algorithm. However, auto-detrending (selecting the best algorithm or none, if detrending is not useful) appears to be the most judicious choice, particularly for analyzing fMRI data with weak activations in the presence of baseline drift.