Wavelet Analysis of 4d Motor Task Fmri Data

We analyze the use of Haar wavelets in classification and clustering of 4D spatiotemporal fMRI motor data, comparing our results with previous studies that applied wavelets to motor task data as well as other techniques that have been previously applied to the same dataset, including voxelwise spatial k-means clustering and classification by subject based on time series analysis. Subject-based classification accuracies using wavelets reached 82% and distinct clusters emerged corresponding to differing anatomical regions of the brain, providing evidence that wavelets are an appropriate technique to use in both classification and clustering analysis of functional brain imaging data.