This dataset was selected by a group of students engaged in analyses of the fMRIDC's dataset library using SPM5 and FSL tools, as part of an fMRI data analysis course at UC Davis.

The comments in this file were made while a reanalysis of the data was attempted. May they prove useful to others venturing into the complexities of this dataset ....

February 10, 2006
This particular dataset illustrates the complexities of maintaining orientation information in the repository. The raw EPI images are flipped in the z-direction, which exacerbates the inherent difficulties of determining which brain hemisphere is which.Without knowing whether the slices were collected in the Inferior to Superior direction (the common assumption) or the other way around (as implied by the images), it is difficult to rely 100% on the orientation implied by associated Analyze format headers which accompany the data. The accompanying protocol information is insufficient to determine slice collection direction.

Anatomical files do not have associated Analyze format headers and cannot be read easily into SPM or FSL.

Numerous runs of data missing for several subjects.

Regular image artifacts occur at one of 3 locations within a run for all runs for all subjects (pdf). The artifacts are an overall increasing shift in the intensity histogram which lasts for one timepoint.