We're using fslmaths bptf to bandpass filter our resting state fMRI data. When running fslmaths bptf version 5 vs. version 4 with the exact same input images and hp_sigma and lp_sigma values, we notice that the resulting image intensity values are slightly different. The differences in intensity are on the order of 1-5%. When we choose a few random voxels inside and outside the brain, and for each voxel correlate its timeseries from the v5 and v4 filtered 4D data, the correlations are high but not perfect (r=.97 to r=1). Is this expected behavior based on some minor modification in the temporal filtering code?
Thanks
Jesse
|