OK, I have no insight into your data, but I am going to guess you did two things:
 1. You analyzed NIfTI images that do not report their modality (rather than mat files created by nii_preprocess or nii_nii2mat that store the modality)
 2. You did not explicity name your stat file.

In this case, prior versions would use the label 'lesion_' for output data. I just changed this so it now defaults to 'nii_' - the results will be the same, but the file name no longer makes any guess regarding modality.