I've built a mask that surrounds my region of interest, then use the mask within the MNF application to zero out those masked areas and force the application to only consider my ROI. This works fine and the MNF results show that the masked areas have values of 0, while the ROI has distinct positive or negative values. However, when I invert the MNF, choosing only the bands I want, the result gives an image that has a spectral profiles for those regions which I had previously masked - thought that these would still be zero? as well as those that were of the ROI . Additionally, I'm wondering if the statistics incorporate the masked regions as well and whether I need to apply the mask again during the inversion? Thanks for your help.