I am trying to run melodic on 3D contrast images. Obviously it is not running because it expects a 4D time series file, however I m inputting it a 3D. The goal is to make clusters in the concatenated contrast files for detections of different IC components that may be used as regions of interests. Would it be okay to concatenate the contrast files across subjects, thus making 4D file and then run melodic ? Or do you suggest any other way to achieve this ?