Uncovering Regions of Maximum Dissimilarity on Random Process Data