So, my dataset is commposed of images with different number of slices (and also different heights and widths). How can i handle this? All the CNN models to segment it I've found uses a fix number of channels.
I thought in creating channels with 0's, but I'm not sure how it would affect the segmentation.