hello,a simple question

[puppy2000]

I wonder what is the gt that is used to train the Deep Label Expansion stage?From my observation,it is seemd that labels generated from torchio.LabelMap are used.Thanks very much!

[LucaLumetti]

Hi @puppy2000, I hope I've understood clearly your question: each patient in the train, test, or eval split has a single gt_alpha.npy file which corresponds to its gt. As you can read in this line, this is exactly what it gets loaded and later used as gt. The torchio.LabelMap is just a class helpful to warp and handle labels, it does not generate anything, you can find more details on the Torchio documentation page.