Can we train 2D images using tlt-train command?

I have 2D grayscale images(PNG) with corresponding masks for 5 classes.I want to train this data using CLARA platform.
Can I achieve this using tlt-train command either by using existing pre-trained models of CLARA or training from scratch??

Is your task segmentation or classification? The pre-trained models and their associated transformations expect nifti image as input data for segmentation task.

I am trying to run segmentation. I can convert them to nifti but, does any of clara’s pretrained model supports traing on 2D images?