What is in this repo? Preprocessing pipeline to process VISCERAL scans for the Spinal CT to MR Translation project and evaluation.
-
Download the visceral dataset and put 36 CT and MRI scans (72 in total) in folder data/visceral/train. For the test set, put the remaining scans in folder data/visceral/test.
-
Run
python prep_data.py
to prepare the training and test sets. -
There should be trainA and trainB folders in data/visceral containing numpy arrays. Each numpy array contains a patch from CT or MRI slice.
-
Transfer the data to the CycleGAN repository and start the training there. The scripts move_data_visceral.sh might be useful for this task.
Some results from this project may be viewed in results.ipynb.