MelanomaDetectionAndSegmentation

Code to reporduce our Melanoma Detection and Segmentation for MAM11A.

Detection (Classification)

The script in the classification folder can be run directly on a cluster with the bash scripts. Test output scores are written to .out files that you can find in the folder as well.

Segmentation

The script in the segmentation folder can be run directly on a cluster with the bash scripts to run a training on a simple unet.

For the nnUNet, we give the scripts that we use to structure our dataset. Further commands to use the nnUNet are

nnUNetv2_plan_and_preprocess -d 11_ID --verify_dataset_integrity

and

python evaluate_predictions.py /home/scur0404/projects/MelanomaDetectionAndSegmentation/nnUNet_results/Dataset011_Melanoma/nnUNetTrainer__nnUNetPlans__2d/test_msk /home/scur0404/projects/MelanomaDetectionAndSegmentation/nnUNet_results/Dataset011_Melanoma/nnUNetTrainer__nnUNetPlans__2d/inference_preds -l 0 1

The unet and nnUNet output scores are written to output files in the segmentation folder in an .out and .json format.