*RuntimeError: More than one of your folds has a prediction for case PETCT_PETCT_113_TP0.nii.gz**

[Preethi2121]

          Traceback (most recent call last):

File "/home/xxxi/xxx/envs/nnUNet/bin/nnUNetv2_find_best_configuration", line 8, in sys.exit(find_best_configuration_entry_point()) File "/xxx/data/xxxxxx/Code/nnUNet/nnunetv2/evaluation/find_best_configuration.py", line 295, in find_best_configuration_entry_point find_best_configuration(dataset_name, model_dict, allow_ensembling=not args.disable_ensembling, File "/mnt/data/Preethi/Code/nnUNet/nnunetv2/evaluation/find_best_configuration.py", line 100, in find_best_configuration accumulate_cv_results(output_folder, merged_output_folder, folds, num_processes, overwrite) File "/xxx/data/xxx/Code/nnUNet/nnunetv2/evaluation/accumulate_cv_results.py", line 40, in accumulate_cv_results raise RuntimeError(f'More than one of your folds has a prediction for case {pf}') RuntimeError: More than one of your folds has a prediction for case PETCT_PETCT_113_TP0.nii.gz, The json split file looks correct. No duplication of files between the split, but I have this error when I tried to find the best configuration. Some of the image files are saved in more than one fold.

Originally posted by @Preethi2121 in https://github.com/MIC-DKFZ/nnUNet/issues/1561#issuecomment-1933669150

[dojoh]

Hey, I'm not sure if I fully understand. The individual test sets of the cv folds should be disjoint. How come you have the imagefiles in multiple folds?

[Preethi2121]

I have no idea... Once the training was done. I ran the code for best configuration and this is what I had

On Tue, 20 Feb 2024, 16:34 dojoh, @.***> wrote:

Hey, I'm not sure if I fully understand. The individual test sets of the cv folds should be disjoint. How come you have the imagefiles in multiple folds?

— Reply to this email directly, view it on GitHub https://github.com/MIC-DKFZ/nnUNet/issues/1942#issuecomment-1954479065, or unsubscribe https://github.com/notifications/unsubscribe-auth/A4CJKG2NHDMBORZCNYYYMYLYUS7CFAVCNFSM6AAAAABC7MGYSOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNJUGQ3TSMBWGU . You are receiving this because you authored the thread.Message ID: @.***>

[Preethi2121]

The original split jason file doesn't have duplicates, it was only after the training, the val folders has duplicated image files

Preethi

On Tue, 20 Feb 2024, 17:14 Preethi Latha, @.***> wrote:

I have no idea... Once the training was done. I ran the code for best configuration and this is what I had

On Tue, 20 Feb 2024, 16:34 dojoh, @.***> wrote:

Hey, I'm not sure if I fully understand. The individual test sets of the cv folds should be disjoint. How come you have the imagefiles in multiple folds?

— Reply to this email directly, view it on GitHub https://github.com/MIC-DKFZ/nnUNet/issues/1942#issuecomment-1954479065, or unsubscribe https://github.com/notifications/unsubscribe-auth/A4CJKG2NHDMBORZCNYYYMYLYUS7CFAVCNFSM6AAAAABC7MGYSOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNJUGQ3TSMBWGU . You are receiving this because you authored the thread.Message ID: @.***>

[dojoh]

could you send us the json files (datset.json, split.json)? and also the commands you used? cheers, Ole

[Preethi2121]

Hi,

Please find the requested files attached. The commands used were

nnUNetv2_plan_and_preprocess -d 852 --verify_dataset_integrity

nnUNetv2_train 852 3d_cascade_fullres 0 [--npz] (Repeated this for 1,2,3,4 folds)

nnUNetv2_find_best_configuration 852 -c 3d_fullres- with this command I had the error

Preethi

On Wed, 28 Feb 2024 at 11:16, dojoh @.***> wrote:

could you send us the json files (datset.json, split.json)? and also the commands you used? cheers, Ole

— Reply to this email directly, view it on GitHub https://github.com/MIC-DKFZ/nnUNet/issues/1942#issuecomment-1968660301, or unsubscribe https://github.com/notifications/unsubscribe-auth/A4CJKG2NCUGQWJV66L2EW3TYV37XZAVCNFSM6AAAAABC7MGYSOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNRYGY3DAMZQGE . You are receiving this because you authored the thread.Message ID: @.***>

[Preethi2121]

Hi,

I am here again with another issue. I seen this specific pattern of artifact on the predicted images from the validation set. Do you have any suggestions or experience why this appears. I have this pattern in all my images of the validation set.

Please share your sugesstions to avoid this.

[dojoh]

Hey,

I could not find the attached files. The commands look good to me. About the dots: is this pattern somehow present in the images? are you talking about the image from the cross validation or the test images? for the cross validation this would be very strange. to me it looks like a difference between the datasets.

[Preethi2121]

Hi,

I am attaching the requested files again. The dots that you see are from the cross validation images. Could you please be more specific about what you mean by different datasets, because I have these patterns in all cross validation images and they are all from a single dataset.

Preethi

On Mon, 6 May 2024 at 13:41, dojoh @.***> wrote:

Hey,

I could not find the attached files. The commands look good to me. About the dots: is this pattern somehow present in the images? are you talking about the image from the cross validation or the test images? for the cross validation this would be very strange. to me it looks like a difference between the datasets.

— Reply to this email directly, view it on GitHub https://github.com/MIC-DKFZ/nnUNet/issues/1942#issuecomment-2095816126, or unsubscribe https://github.com/notifications/unsubscribe-auth/A4CJKG53EJ6Z3KZSVSDMADTZA5T57AVCNFSM6AAAAABC7MGYSOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAOJVHAYTMMJSGY . You are receiving this because you authored the thread.Message ID: @.***>