Dear participants,

due to a copy-paste error on our side we accidentally included 3 cases, that did not match our quality and inclusion criteria. We hereby retract the files listed below and ask you to remove these files from your training setup and update the affected metadata files (see below for details).

Please note that these erroneously included files are different from the correct training and test data and therefore might negatively affect your results.

We just updated the download links to provide you with the correct dataset, if you start out with the challenge now, you will obtain the correct data. In case you already downloaded the data, you do not need to download the full dataset again, but follow these steps (or use the script shown below):

  • Download the files "dataset.json", "dataset_fingerprint.json", and "splits_final.json" from https://syncandshare.lrz.de/getlink/fiCJ6mQcjefMTQdKJsBSys/ and replace the corresponding files in your download directory
  • Go to the downloaded dataset and delete these files:
    • imagesTr/psma_a96814a79aa26c8f_2022-06-20_0000.nii.gz
    • imagesTr/psma_a96814a79aa26c8f_2022-06-20_0001.nii.gz
    • labelsTr/psma_a96814a79aa26c8f_2022-06-20.nii.gz
    • imagesTr/psma_c878220db97c05e1_2019-09-02_0000.nii.gz
    • imagesTr/psma_c878220db97c05e1_2019-09-02_0001.nii.gz
    • labelsTr/psma_c878220db97c05e1_2019-09-02.nii.gz
    • imagesTr/psma_f77011649a56cf08_2020-11-16_0000.nii.gz
    • imagesTr/psma_f77011649a56cf08_2020-11-16_0001.nii.gz
    • labelsTr/psma_f77011649a56cf08_2020-11-16.nii.gz

Alternatively, save the following code as filename.sh and execute it with bash (bash filename.sh) in the directory of your dataset (use with caution):

echo "Download new json files:"

curl -o dataset.json https://syncandshare.lrz.de/dl/fiCJ6mQcjefMTQdKJsBSys/dataset.json
curl -o dataset_fingerprint.json https://syncandshare.lrz.de/dl/fiCJ6mQcjefMTQdKJsBSys/dataset_fingerprint.json
curl -o splits_final.json https://syncandshare.lrz.de/dl/fiCJ6mQcjefMTQdKJsBSys/splits_final.json

echo "Remove affected niftis:"

rm imagesTr/psma_a96814a79aa26c8f_2022-06-20_0000.nii.gz
rm imagesTr/psma_a96814a79aa26c8f_2022-06-20_0001.nii.gz
rm labelsTr/psma_a96814a79aa26c8f_2022-06-20.nii.gz
rm imagesTr/psma_c878220db97c05e1_2019-09-02_0000.nii.gz
rm imagesTr/psma_c878220db97c05e1_2019-09-02_0001.nii.gz
rm labelsTr/psma_c878220db97c05e1_2019-09-02.nii.gz
rm imagesTr/psma_f77011649a56cf08_2020-11-16_0000.nii.gz
rm imagesTr/psma_f77011649a56cf08_2020-11-16_0001.nii.gz
rm labelsTr/psma_f77011649a56cf08_2020-11-16.nii.gz


Although we do not recommend, we do allow models that have been trained already on the old dataset to be submitted, but we will ask you to be transparent and report it.

We are very sorry for this inconvenience and thank you for understanding!

Best, Your challenge organizers