Running nnUNet tests - error in pre-processing

[andreabnes]

Hi All,

I am absolutely new to nnUNet, and after installing it, the first thing I wanted to try are the integration tests under nnunet2/tests/

My environment is based on Ubuntu 22.04, nnunetv2 2.5, pytorch 2.3.0

I have read the test instructions and followed the following steps:

• downloaded Task04_Hippocampus from http://medicaldecathlon.com/dataaws/
• converted it to a nnunet2 dataset with nnUNetv2_convert_old_nnUNet_dataset
• named the new dataset as Dataset004__Hippocampus and placed it under nnUNET_raw

I have then run nnunetv2/tests/integration_tests/prepare_integration_tests.sh

The script generates 4 new Dataset folders with IDs 996, 997, 998, and 999, each with subdirectories containing training images and labels. The script fails though with the following error:

(ai) andrea@ubuntu-2024-uefi:/mnt/S/WorkFolder_Andrea/Image_Segmentation/nnUNet/nnUNet$ ./nnunetv2/tests/integration_tests/prepare_integration_tests.sh
Traceback (most recent call last):
File "/mnt/S/WorkFolder_Andrea/Image_Segmentation/nnUNet/nnUNet/nnunetv2/dataset_conversion/datasets_for_integration_tests/Dataset996_IntegrationTest_Hippocampus_regions_ignore.py", line 69, in
seg_itk = sitk.ReadImage(s)
^^^^^^^^^^^^^^^^^
File "/home/andrea/anaconda3/envs/ai/lib/python3.11/site-packages/SimpleITK/extra.py", line 375, in ReadImage
return reader.Execute()
^^^^^^^^^^^^^^^^
File "/home/andrea/anaconda3/envs/ai/lib/python3.11/site-packages/SimpleITK/SimpleITK.py", line 8430, in Execute
return _SimpleITK.ImageFileReader_Execute(self)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: Exception thrown in SimpleITK ImageFileReader_Execute: /tmp/SimpleITK/Code/IO/src/sitkImageReaderBase.cxx:105:
sitk::ERROR: Unable to determine ImageIO reader for "/mnt/NEB/DeepLearning_Datasets_nnUNet/Dataset996_IntegrationTest_Hippocampus_regions_ignore/labelsTr/._hippocampus_003.nii.gz"
Fingerprint extraction...

It appears that SimpleITK is unable to read-in the nifti images. I have SimpleITK version 2.3.1. Has anyone incurred in the same problem? Does anyone have a suggestion?

Thanks for any advice, and congratulations for the great work,

Best Regards,

Andrea