Sejin Kim

Missing class/labels

2

How does nnU-Net handle missing labels for samples during multi-class training? For example: - Training 10-class segmentation model - Some samples are missing 1-4 classes due to missing labels /...

RuntimeError when loading trained weights

5

``` Traceback (most recent call last): File "train.py", line 21, in trainer.build_network() File "/cluster/projects/radiomics/Temp/sejin/AtlasNet/model/trainer_model.py", line 31, in build_network self.reload_network() File "/cluster/projects/radiomics/Temp/sejin/AtlasNet/model/trainer_model.py", line 42, in reload_network self.network.module.load_state_dict(torch.load(self.opt.reload_model_path, map_location='cuda:0')) File "/cluster/home/sejinkim/miniconda3/envs/atlasnet/lib/python3.6/site-packages/torch/nn/modules/module.py", line...

clearer autopipeline example

DICOM-Segmentation Module IO

1

Test robustness with multiple datasets.

To-do - Write documentation for H4H Public/Processed/Internal_Datasets management - Integrate docstring into official repo docs - [CARYN] QC checks to output FAILURE - assert image.shape == theoretical_shape - ex) assert...

- does contour fall within image bounds? - other TCIA QC check

assert image.shape == theoretical_shape - ex) assert spacing = image.GetSpacing() - ex) spacing = 1, 1, 2 -> orig_img.shape / spacing = image.GetSpacing()