kcl-bmeis / vs_seg Goto Github PK

I am following the instructions in the preprocessing folder to convert the data to Nifti but it does not work. I am using slicer 5.2.2 on mac os. and the slicer remains still and does nothing after giving these outputs,

Switch to module:  "Welcome"
['1']
case: 1
> <string>(440)main()

any idea what maybe wrong?

Hi, my computer GPU is 24GB, but when code run to epoch 2, loss.backward() will show CUDA out of memory.the batch size is one. I do bot know how to fix this! thanks:)

Goodmorning, when I was doing data preprocessing , to convert DICOM images to NIFTI, it raise an error. It's very nice of you to help me solve this problem, thanks.

Switch to module:  "Welcome"
TagCacheDatabase adding table

['1']
case: 1
TagCacheDatabase adding table

W: DcmItem: Length of element (4923,736e) is odd
E: DcmElement: Unknown Tag & Data (4923,736e) larger (1952999273) than remaining bytes in file
Could not load  "/Users/yingmuzhi/fsdownload/target/vs_gk_1_t1/inv_T1_LPS_to_T2_LPS.tfm" 
DCMTK says:  I/O suspension or premature end of stream
Could not read DICOM file:/Users/yingmuzhi/fsdownload/target/vs_gk_1_t1/inv_T1_LPS_to_T2_LPS.tfm
E: DcmElement: Unknown Tag & Data (0a5b,7b20) larger (572530698) than remaining bytes in file
Could not load  "/Users/yingmuzhi/fsdownload/target/vs_gk_1_t1/contours.json" 
DCMTK says:  I/O suspension or premature end of stream
Could not read DICOM file:/Users/yingmuzhi/fsdownload/target/vs_gk_1_t1/contours.json
"DICOM indexer has successfully inserted 123 files [0.09s]"
"DICOM indexer has successfully processed 125 files [0.50s]"
"DICOM indexer has updated display fields for 123 files [0.04s]"
W: DcmItem: Length of element (4923,736e) is odd
E: DcmElement: Unknown Tag & Data (4923,736e) larger (1952999273) than remaining bytes in file
Could not load  "/Users/yingmuzhi/fsdownload/target/vs_gk_1_t2/inv_T2_LPS_to_T1_LPS.tfm" 
DCMTK says:  I/O suspension or premature end of stream
Could not read DICOM file:/Users/yingmuzhi/fsdownload/target/vs_gk_1_t2/inv_T2_LPS_to_T1_LPS.tfm
E: DcmElement: Unknown Tag & Data (0a5b,7b20) larger (572530698) than remaining bytes in file
Could not load  "/Users/yingmuzhi/fsdownload/target/vs_gk_1_t2/contours.json" 
DCMTK says:  I/O suspension or premature end of stream
Could not read DICOM file:/Users/yingmuzhi/fsdownload/target/vs_gk_1_t2/contours.json
"DICOM indexer has successfully inserted 83 files [0.06s]"
"DICOM indexer has successfully processed 85 files [0.32s]"
"DICOM indexer has updated display fields for 83 files [0.02s]"
Traceback (most recent call last):
  File "<string>", line 5, in <module>
  File "<string>", line 531, in <module>
  File "<string>", line 440, in main
  File "<string>", line 144, in import_T1_and_T2_data
AssertionError: Not exactly 4, but 2 files selected for loading of case 1. 
Selected files are ['2: t1_mpr_tra_gk_v4', '4: t2_ci3d_tra_1.5mm_v1']

Dear all,

Thank you for the dataset and code, I could imagine what a hard job it was to deliver that.

We are reproducing your pipeline and found that some subjects have zero DICE on inference.
We are trying to explore the issue with failing segmentations for a while, maybe you've already faced that. So we try to (1) reproduce your pipeline and after to train (2) nnUnet with custom data preprocessing.

(1) While reproducing your pipeline on T1 subjects we found that the network will not predict any tumour in 103 subjects. The overall inference quality will be around 0.3 DICE. During training, the quality reaches DICE 0.9.

By inference, I mean predicting the whole dataset after the network is trained.

(2) When we trained nnUnet with a similar preprocessing to yours on T1 and T2 and we get DICE 0.83 and have ~9 subjects predicted with DICE 0.

We train nnUnet from MONAI on T1 data after your preprocessing, and again 103 subjects will be poorly predicted (with not DICE 0, but DICE 0.4)

During the inference we reduce the sliding window size to

self.sliding_window_inferer_roi_size = [128, 128, 32]``` 

This was reduced to fit into GPU. 

May be you can guide us - why subjects on inference get zeo DICEs? Maybe it is becouse you were training and predicting on the whole size image?

I have been trying to use the preprocessing code and I have been getting these errors:
When I enter <data/Vestibular-Schwannoma-SEG>:

Traceback (most recent call last):
  File "TCIA_data_convert_into_convenient_folder_structure.py", line 125, in <module>
    assert(all(found)), f"Not all required files found"
AssertionError: Not all required files found

When I point to the folder <data_path>:

Traceback (most recent call last):
  File "TCIA_data_convert_into_convenient_folder_structure.py", line 42, in <module>
    dd = pydicom.read_file(first_file)
  File "/Users/mabbasi6/opt/anaconda3/envs/momo_seg/lib/python3.6/site-packages/pydicom/filereader.py", line 993, in dcmread
    fp = open(fp, 'rb')
IsADirectoryError: [Errno 21] Is a directory: '/Users/mabbasi6/Downloads/VS_Seg/data/new/manifest-1614264588831/Vestibular-Schwannoma-SEG/VS-SEG-061/03-17-1996-NA-Avanto RoutineImage Guidance-11244'

Could you please let me know how I could solve it? and/or the data has changed causing some errors?

The README should point to https://doi.org/10.1038/s41597-021-01064-w

We should also include a cff file:
https://docs.github.com/en/repositories/managing-your-repositorys-settings-and-features/customizing-your-repository/about-citation-files

Hi, there. I appreciate the paper Wang, G. et al. Automatic Segmentation of Vestibular Schwannoma from T2-Weighted MRI by Deep Spatial Attention with Hardness-Weighted Loss, MICCAI, pp 264-272, 2019. and I am interested in Unet2d5_spvPA. But I found in this project VS_Seg missed the attention module, why? Is there any improvement for disable this module? I also want to see the net with attention module , thanks:)