-
Notifications
You must be signed in to change notification settings - Fork 1
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #66 from ivadomed/plb/nnunet-region-based
Region based nnUNet scripts
- Loading branch information
Showing
8 changed files
with
3,227 additions
and
0 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,171 @@ | ||
| # PSIR | ||
| - sub-mon118_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/43 | ||
| - sub-mon006_ses-M0 # FOV covers the whole brain. The SC has only C1-C6; see https://github.com/ivadomed/canproco/issues/22#issue-1568568207 | ||
| - sub-mon009_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1758684880 | ||
| - sub-mon032_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1758686509 | ||
| - sub-mon097_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1758686772 | ||
| - sub-mon113_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1760175861 | ||
| - sub-mon148_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1758686965 | ||
| - sub-mon152_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1760248384 | ||
| - sub-mon168_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1758687112 | ||
| - sub-mon191_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1758687352 | ||
| - sub-van134_ses-M0 # No disc labeling file : impossible to do it : https://github.com/ivadomed/canproco/issues/53#issuecomment-1761698648 | ||
| - sub-van135_ses-M0 # No disc labeling file : impossible to do it : https://github.com/ivadomed/canproco/issues/53#issuecomment-1761702591 | ||
| - sub-van171_ses-M0 # No disc labeling file : impossible to do it : https://github.com/ivadomed/canproco/issues/53#issuecomment-1761910863 | ||
| - sub-van176_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1758687602 | ||
| - sub-van181_ses-M0 # No disc labeling file : impossible to do it : https://github.com/ivadomed/canproco/issues/53#issuecomment-1767235494 | ||
| - sub-van201_ses-M0 # No disc labeling file : impossible to do it : https://github.com/ivadomed/canproco/issues/53#issuecomment-1767238882 | ||
| - sub-van206_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1758688851 | ||
| - sub-van207_ses-M0 # No disc labeling file : impossible to do it : https://github.com/ivadomed/canproco/issues/53#issuecomment-1761973064 | ||
| - sub-tor014_ses-M0 # Poor data quality and no disc labeling file; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1760420705 | ||
| - sub-tor133_ses-M0 # Poor data quality; see https://github.com/ivadomed/canproco/issues/53#issuecomment-1758689928 | ||
| # STIR | ||
| - sub-cal149_ses-M0 # FOV does not cover the whole SC in sagittal plane; see https://github.com/ivadomed/canproco/issues/22#issue-1568568207 | ||
| # Similar accross timepoints | ||
| - sub-edm005_ses-M0 | ||
| - sub-edm031_ses-M0 | ||
| - sub-edm035_ses-M0 | ||
| - sub-edm054_ses-M0 | ||
| - sub-edm065_ses-M0 | ||
| - sub-edm088_ses-M0 | ||
| - sub-mon002_ses-M0 | ||
| - sub-mon003_ses-M0 | ||
| - sub-mon005_ses-M0 | ||
| - sub-mon006_ses-M0 | ||
| - sub-mon007_ses-M0 | ||
| - sub-mon009_ses-M0 | ||
| - sub-mon010_ses-M0 | ||
| - sub-mon011_ses-M0 | ||
| - sub-mon014_ses-M0 | ||
| - sub-mon031_ses-M0 | ||
| - sub-mon032_ses-M0 | ||
| - sub-mon041_ses-M0 | ||
| - sub-mon052_ses-M0 | ||
| - sub-mon060_ses-M0 | ||
| - sub-tor010_ses-M0 | ||
| - sub-tor012_ses-M0 | ||
| - sub-tor014_ses-M0 | ||
| - sub-tor016_ses-M0 | ||
| - sub-tor021_ses-M0 | ||
| - sub-tor029_ses-M0 | ||
| - sub-tor036_ses-M0 | ||
| - sub-tor038_ses-M0 | ||
| - sub-tor039_ses-M0 | ||
| - sub-tor043_ses-M0 | ||
| - sub-tor044_ses-M0 | ||
| - sub-tor051_ses-M0 | ||
| # Subjects only present at M12 | ||
| - sub-cal064_ses-M12 | ||
| - sub-edm161_ses-M12 | ||
| - sub-mon164_ses-M12 | ||
| # Subject with 4 dimensions | ||
| - sub-cal123_ses-M12 | ||
| # Subject only present at M0 | ||
| - sub-cal080_ses-M0 | ||
| - sub-cal113_ses-M0 | ||
| - sub-cal125_ses-M0 | ||
| - sub-cal130_ses-M0 | ||
| - sub-cal131_ses-M0 | ||
| - sub-cal136_ses-M0 | ||
| - sub-cal140_ses-M0 | ||
| - sub-cal151_ses-M0 | ||
| - sub-cal153_ses-M0 | ||
| - sub-cal154_ses-M0 | ||
| - sub-cal156_ses-M0 | ||
| - sub-cal162_ses-M0 | ||
| - sub-cal173_ses-M0 | ||
| - sub-cal185_ses-M0 | ||
| - sub-cal191_ses-M0 | ||
| - sub-cal197_ses-M0 | ||
| - sub-cal198_ses-M0 | ||
| - sub-cal199_ses-M0 | ||
| - sub-cal200_ses-M0 | ||
| - sub-cal206_ses-M0 | ||
| - sub-cal207_ses-M0 | ||
| - sub-cal209_ses-M0 | ||
| - sub-cal210_ses-M0 | ||
| - sub-cal213_ses-M0 | ||
| - sub-cal214_ses-M0 | ||
| - sub-edm008_ses-M0 | ||
| - sub-edm010_ses-M0 | ||
| - sub-edm018_ses-M0 | ||
| - sub-edm023_ses-M0 | ||
| - sub-edm029_ses-M0 | ||
| - sub-edm034_ses-M0 | ||
| - sub-edm038_ses-M0 | ||
| - sub-edm053_ses-M0 | ||
| - sub-edm064_ses-M0 | ||
| - sub-edm067_ses-M0 | ||
| - sub-edm089_ses-M0 | ||
| - sub-edm095_ses-M0 | ||
| - sub-edm098_ses-M0 | ||
| - sub-edm107_ses-M0 | ||
| - sub-edm138_ses-M0 | ||
| - sub-edm157_ses-M0 | ||
| - sub-edm162_ses-M0 | ||
| - sub-edm164_ses-M0 | ||
| - sub-edm169_ses-M0 | ||
| - sub-edm178_ses-M0 | ||
| - sub-edm181_ses-M0 | ||
| - sub-edm183_ses-M0 | ||
| - sub-mon013_ses-M0 | ||
| - sub-mon027_ses-M0 | ||
| - sub-mon064_ses-M0 | ||
| - sub-mon107_ses-M0 | ||
| - sub-mon108_ses-M0 | ||
| - sub-mon109_ses-M0 | ||
| - sub-mon139_ses-M0 | ||
| - sub-mon163_ses-M0 # parent folder exist, missing only PSIR contrast | ||
| - sub-mon180_ses-M0 | ||
| - sub-mon181_ses-M0 | ||
| - sub-tor004_ses-M0 | ||
| - sub-tor018_ses-M0 | ||
| - sub-tor024_ses-M0 | ||
| - sub-tor025_ses-M0 | ||
| - sub-tor031_ses-M0 | ||
| - sub-tor032_ses-M0 | ||
| - sub-tor041_ses-M0 | ||
| - sub-tor055_ses-M0 | ||
| - sub-tor063_ses-M0 | ||
| - sub-tor066_ses-M0 | ||
| - sub-tor073_ses-M0 | ||
| - sub-tor076_ses-M0 | ||
| - sub-tor077_ses-M0 | ||
| - sub-tor106_ses-M0 | ||
| - sub-tor109_ses-M0 | ||
| - sub-tor110_ses-M0 | ||
| - sub-tor112_ses-M0 | ||
| - sub-tor124_ses-M0 | ||
| - sub-tor125_ses-M0 | ||
| - sub-tor147_ses-M0 | ||
| - sub-van009_ses-M0 | ||
| - sub-van093_ses-M0 | ||
| - sub-van112_ses-M0 | ||
| - sub-van116_ses-M0 | ||
| - sub-van123_ses-M0 | ||
| - sub-van125_ses-M0 | ||
| - sub-van129_ses-M0 | ||
| - sub-van131_ses-M0 | ||
| - sub-van134_ses-M0 | ||
| - sub-van144_ses-M0 | ||
| - sub-van145_ses-M0 | ||
| - sub-van147_ses-M0 | ||
| - sub-van149_ses-M0 | ||
| - sub-van150_ses-M0 | ||
| - sub-van151_ses-M0 | ||
| - sub-van154_ses-M0 | ||
| - sub-van168_ses-M0 | ||
| - sub-van172_ses-M0 | ||
| - sub-van173_ses-M0 | ||
| - sub-van178_ses-M0 | ||
| - sub-van184_ses-M0 | ||
| - sub-van186_ses-M0 | ||
| - sub-van204_ses-M0 | ||
| - sub-van210_ses-M0 | ||
| - sub-van211_ses-M0 | ||
| - sub-van214_ses-M0 | ||
| - sub-van215_ses-M0 | ||
| - sub-van217_ses-M0 | ||
| - sub-van219_ses-M0 | ||
| - sub-van221_ses-M0 | ||
| - sub-van222_ses-M0 |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,132 @@ | ||
| # Installation instructions | ||
|
|
||
| ## Installation of Anima metrics | ||
|
|
||
| Installation of | ||
|
|
||
| ``` | ||
| cd ~ | ||
| mkdir anima/ | ||
| cd anima/ | ||
| wget -q https://github.com/Inria-Empenn/Anima-Public/releases/download/v4.2/Anima-macOS-4.2.zip # for MACOS | ||
| unzip Anima-macOS-4.2.zip | ||
| rm Anima-macOS-4.2.zip | ||
| git lfs install | ||
| git clone --depth 1 https://github.com/Inria-Visages/Anima-Scripts-Public.git | ||
| git clone --depth 1 https://github.com/Inria-Visages/Anima-Scripts-Data-Public.git | ||
| ``` | ||
|
|
||
| Configure directories | ||
|
|
||
| ``` | ||
| cd ~ | ||
| mkdir .anima/ | ||
| touch .anima/config.txt | ||
| echo "[anima-scripts]" >> .anima/config.txt | ||
| echo "anima = ${HOME}/anima/Anima-Binaries-4.2/" >> .anima/config.txt | ||
| echo "anima-scripts-public-root = ${HOME}/anima/Anima-Scripts-Public/" >> .anima/config.txt | ||
| echo "extra-data-root = ${HOME}/anima/Anima-Scripts-Data-Public/" >> .anima/config.txt | ||
| ``` | ||
|
|
||
| ## Installation of required libraries | ||
|
|
||
| Create a virtual invironment: | ||
| ~~~ | ||
| conda create -n venv_nnunet python=3.9 | ||
| ~~~ | ||
|
|
||
| Activate the environment with the following command: | ||
| ~~~ | ||
| conda activate venv_nnunet | ||
| ~~~ | ||
|
|
||
| To install required libraries to train an nnUNet v2: | ||
|
|
||
| ``` | ||
| pip install -r requirements_nnunet.txt | ||
| ``` | ||
|
|
||
| Install SpinalCordToolbox 6.0 : | ||
|
|
||
| Installation link : https://spinalcordtoolbox.com/user_section/installation.html | ||
|
|
||
|
|
||
| # Data preparation | ||
|
|
||
| Create the following folders: | ||
|
|
||
| ~~~ | ||
| mkdir nnUNet_raw | ||
| mkdir nnUNet_preprocessed | ||
| mkdir nnUNet_results | ||
| ~~~ | ||
|
|
||
| We are training a region-based nnUNet taking an image from contrasts PSIR or STIR and creating a mask with 0=background, 1=spinal cord and 2=MS lesion. | ||
|
|
||
| Convert the data to the nnUNet format : | ||
|
|
||
| ~~~ | ||
| python convert_BIDS_to_nnunet.py --path-data /path/to/BIDS/dataset --path-out /path/to/nnUNet_raw --taskname TASK-NAME --tasknumber DATASET-ID --contrasts PSIR,STIR --test-ratio XX --time-point ses-XX --type training --exclude-file /path/to/exclude_file.yml | ||
| ~~~ | ||
|
|
||
| > **Note** | ||
| > The test ratio is 0.2 for 20% (train ratio is therefore 80%). For M0 images, the time point is ses-M0. | ||
| To mutliply PSIR images by -1 before training and convert the data to the nnUNet format : | ||
|
|
||
| ~~~ | ||
| python convert_BIDS_to_nnunet_with_mul_PSIR.py --path-data /path/to/BIDS/dataset --path-out /path/to/nnUNet_raw --taskname TASK-NAME --tasknumber DATASET-ID --contrasts PSIR,STIR --test-ratio XX --time-point ses-XX --type training --exclude-file /path/to/exclude_file.yml | ||
| ~~~ | ||
|
|
||
| # Model training | ||
|
|
||
| Before training the model, nnU-Net performs data preprocessing and checks the integrity of the dataset: | ||
|
|
||
| ~~~ | ||
| export nnUNet_raw="/path/to/nnUNet_raw" | ||
| export nnUNet_preprocessed="/path/to/nnUNet_preprocessed" | ||
| export nnUNet_results="/path/to/nnUNet_results" | ||
| nnUNetv2_plan_and_preprocess -d DATASET-ID --verify_dataset_integrity | ||
| ~~~ | ||
|
|
||
| You will get the configuration plan for all four configurations (2d, 3d_fullres, 3d_lowres, 3d_cascade_fullres). | ||
|
|
||
| To train the model, use the following command: | ||
| ~~~ | ||
| CUDA_VISIBLE_DEVICES=XXX nnUNetv2_train DATASET-ID CONFIG FOLD --npz | ||
| ~~~ | ||
|
|
||
| > **Note** | ||
| > Example for Dataset 101, on 2d config on fold 0: CUDA_VISIBLE_DEVICES=2 nnUNetv2_train 101 2d 0 --npz | ||
| # Model inference | ||
|
|
||
| Convert data to nnUNet format for inference using `convert_BIDS_to_nnunet.py` or `convert_BIDS_to_nnunet_with_mul_PSIR.py` with `--type=inference`. | ||
|
|
||
| Then perform inference: | ||
| ~~~ | ||
| CUDA_VISIBLE_DEVICES=XXX nnUNetv2_predict -i /path/to/image/folder -o /path/to/predictions -d DATASET_ID -c CONFIG --save_probabilities -chk checkpoint_best.pth -f FOLD | ||
| ~~~ | ||
|
|
||
| # Inference evaluation | ||
|
|
||
| First, convert the predictions back the BIDS format ; this only keeps the lesion segmentation and discards spinal cord segmentation : | ||
|
|
||
| ~~~ | ||
| python convert_predictions_to_BIDS.py --pred-folder /path/to/predictions --out-folder /path/to/output/folder --conversion-dict /path/to/conversion/dict | ||
| ~~~ | ||
|
|
||
| If you are converting predictions which are not the evaluations set from nnUNet, use the flag `--not-imageTs`. | ||
|
|
||
| Then, you can evaluate the lesion prediction with Anima metrics | ||
|
|
||
| ~~~ | ||
| python evaluate_lesion_seg_prediction.py --pred-folder path/to/predictions --dataset path/to/dataset --animaPath path/to/animaSegPerfAnalyzer --output-folder path/to/output_folder | ||
| ~~~ | ||
|
|
||
| # Evaluation analysis | ||
|
|
||
| The following Notebook `nnUNet_inference_analysis.ipynb` was used to perform analysis of the nnUNet segmentations. | ||
| To use it, change the path to the csv files. |
Oops, something went wrong.