G-UNETR++: Liver tumor segmentation

---

🔥G-UNETR++: Liver tumor segmentation from CT images
Paper: G-UNETR++

---

Requirements

Our code is based on UNETR++ code.
But, we modified the code for easy implementation. Our GPU is RTX 3090 GPU.

Environment

1. Create and activate conda environment

conda create --name gunetr_pp python=3.9
conda activate gunetr_pp

2. Install pytorch

# cuda 11.3
conda install pytorch==1.12.0 torchvision==0.13.0 torchaudio==0.12.0 cudatoolkit=11.3 -c pytorch

It is important that check your cuda version.
Please, see the pytorch document.

3. Install other dependencies

pip install -r requirements.txt

---

Dataset

In paper, we teseted LiTS, and 3Dircadb.

Dataset format

GUNETR_pplus_LiTS
├── DATASET_Synapse                  
│   ├── unetr_pp_raw
│       ├── unetr_pp_raw_data           
│           ├── Task02_Synapse           
│               ├── Task002_Synapse         
│                   ├── seg_gt
│                       ├── 3Dircadb
│                       ├── LiTS
│                           ├── segmentation-3.nii
│                           ├── ...
│                           └── segmentation-123.nii
│                   ├── unetr_pp_Data_plans_v2.1_stage1
│                       ├── 3Dircadb
│                       ├── LiTS
│                           ├── volume-3.nii
│                           ├── ...
│                           └── volume-123.nii
│                   ├── masking
│                       ├── 3Dircadb
│                       ├── LiTS
│                           ├── masking-3.nii
│                           ├── ...
│                           └── masking-123.nii
│                   └── unetr_pp_Plansv2.1_plans_3D.pkl

LiTS dataset: 131 cases.
3Dircadb link: 20 cases.

Our LiTS-testset number is 3, 17. 18, 28, 33, 37, 43, 64, 70, 77, 80, 90, 100, 104, 110, and 123.

Make masking files

Please see whole_liver_segmentation.

Model Checkpoint

GUNETR_pplus_LiTS
├── output_synapse                 
│   ├── 3d_fullres
│       ├── Task002_Synapse                   
│           ├── unetr_pp_trainer_synapse__unetr_pp_Plansv2.1        
│               ├── fold_4
│                   ├── validation_raw
│                   ├── model_final_checkpoint.model
│                   └── model_final_checkpoint.model.pkl

Final-model-chekcpoint: link.

---

Implementation

1. Make masking files
   Using LiTS Whole liver segmentation model.

2. Make npy files (automatic apply masking)

$> python LiTS_npy_make.py

You select the options, LiTS, and 3Dircadb.

3. Evaluation script

$> cd ./evaluation_scripts
$> sh run_evaluation_synapse.sh

You select the options, LiTS, and 3Dircadb.

4. Calculation metrics Please see our jupyter notebook.
   We implemented all of metric classes.

You can control post-processing option through flag_post = True.

---

Result

LiTS

Model	DSC	VOE	RAVD	ASSD	RMSD
Chen et al.	0.711	0.401	0.023	7.201	13.445
Chen et al.	0.705	0.395	0.534	8.286	13.680
Chen et al.	0.742	0.367	0.107	5.996	10.853
Jiang et al.	0.762	0.371	0.012	---	---
Ours (G-UNTER++)	0.844	0.263	0.133	1.317	3.189

---

References

UNETR++

---

Citation

@ARTICLE{
  title={G-UNETR++}, 
  author={},
  journal={}, 
  year={2024},
  doi={}}