In this research work, we propose a style matching U-Net to address the missing modality challenge for brain tumour segmentation on MRI modalities. The proposed Style matching mechanism decomposes the representational space into content and style representation and then uses conceptual loss to enforce knowledge distillation in a co-training strategy.
If this code helps with your research please consider citing the following paper:
R. Azad, Nika Khosravi and Dorit Merhof , "SMU-Net: Style matching U-Net for brain tumor segmentation with missing modalities", download link.
soon- March 30, 2022: First release will announce.
This code has been implemented in python language using Pytorch library and tested in ubuntu OS, though should be compatible with related environment. following Environement and Library needed to run the code:
- Python 3
- Pytorch
For training deep model and evaluating on the BraTA 2018 dataset set follow the bellow steps:
1- Download the BraTS 2018 train dataset from this link and extract it inside the dataset_BraTS2018 folder.
2- Run train.ipynb for training the model.
3- For performance calculation and producing segmentation result, run evaluation.ipynb.
Notice: our implementation uses the ACN codes: https://github.com/Wangyixinxin/ACN
For evaluating the performance of the proposed method, Two challenging task in medical image segmentaion has been considered. In bellow, results of the proposed approach illustrated.
All implementations are done by Reza Azad. For any query please contact us for more information.
rezazad68@gmail.com