Code for paper "Self-Supervised Learning for Medical Image Data with Anatomy-Oriented Imaging Planes." （See fig below）
Abstract. Self-supervised learning has emerged as a powerful tool for pretraining deep networks on unlabeled data, prior to transfer learning of target tasks with limited annotation. The relevance between the pretraining pretext and target tasks is crucial to the success of transfer learning. Various pretext tasks have been proposed to utilize properties of medical image data (e.g., three dimensionality), which are more relevant to medical image analysis than generic ones for natural images. However, previous work rarely paid attention to data with anatomy-oriented imaging planes, e.g., standard cardiac magnetic resonance imaging views. As these imaging planes are defined according to the anatomy of the imaged organ, pretext tasks effectively exploiting this information can pretrain the networks to gain knowledge on the organ of interest. In this work, we propose two complementary pretext tasks for this group of medical image data based on the spatial relationship of the imaging planes. The first is to learn the relative orientation between the imaging planes and implemented as regressing their intersecting lines. The second exploits parallel imaging planes to regress their relative slice locations within a stack. Both pretext tasks are conceptually straightforward and easy to implement, and can be combined in multitask learning for better representation learning. Thorough experiments on two anatomical structures (heart and knee) and representative target tasks (semantic segmentation and classification) demonstrate that the proposed pretext tasks are effective in pretraining deep networks for remarkably boosted performance on the target tasks, and superior to other recent approaches.
- Data Science Bowl Cardiac Challenge (DSBCC) dataset: https://www.kaggle.com/c/second-annual-data-science-bowl/data
- Automated Cardiac Diagnosis Challenge (ACDC) dataset: https://www.creatis.insa-lyon.fr/Challenge/acdc/databases.html
- Data and Enverioment preparing: More requirements and virtual environment creation cound be seen in requirements.txt. Download the dataset and run
preprocess.pyto generate label for Pretext tasks. - Pretraining by pretext tasks: Replace the datapath in the training command of
./CMR_pretrain/pretrain.py, and run it. After training, the pretrained model will be stored in./CMR_pretrain/experiments. - Training for Downstream task: Replace the datapath and pretrain model path in the training command of
./CMR_downstrem/train.py, and run it. After training, the segmentation model will be stored in./CMR_downstrem/experiments. - Testing for Downstream task: Replace the datapath and checkpoint path in the testing command of
./CMR_downstrem/predict_acdc_V1.py, and run it, which will do the test and print the result of downstream segmentation task.
@article{zhang2024self,
title={Self-supervised learning for medical image data with anatomy-oriented imaging planes},
author={Zhang, Tianwei and Wei, Dong and Zhu, Mengmeng and Gu, Shi and Zheng, Yefeng},
journal={Medical Image Analysis},
pages={103151},
year={2024},
publisher={Elsevier}
}
If you have any questions, please do not hesitate to contact gsbyztw@163.com