This is the implementation of a paper titled: An Unsupervised Domain Adaptation Brain CT Segmentation Method Across Image Modalities and Diseases, currently under review by Journal Expert Systems with Applications. More details can be seen in our upcoming paper.
Python 3
CPU or NVIDIA GPU + CUDA CuDNN
In the first stage, we train a pseudo-CT image synthesis network to minimize the difference between the two modalities.
In the second stage, we use labelled pseudo-CT images (obtained from the first stage network) and unlabeled CT images to train domain adaptation segmentation network.
In the training process of the image synthesis network, inputting the image pairs selected by this method can improve the performance of downstream tasks.
Daqiang Dong: dongdaqiang@emails.bjut.edu.cn
Guanghui Fu: guanghui.fu@icm-institute.org; aslanfu123@gmail.com