zax130 / attention-reg

Attention-Reg

This repository is an implementation of "Cross-modal Attention for MRI and Ultrasound Volume Registration" accepted by MICCAI 2021.

• We have prepared 9 dummy samples for demo, which can be viewed by running "python view_sample.py". Samples are stored in the sample folder. Six samples are for training, two for validation, and one for testing.

• In practice, we initialize new transformations and resample for network input every epoch. For the purpose of demonstration, the dummy samples are the resampled result of random initialzations.

• To train a model, run "python train_network.py". To test your model, add the model name in test_network.py and run python test_network.py.

Our network is implemented through PyTorch 1.6.0, and the code for the network can be found at "networks/generator.py".