LSDM: Long-Short Diffeomorphism Memory Network for Weakly-Supervised Ultrasound Landmark Tracking

This is the official public PyTorch implementation for our paper LSDM: Long-Short Diffeomorphism Memory Network for Weakly-Supervised Ultrasound Landmark Tracking, which was accepted by Medical Image Analysis.

TODO:

• Dependencies
• LSDM Code
• CLUST2D Dataset Preparation
• CLUST2D Submission Format

---

Acknowledgement

Part of the code is modified from DeepTag, SiamFC-PyTorch and SiamTrackers. Baseline models are from SiamTrackers, KCF, LCT. Thanks a lot for the great work!

The authors also thank Dr. Lin Zhang for her generous help during CLUST2D submission stage.

---

If you find LSDM useful in your research, please consider citing:

@article{liu2024lsdm,
        title = {Long-short diffeomorphism memory network for weakly-supervised ultrasound landmark tracking},
        author = {Zhihua Liu and Bin Yang and Yan Shen and Xuejun Ni and Sotirios A. Tsaftaris and Huiyu Zhou},
        journal = {Medical Image Analysis},
        volume = {94},
        pages = {103138},
        year = {2024},
        doi = {https://doi.org/10.1016/j.media.2024.103138},
}