- Image registration's importance in medical imaging is underrepresented at MICCAI, attributed to the complexity of applying deep learning to its ill-posed nature w/o real ground truth.
- There is still much scope for novel methods (research gap) to improve clinical impact and there is not yet a one-fits all solution available
- The fragmented image registration community would benefit from specialised gatherings to enhance collaboration and research ties.
- Challenges: Learn2Reg ’20-‘24, OncoReg ’23-‘24 to provide much needed standardised training data and evaluation metrics, comprising 12 datasets with labels and many active leaderboards participate at learn2reg.grand-challenge.org
- 10th Workshop on Biomedical Image Registration organised by Marc Modat, Ivor Simpson, Žiga Špiclin and many more for the first time at MICCAI ’24 see https://wbir.info for programme and papers
- New 2024 initiative to collect well-explained code that is applicable out-of-the-box to at least two Learn2Reg tasks ➞ github.com/sigbir
We always welcome new members to SIGBIR, just send an email to miccai (dot) sigbir (at) gmail (dot) com to be invited to our next meeting.