SIG Biomedical Image Registration (SIG-BIR)

MICCAI Special Interest Group in Biomedical Image Registration (SIG-BIR)

Motivation:

Image registration is a vital part in medical image analysis and computer-assisted interventions. Yet, with the advent of deep learning in image analysis that has shifted focus to automatic segmentation and classification techniques, the impact of registration is not well reflected in the number of MICCAI papers, workshops and challenges. We believe this is due to the difficulties in advancing learning techniques for a problem that is inherently ill-posed and has no viable ground truth training data. In addition, the connections across worldwide research groups are only loose and as a smaller community would benefit from specialised gatherings.

Mission: 

The main mission of SIG-BIR will be the strengthening of exchange between different research groups in the field of biomedical image registration. This will be facilitated by organising one satellite event at MICCAI conferences and one biennial international workshop (WBIR). In addition tutorials and smaller workshops organised by members/interested individuals will be considered for financial support.

Secondly, we aim to provide and curate public data sets for evaluation and training of medical registration algorithms. Multiple examples, where datasets, annotations and an evaluation platform is already provided can be found at our learn2reg.grand-challenge.org website. Hence, SIG-BIR will support the organisation of future Learn2Reg Challenges or similar events and will be liasing closely with the MICCAI SIG on Challenges

In addition, we want to develop best practices and educational initiatives to facilitate entry into the field of image registration and open community for more diverse participation. In particular, we plan to enable research from low-income countries through sponsored access to GPU cloud computing for training registration networks in a federated data-privacy-preserving way. 

Committee: