COMULIS Multimodal Image Registration Challenge

Image registration/alignment is a key requirement for correlative analysis. Today this is often performed using ad-hoc solutions, with a lot of manual effort and poor reproducibility. Aiming to reach more efficient and standardized approaches, including objective evaluation, COMULIS has initiated work towards a COMULIS Multimodal Image Registration Challenge,  to identify the best available techniques to address this task.

To organize such a challenge, we need suitable Multimodal Datasets for methods’ evaluation, and we are therefore asking for your support to help us collect such data. We observe the following requirements on the data:

1. At least two modalities showing the same sample, preferably in 3D
   • We look for a broad range of different types of image data at all scales
2. Can be publicly shared (presumably already published at least once)
3. Ground truth alignment information is available
   • either already aligned image volumes (e.g. from integrated imaging)
   • or reliable fiducials/ROIs (withheld markers or reliable expert annotations)

In organizing a Challenge that utilizes your dataset, we envision world-wide attention to your data (leading to citations), a publication in a high-ranked journal about the challenge & results with you, and in particular, focused effort on finding, refining and developing methods for solving the image registration problems which you and your colleagues may be struggling with.

In case you have a dataset that fits the above three requirement, and would like to support the community (and receive fame, publications and citations) please write a short email to Joakim Lindblad  (with andreas.walter@vbcf.ac.at in Cc) and we can discuss how to best proceed.