Computer Vision News - February 2022

51 METGAN: Generative Tumour Inpainting boss , Daniel Rueckert ! I will be starting a postdoc at Imperial College London soon while retaining my remote position as team lead for AI in the Erturk-Lab. ” Thinking about next steps for this work, the team say their approach could be extended to other state of the art medical imaging domains. “ Themultimodal datawe see inmicroscopic images are also present in other imaging domains, likeMRI, where you have different kinds of sequences, ” Johannes explains. Izabela agrees: “ Yes, it would be interesting to try contrast MRI. Also, in our lab, we already have novel experiments in the pipeline, and we want to apply this method to improve our segmentations on those – not only for cancer imaging, but for other pathologies as well. ” not only in computer science, but also in biology. I think that’s a very cool thing! ” Outside of this paper, Izabela’s usual work involves research on machine learning for light sheet microscopy applications. Johannes’smaininterest isinunderstanding large biological network structures, such as vessels and neurons, using machine learning applications. “ My favorite tools for this used to be CNNs and segmentation, but I have shifted my research focus more towards graph learning or geometric deep learning, ” he reveals. We tell Johannes we have heard on the grapevine he has an exciting new position in the pipeline. “ Yes, in fact I was really interested to read your interview last month with my future Detection of cancer micrometastases in mouse lungs. Source: Cell .