Computer Vision News - July 2020

2 We Tried For You 20 Note that, in order for this code to run, the training images and labels need to be given as a dictionary with keys ‘ image ’ and ‘ label ’. MONAI applications: Automatic Segmentation of COVID-19 Pneumonia Lesions from CT Images by G. Wang et al (2020) Conclusion We believe that Monai embodies a new essential type of framework that merges the best outcomes of optimal research environments. Its goal to integrate different available software rather than competing with them, well reflects the current desires and expectations of a community that is continuously evolving, and which needs, more than anything, a unified reference system. S tephen Aylward and Jorge Cardoso highlighted how big the community of MONAI already is and how it is expanding: “The most exciting news is the large number of existing code bases and lead developers that have already chosen to interface with or directly become a part of Project MONAI,” says Stephen. “Via their participation, MONAI has already become a vibrant community, with significant contributions from the developers of NiftyNet at KCL, BatchGenerator from DKFZ, ITK from the University of Iowa and Kitware, and many others. At this early stage, creating and maintaining an active community is the best sign of our potential for success that I could imagine. We welcome others to join us and bring their energy and code bases as well!” Jorge adds that MONAI is “interfacing with popular community tools suchas (Ignite, Lightning, BatchGenerators, Rising, TorchIO, ITK, OpenMined, Substra), and improving its educational/ tutorial section to also engage with the new members of the community.” Prerna concludes: “So far the project is well received and is part of official PyTorch ecosystem (https://pytorch.org/ ecosystem/) with more than 1K GitHub stars. All these early indicators validate our hypothesis that to truly accelerate the pace of innovation for Medical Imaging Researcher joining forces for an open- source initiative like MONAI is the right way forward.” Jorge Cardoso