Computer Vision News - March 2023

38 Medical Intelligent Imaging concepts over the numerous physical issues that still need to be addressed globally. “ It’s important because if algorithms are biased, they can perpetuate health inequalities in our country in other places, ” Paul responds. “ We want to improve human health, but pieces like fairness and explainability are just as important as diagnosing disease because if these things are not solved, then we can’t achieve the ultimate goal. If an AI model says you have cancer, but we can’t confirm that, it will be hard to get anyone to trust it. If they don’t trust it, they won’t use it! ” Theteam’swork interestsabroadaudience, including journals such as RSNA’s Radiology and Radiology: Artificial Intelligence , and conferences and meetings, including the annual meeting of the Society for Imaging Informatics in Medicine ( SIIM ), the Conference on Machine Intelligence in Medical Imaging ( CMIMI ), the Medical Imaging with Deep Learning ( MIDL ) conference, and of course the Conference on Medical Image Computing and Computer Assisted Intervention ( MICCAI ). With these events still a few months away, what might we expect to see Paul and Vishwa presenting this year? “ We’re developing a federated perform better on males vs. females, or on one ethnic group compared to another? Few of us on the MD side evaluate these pitfalls. ” The team recently published a paper in RSNA’s Radiology journal showing biases in a leading bone age prediction model, indicating inconsistent performance across diverse populations. While they work to solve this problem, one may wonder why they have prioritized such intangible