MIDL Vision 2021

This paper is based on work he had published in IEEE Transactions on Medical Imaging (T-MI) earlier this year, titled ‘ Looking for abnormalities in mammograms with self-and weakly supervised reconstruction ’ . Mickael recommends that anyone interested in mammography should read the full version. Looking ahead, Mickael would like to apply these kinds of losses to other types of imaging, including MRI, and explore different operators. “ We focused on two operators, closing and top-hat , and they worked really well for us, ” he explains. “ But exploring different operators would be great and would open up opportunities to explore something trainable. We set a decent amount of hyperparameters in our network, which is fine, and everybody does this, but we would like to make something more generalizable , and next step would be to do this. ” To learn more about Mickael’s work, you can view his short oral video [H8] anytime and visit his discussion session during Application: Radiology today at 16:45-17:30 (UTC+2) 25 Mickael Tardy VISION MIDL