ISBI Daily - Friday

16 Friday ISBI DAILY At the poster session Computer Aided Diagnosis, Mariëlle J.A. Jansen (University Medical Center Utrecht) presented her paper about liver lesion classification using features derived from dynamic contrast MRI added to T2-weighted MRI. Mariëlle’s supervisor is Hugo Kuijf . At the poster session Image Reconstruction and Restoration, Haris Jeelani and Jonathan Martin (University of Virginia) presented their paper about how image quality affects deep learning reconstruction of magnetic resonance imaging. “ What happens in clinical settings is that images have a lot of artefacts, so if you use these images to train your deep neural network, the predictions of the neural networks are going to be worse. We have come up with the weighted loss function which tries to avoid learning from artefacts and that’s how it improves the image quality. ” Images and more