The Real World Strikes Back – Failures and Biases of Brain Segmentation

The next LIVIA seminar will be held on Thursday, July 20 at 12h00 in hybrid mode.

Title: The Real World Strikes Back – Failures and Biases of Brain Segmentation

by Clemens Pollak, Research Associate, Deep-MI Lab, German Center for Neurodegenerative Diseases

Abstract: 

In the age of deep learning the accuracy and speed of brain segmentation is ever increasing. In the medical imaging domain, however, high average accuracy on pre-defined datasets is not good enough. Large population studies require performant tools that avoid or reduce potential biases in downstream analysis and can be applied to data from all participants, indifferent of large anatomical or pathological variance. We will approach this challenge, first by quantifying motion artifacts, caused by head motion during the MR acquisition, to reduce detrimental biases in analysis of brain changes; and second, by improving the performance of deep learning-based brain segmentation in the presence of large lesions, by generating healthy looking priors.

Bio: 

Pierre LE JEUNE is a PhD student at L2TI laboratory, University Sorbonne Paris Nord while working at COSE company. He received the M.Sc. degree in Mathematical Modelling and Computation from Danish Technical University (Copenhagen) and the M.Sc. in engineering from Centrale Nantes. His current research interests include Few-Shot Learning, Computer Vision and Deep Learning

Paper:  https://arxiv.org/abs/2306.11982

* In person: ETS-LIVIA, room A-3600.

* Zoom link: https://etsmtl.zoom.us/j/84820130813

Meeting ID: 848 2013 0813