Geometric Machine Learning for Medical Image Analysis
Hervé Lombaert - Polytechnique Montréal
Nov. 6, 2026, 3:30 p.m. - Nov. 6, 2026, 4:30 p.m.
ENGMC 11
Hosted by: Kaleem Siddiqi
How to analyze the shapes of complex organs, such as the highly folded surface of the brain? This talk will show how spectral shape analysis can benefit general learning problems where data fundamentally lives on surfaces. We exploit spectral coordinates derived from the Laplacian eigenfunctions of shapes. Spectral coordinates have the advantage over Euclidean coordinates, to be geometry aware, invariant to isometric deformations, and to parameterize surfaces explicitly. This change of paradigm, from Euclidean to spectral representations, enables a classifier to be applied *directly* on surface data, via spectral coordinates. Brain matching and learning of surface data will be shown as examples. The talk will focus, first, on the spectral representations of shapes, with an example on brain surface matching; second, on the basics of geometric deep learning; and finally, on the learning of surface data, with an example on automatic brain surface parcellation.
Hervé Lombaert is a Professor at Polytechnique Montreal, Associate Member of Mila, and holds a Research Chair in Shape Analysis in Medical Imaging, previously served as Canada Research Chair. His research intersects machine learning, statistical geometry, and medical imaging. His key achievements range from early graph-cut segmentation methods, deployed in hospitals worldwide, to recent spectral graph-based analysis and the first human atlas of the cardiac fibers, advancing both neurology and cardiology. A recipient of the Erbsmann Prize, Hervé has published over 80 articles, holds 5 patents, and serves on the editorial boards of the journals IEEE TMI and MedIA. He brings extensive experience from past roles at Microsoft Research, Siemens, Inria, and McGill, and has mentored multiple alumni to best-thesis awards.
