Keisuke presenting at NITMB workshop on morphology and shape dynamics

Keisuke will be speaking at Representing and Learning Morphology and Shape Dynamics from Biological Data, a workshop hosted by the National Institute for Theory and Mathematics in Biology (NITMB), October 19-23, 2026, in Chicago, IL.

Talk Title: “An integral geometry approach to 3D biological shape analysis with Minkowski tensors”

Keisuke’s talk is on Tuesday, October 20.

As 3D microscopy generates ever more volumetric imaging data, shape has become a powerful readout of cell state and tissue health, but turning segmented 3D structures into meaningful, comparable shape measurements remains surprisingly hard. Keisuke will present Minkowski tensors (MinT), a shape descriptor from integral geometry that captures volume, surface area, curvature, and anisotropy in one compact representation. He will also introduce pykarambola, the lab’s open-source Python package for computing these descriptors directly from 3D segmentations. Across several benchmark datasets, this approach classifies cell and tissue shapes more accurately than standard tools, using far fewer features and no network training, pointing to a simple, interpretable foundation for shape analysis in biology.

Keisuke is looking forward to the discussions at the workshop.