Kilian Biasuz

MorphoDeep improves 3D cell segmentation by learning cell interfaces and junctions, achieving higher accuracy and speed than existing methods.

MorphoNet 2.0 improves segmentation accuracy and interpretability for 3D time-lapse imaging datasets through self-evaluation and correction tools.

This paper introduces an automated framework for naming cells in ascidian embryos, revealing consistent and variable features of embryogenesis across species and conditions.

The paper investigates how contact-dependent cell communications maintain morphological invariance during ascidian embryogenesis despite genomic divergence.