Related Experiment Video
Updated: Sep 3, 2026
Label-free, High-Resolution 3D Imaging and Machine Learning Analysis of Intestinal Organoids via Low-Coherence Holotomography
Published on: August 12, 2025
A Single-organoid Workflow for quantitative Imaging classiFication and Tracking (SWIFT)
Lucie Bracq1, Romain Guiet2, Sandra Offner1
1Global Health Institute, School of Life Sciences, EPFL, Lausanne, Switzerland.
Abstract:
SWIFT (Single-organoid Workflow for quantitative Imaging classiFication and Tracking) is a fast and modular pipeline for quantitative, time-resolved organoid analysis from brightfield images. Requiring only minimal training annotations, SWIFT combines YOLOv8, SAM, and application-specific object classification to deliver accurate single-organoid segmentation, phenotyping, and tracking. The workflow is robust across organoid types and microscope systems. Applied to intestinal organoids, SWIFT uncovers dynamic Wnt-dependent morphological transitions, enabling scalable, high-throughput studies of epithelial plasticity.
