Segmentation and Analysis Method for Mitochondria Enhanced with Open-Source AI Tools in Holotomography Images

20241125 CBIAS Analysis Poster thumb nail.png

Demonstrates a multi-organelle segmentation strategy for HT images that combines CNN-based (StarDist, Cellpose), rule-based, and pixel-classification (ilastik) methods to simultaneously segment the cell body, nucleus, lipid droplets, and mitochondria, then applies the pipeline to quantify mitochondrial fission/fusion dynamics and lipid droplet accumulation during adipogenesis.

What's inside

- RI-to-molecular-concentration calibration for lipid droplets and mitochondria in an adherent Hep3B cell

- Segmentation schematics per compartment: CNN-based (StarDist/Cellpose) for cell/nucleus, rule-based threshold+TopHat for lipid droplets, pixel classification (ilastik) for mitochondria

- Dose-dependent (Mdivi-1 fusion inducer / MPP+ fission inducer) and time-dependent (0–32 min) quantification of mitochondrial morphology classes (fragmented/normal/elongated)

- Lipid droplet accumulation quantification (concentration and volume) over a 96-hour adipogenesis time course

Keywords

open-source AI segmentation StarDist Cellpose ilastik holotomography mitochondrial dynamics lipid droplet accumulation label-free imaging cell segmentation refractive index tomography adipogenesis

Download the Application Note

First name*

Last name*

Title

Company*

Region*

Country*

Interested Product*

E-mail*

Comment

I have read and agree to the Terms and Policies.