Hello Biostars Community!
Analyzing live-cell imaging data, especially at the single-cell level, can be incredibly challenging. Whether it’s dealing with detailed segmentation of individual cells or struggling with irregular and strong background noise in fluorescence channels, we know how frustrating it can be. That’s why we released CellLocator, a free and open-source tool tailored specifically for images from the Incucyte® system!
Here’s what CellLocator can do to make your life easier:
- Precise label-free segmentation: Identify living and dead cells, calculate confluence, and analyze cell state directly from phase images.
- Robust denoising: Improve signal clarity and handle complex fluorescence backgrounds for consistent, reliable results.
- Detailed single-cell data export: Analyze fluorescence kinetics and cell viability, with per-image and single-cell data saved in CSV format.
- Fast and lightweight: Processes ~50 images per minute, even on a low-end laptop.
Ready to simplify your live-cell imaging analysis? Download CellLocator for free: CellLocator GitHub Repository
CellLocator has already proven its value in high-impact research:
- Pre-print: Ferroptosis propagates to neighboring cells via cell-cell contacts By Bernhard Röck and collaborators. https://www.biorxiv.org/content/10.1101/2023.03.24.534081v1.abstract
- Published: TBK1-associated adapters TANK and AZI2 protect mice against TNF-induced cell death and severe autoinflammatory diseases Led by Andrea Ujevic and team. https://www.nature.com/articles/s41467-024-54399-4
Project History
CellLocator originated in Spring 2021, when Bernhard Röck and I (Michael Vorndran) began developing an AI-powered tool for analyzing brightfield microscopy images.
Our initial vision was ambitious: a versatile platform for single-cell analysis that could also classify different types of cell death (e.g., ferroptosis, apoptosis). However, creating a reliable classifier for cell death types required a vast amount of meticulously labeled training data. Due to resource limitations, we were unable to generate a sufficiently large and diverse dataset to achieve this goal with the desired accuracy.
Consequently, we shifted our focus to providing a highly accurate and efficient tool for cell segmentation, fluorescence quantification, and kinetic analysis, which we believe offers significant value to researchers even without cell death type classification. While our early work included images from an ImageXpress® Micro 4 MD system, we focused CellLocator’s development and training specifically on Incucyte® brightfield images, due to the widespread use of this platform.
In 2022, our team (then called "Cell ImAIging") won the "Start-up Your Idea" competition and secured seed funding, followed by a GO-Bio initial grant.
Despite this promising start, we were unable to secure further funding in 2023 to continue the startup. Rather than abandoning the project, we decided to open-source CellLocator, making our powerful analysis tools freely available to the research community.
CellLocator’s deep learning models were trained using a novel method described in our paper on "Inconsistency Masks", enabling accurate segmentation and analysis even with limited training data. Its robustness and effectiveness have already been demonstrated through its use in several scientific publications.
Feel free to share your feedback and ask questions—we’d love to hear your thoughts!
Awesome news!
CellLocator's latest release just hit 10+ downloads! A massive thank you to everyone who's given our little cell segmentation app a try.
Don't forget to share your feedback and spread the word to your lab mates and colleagues!
Keep on segmenting! :D