StarDist_BF_cancer_cell_dataset_20x

This repository contains a StarDist deep learning model and its training and validation datasets designed for segmenting cancer cells perfused over an endothelial cell monolayer captured at 20x magnification. Using computational methods, the initial dataset of 20 manually annotated images was augmented to 160 paired images. The model was trained over 400 epochs and achieved an average F1 Score of 0.921, demonstrating high accuracy in cell segmentation tasks.

Specifications

Model: StarDist for cancer cell segmentation on endothelial cells (20x magnification)

Training Dataset:

Number of Original Images: 20 paired brightfield microscopy images and label masks

Microscope: Nikon Eclipse Ti2-E, 20x objective

Data Type: Brightfield microscopy images with manually segmented masks

File Format: TIFF (.tif)

Brightfield Images: 16-bit

Masks: 8-bit

Image Size: 1024 x 1022 pixels (Pixel size: 650 nm)

Training Parameters:

Epochs: 400

Patch Size: 992 x 992 pixels

Batch Size: 2

Performance:

Average F1 Score: 0.921

Average IoU: 0.793

Model Training: Conducted using ZeroCostDL4Mic (https://github.com/HenriquesLab/ZeroCostDL4Mic/wiki)

Reference

Fast label-free live imaging reveals key roles of flow dynamics and CD44-HA interaction in cancer cell arrest on endothelial monolayers

Gautier Follain, Sujan Ghimire, Joanna W. Pylvänäinen, Monika Vaitkevičiūtė, Diana Wurzinger, Camilo Guzmán, James RW Conway, Michal Dibus, Sanna Oikari, Kirsi Rilla, Marko Salmi, Johanna Ivaska, Guillaume Jacquemet

bioRxiv 2024.09.30.615654; doi: https://doi.org/10.1101/2024.09.30.615654