OrganoLabeler: A Quick and Accurate Annotation Tool for Organoid Images.

Area of Science:

• Biotechnology
• Computational Biology
• Regenerative Medicine

Background:

• Organoids offer promising in vitro models for organ function and disease research.
• Generating organoids from human cells is often inefficient, costly, and time-consuming.
• AI tools can predict and select optimal cellular aggregates for functional organoid development.

Purpose of the Study:

• To develop an automated, user-friendly application for segmenting 3D cellular construct images.
• To create large, reliable image datasets for training deep learning models in organoid research.
• To provide a faster and more accurate alternative to manual image labeling for organoid studies.

Main Methods:

• Developed OrganoLabeler, an application utilizing contrast adjustment, K-means clustering, CLAHE, binary, and Otsu thresholding.
• Created embryoid body and brain organoid datasets for comparison.
• Trained U-Net deep learning models using both OrganoLabeler-segmented and manually labeled images.

Main Results:

• OrganoLabeler generated segmented images consistently, reliably, and rapidly.
• U-Net models trained with OrganoLabeler-segmented data achieved comparable or superior segmentation accuracy to those trained with manual labels.
• The application demonstrated significant improvements in speed and accuracy over manual labeling.

Conclusions:

• OrganoLabeler effectively automates image segmentation for organoid research.
• The tool can replace manual labeling, enhancing efficiency and accuracy in dataset creation.
• OrganoLabeler provides a free, valuable resource for advancing organoid development and AI-driven biological research.