SegmentAnything helps microscopy images based automatic and quantitative organoid detection and analysis

Xiaodan Xing,Chunling Tang,Yunzhe Guo,Nicholas Kurniawan,Guang Yang

2024-04-08

Abstract:Organoids are self-organized 3D cell clusters that closely mimic the architecture and function of in vivo tissues and organs. Quantification of organoid morphology helps in studying organ development, drug discovery, and toxicity assessment. Recent microscopy techniques provide a potent tool to acquire organoid morphology features, but manual image analysis remains a labor and time-intensive process. Thus, this paper proposes a comprehensive pipeline for microscopy analysis that leverages the SegmentAnything to precisely demarcate individual organoids. Additionally, we introduce a set of morphological properties, including perimeter, area, radius, non-smoothness, and non-circularity, allowing researchers to analyze the organoid structures quantitatively and automatically. To validate the effectiveness of our approach, we conducted tests on bright-field images of human induced pluripotent stem cells (iPSCs) derived neural-epithelial (NE) organoids. The results obtained from our automatic pipeline closely align with manual organoid detection and measurement, showcasing the capability of our proposed method in accelerating organoids morphology analysis.

Image and Video Processing,Computer Vision and Pattern Recognition,Quantitative Methods

What problem does this paper attempt to address?

The paper aims to address the following issues:

1. **Limitations of manual analysis**: Currently, the process of manually labeling organoids in microscope images is very time-consuming and labor-intensive. To overcome this challenge, researchers have proposed an automated method to identify and analyze the morphology of organoids.

2. **Limitations of existing algorithms**: Although deep learning algorithms such as StarDist and Cellos are used for organoid detection, these methods require a large amount of annotated data for training and have limitations when processing different modalities of microscope images. Additionally, each different type of microscope image requires retraining the model, which limits their practical application scope.

To address the above issues, this paper proposes a comprehensive pipeline based on the SegmentAnything model, leveraging its strong generalization ability to automatically identify and segment individual organoids in microscope images. Specifically, this method not only accurately identifies organoids but also introduces a series of morphological attributes (including perimeter, area, radius, roughness, and non-circularity) to achieve quantitative and automated analysis of organoid structures. Experimental results show that this automated pipeline is highly consistent with manual detection and measurement results, demonstrating its effectiveness in accelerating organoid morphology analysis. Moreover, this method can achieve efficient application without requiring a large amount of annotated data.