Smart machine vision system for lung organoids analysis

Abstract

Organoids are cell structures generated in the laboratory through three-dimensional culture techniques that simulate the microenvironment and biological characteristics of in vivo organs. This study developed a machine vision system to analyse and quantify the growth dynamics of A549 lung cancer organoids in Matrigel spherical dishes. The system uses a Mask R-CNN deep learning network for instance segmentation, achieving precise localization of A549 organoids and tracking their growth over time. It automatically calculates key parameters such as the number, average size, and average roundness of organoids. By analyzing samples at different time points, the system provides quantitative descriptions of organoid growth dynamics. Preliminary results show that the system exhibits high accuracy in organoid identification and parameter measurement. This method enhances data collection and analysis efficiency, reduces human error, and offers a reliable tool for lung cancer research and drug screening.