A Large Model-Derived\nAlgorithm for Complex Organoids\nwith Internal Morphogenesis and Digital Marker Derivation

Automated segmentation and evaluation algorithms have\nbeen demonstrated\nto enhance the simplicity and translational utility of organoid technology.\nHowever, there is a pressing need for the development of complex organoids\nthat possess epithelium environmental elements, dense regional cell\naggregation, and intraorganoid morphologies. Nevertheless, there has\nbeen limited progress, including both the construction of data sets\nand the development of algorithms, in the use of user-friendly microscopy\nto address such complex organoids. In this study, a data set of bright-field\nand living cell fluorescence images in paired forms and with temporal\nvariance was constructed using droplet-engineered lung organoids.\nAdditionally, a large model-based algorithm was developed. Both the\norganoid contours and intraorganoid morphologies were included in\nthe data set, and their physical parameters were included and screened\nto form multiplex digital markers for organoid evaluation. The algorithm\nhas been demonstrated to outperform existing methods and is therefore\nsuitable for the evaluation of complex organoids. It is expected that\nthe algorithm will facilitate the successful demonstration of AI in\norganoid evaluation and decision-making regarding their status.