Sr. Scientist, Sr. Manager of Assay Development Molecular Devices, LLC San Jose, California, United States
Abstract: 3D organoid models are increasingly important for biological research and drug development, however because of complexity of steps those processes are difficult to automate. To enable automated control of cell culture, we developed the automation solution CellXpress.ai. CellXpress.ai contains four essential components for automated organoid culture: liquid handler, automated incubator and imager, plus integrated AI-powered software that that provides automated processing of complex protocols. Automated 3D organoid culture includes processes of plating organoid domes, periodic media exchanges, and periodic monitoring by imaging and analysis. It provides automated passaging of organoids, which can be triggered by AI-based organoid classification based on organoid phenotypes. We developed AI-based protocols and present results from the automation of three different organoid types: mouse intestinal organoids, human intestinal organoids, and patient-derived colorectal tumor organoids. 3D organoid cultures were started from seeding organoids into matrigel domes in 24 well plate format. We automated organoid culture protocols using media kits for mouse and human organoids recommended by STEMCELL Technologies. Media exchanges were done automatically every 24 hours. Passaging organoids were also performed automatically using liquid handling, but timing for passaging depends on maturation of organoids, which requires decision action done either by scientists of by software. Cultures were monitored by imaging every 12 hours with transmitted light and 4X magnification by integrated imager. Then machine-learning based image analysis allowed to detect organoids and provide analysis of organoid objects to measure variety of phenotypic read-outs including size, density, and texture measurements. Phenotypic classification of organoids for mature and immature phenotypes was done by pre-trained model that combined un-supervised and supervised machine learning. We created models for three tested organoid types. Passaging steps were then triggered automatically by a user-defined percentage and number of mature organoids in the culture, typically >50%. The AI-based classification allowed to fully automate 3D culture and expansion of organoids, also to increase productivity and throughput.
