Design and development of a flexible autonomous cell culture system

概要

Cell culture is a basic experimental technique in cell biology and regenerative medicine. However, culturing high-quality cells while maintaining reproducibility relies heavily on expert skills and tacit knowledge. Combining a newly developed AI software with LabDroid, a versatile dual-arm experimental robot, a variable-scheduling autonomous cell culture system was developed that can continuously produce subcultures of cell lines without human intervention. The system periodically observes the cells on plates with a microscope, predicts the cell growth curve by processing cell images, and determines the best times for passage. This system has been successful in maintaining the cultures of two different HEK293A cell lines with no human intervention for 192 hours.

This autonomous culture system can also use different modules. Convolutional Neural Network (CNN) is one method that can be applied as a module. The effects of different imaging methods on the accuracy of image processing have not been examined systematically. The effects of different imaging methods on the performance of machine learning-based cell type classifiers were studied. Lymphoid-primed multipotent progenitor (LMPP) and pro-B cells were observed using three imaging methods: differential interference contrast (DIC), phase contrast (Ph), and bright-field (BF) imaging. The classification performances of CNNs were examined with each of these imaging methods and their various combinations.

CNN achieved the same level of classification performance when using images captured by any of the BF, Ph, and DIC methods to infer the differentiation status of cells. Therefore, there is little need for hardware modification, such as the addition of a microscope, when applying CNN to automated experimental systems. By applying this CNN to the autonomous culture system, it is possible to discard cells, expand cultures, or change culture medium, depending on the quality as well as the quantity of cells, by changing the software. In future, the autonomous culture system developed in this study may alleviate the problems caused by a lack of human resources and high educational costs in cell biology research and regenerative medicine. It may also lead to acceleration of scientific discovery by the automation of exploratory experiments.