Automated Computer Vision and Dose-Response Modeling Improve Throughput and Accuracy of an Ex Vivo Functional Precision Medicine Platform

Functional Precision Medicine platforms, which investigate the dynamic behavior of a patient’s tumor ex vivo to inform personalized treatment, face unique obstacles to clinical translation. These include limited access to patient tissue and stringent demands for intra-platform accuracy and consistency. In this study, an automated data analysis pipeline addresses these concerns for an organotypic brain slice culture-based functional assay by combining computer vision and dose-response modeling approaches. A 99% reduction in analysis time increases the amount of patient tissue that can be processed on the platform. Comparing automated measurements to previously published manual results revealed that automation increased consistency both within experiments and across replicate experiments. This pipeline also explores implementing complex CV with limited resources, modeling a unique and diverse dataset, and validating automated analysis when no gold standard measurements exist, obstacles that hinder automation efforts across scientific disciplines.