Jump to content

Label-Free Live Cell Imaging: AI Boosts Analysis of T-Cell Killing Assays

SBI2, Boston, MA, USA
September 19, 2022

To best facilitate personalized cancer treatment, functional activity of immune cell therapeutics must be confirmed in immune-mediated killing assays, in which diverse cell line and patient-derived primary tumor co-cultures are often live imaged using automated microscopy. Traditionally, such assays require labeling and detection of cells using fluorescent dyes or expressed nuclear proteins and are therefore subject to artifacts like phototoxicity and bleaching. Furthermore, the visual phenotype in such assays can vary along the measured time course as well as across different human tissue types or donors, creating the need to constantly adapt analysis parameters when using classical image analysis methods reliant on cell segmentation.
Here we show a new experimental approach employing brightfield images only and thus requiring no fluorescent markers for analysis. To enable this experimental approach, we developed a hands-free, scalable AI-based analysis workflow for quantification of T-cell mediated killing in brightfield-acquired images. We benchmarked this new workflow against current standard, semi-manual, cell segmentation-based analysis of fluorescent images. We found that the new workflow performs well on phenotypically diverse tumor cells, does so with greater efficiency, and produces results of equivalent consistency. We thus conclude that our AI-based analysis workflow is suitable for image-based T-cell mediated killing assays and allows efficient analysis of brightfield images, which can be time-consuming and difficult to analyze using classical segmentation-based analysis.

Request Resource

By submitting my data, I give consent to the collection, processing and use of my personal data in accordance with the Genedata privacy policy