Presented at The Society of Biomolecular Imaging and Informatics (SBI2)
Teresa Findley1, Rebecca Winfree1, Martin Engel2, & Ilya Goldberg1
1ViQi Inc., 2Inventia Life Science
Modern biomedical research increasingly relies on automation, with robust image-based assays driving rapid discovery. AI-driven assays can enhance both biopharma discovery and manufacturing.
In drug discovery, phenotypic profiling is essential for effectively identifying subtle effects in high-content drug screens. Existing imaging and analysis technologies are being challenged by the increased use of 3D cultures to benefit from their significant richness and corresponding complexity. ViQi’s AI-based toolkit AutoHCSTM has been used both for high-content assay optimization and treatment effect detection: when presented with bioprinted 3D brain cancer cultures treated with test compounds, it reliably distinguished candidate anti-cancer compounds across multiple doses that differ in mechanism. While phenotypic separation between compounds improved with increasing z-planes, it did not scale linearly with increasing z-planes. This finding enables assay workflow improvements optimized for image acquisition time, computing, and storage costs.
At the stage of biomanufacturing in biopharma, optimization relies on automated assays to monitor the process . Cells are ideal reporters for key variables that characterize the biomanufacturing workflow. ViQi’s Automated Viral Infectivity Assay (AVIA™) uses machine learning and brightfield microscopy to detect viral infection in live cells, outputting a quantitative infectivity measure without the need for sample preparation or fluorescent markers. Alternatively, AVIA can also be used for viral clearance through extended incubations, which can increase sensitivity to the limiting dilution range.
Both assays use high-throughput, image-based analysis, with AI interpreting complex visual data faster,more precisely and more generally than conventional methods.