Light Microscopy Combined with Computational Image Analysis Uncovers Virus-Specific Infection Phenotypes and Host Cell State Variability

Abstract

Abstract:
The study of virus infection phenotypes and variability plays a critical role in understanding viral pathogenesis and host response. Virus-host interactions can be investigated by light and various label-free microscopy methods, which provide a powerful tool for the spatiotemporal analysis of patterns at the cellular and subcellular levels in live or fixed cells. Analysis of microscopy images is increasingly complemented by sophisticated statistical methods and leverages artificial intelligence (AI) to address the tasks of image denoising, segmentation, classification, and tracking. Work in this thesis demonstrates that combining microscopy with AI techniques enables models that accurately detect and quantify viral infection due to the virus-induced cytopathic effect (CPE). Furthermore, it shows that statistical analysis of microscopy image data can disentangle stochastic and deterministic factors that contribute to viral infection variability, such as the cellular state. In summary, the integration of microscopy and computational image analysis offers a powerful and flexible approach for studying virus infection phenotypes and variability, ultimately contributing to a more advanced understanding of infection processes and creating possibilities for the development of more effective antiviral strategies.