PROJECT SUMMARY/ABSTRACT High-throughput microscopy (HTM) can generate vast spatiotemporal information on cellular structures and dynamics under various conditions. Computational analyses of the whole aspect of such data could produce unbiased, systematic representations of cellular heterogeneity across multiple scales. Current HTM, however, is constructed by combining high-magnification microscopes with scanning stages; this configuration would entail high complexity in the system design and operation, high cost, and slow image acquisition rates. We reason that we can transcend the current HTM by integrating light microscopy and machine learning (ML). ML is potent in discovering intricate, hidden structures in high-dimensional datasets with limited human supervision. We will leverage ML’s power to learn important cellular features to create a high-throughput high- resolution live cell imaging system. Our goal is to develop a new HTM platform, termed Machine Learning Microscopy (MLM), that autonomously acquires high-resolution live cell movies in a high-throughput manner. The proposed MLM platform will leverage deep neural networks to allow for i) high-resolution, continuous imaging on live cells and ii) automated acquisition of single-cell and subcellular movies specific to target phenotypes. In the parent award, we will apply MLM to construct a detailed phenotypic map of cell migration and subcellular morphodynamics. The MLM will bring unprecedented analytical power to HTM by imaging large numbers of single cells at high spatial resolution and facilitating extracting many cellular and subcellular phenotypes. MLM can be applied to various areas of cell biology, such as cell division, cytoskeleton, membrane remodeling, and membrane-bound organelles.