Practical fluorescence reconstruction microscopy for large samples and low-magnification imaging

Fluorescence reconstruction microscopy (FRM) describes a class of techniques where transmitted light images are passed into a convolutional neural network that then outputs predicted epifluorescence images. This approach enables many benefits including reduced phototoxicity, freeing up of fluorescence channels, simplified sample preparation, and the ability to re-process legacy data for new insights. However, FRM can be complex to implement, and current FRM benchmarks are abstractions that are difficult to relate to how valuable or trustworthy a reconstruction is. Here, we relate the conventional benchmarks and demonstrations to practical and familiar cell biology analyses to demonstrate that FRM should be judged in context. We further demonstrate that it performs remarkably well even with lower-magnification microscopy data, as are often collected in screening and high content imaging. Specifically, we present promising results for nuclei, cell-cell junctions, and fine feature reconstruction; provide data-driven experimental design guidelines; and provide researcher-friendly code, complete sample data, and a researcher manual to enable more widespread adoption of FRM. Author summary Biological research often requires using fluorescence imaging to detect fluorescently labeled proteins within a cell, but this kind of imaging is inherently toxic and complicates the experimental design and imaging. Advances in machine learning and artificial intelligence can help with these issues by allowing researchers to train neural networks to detect some of these proteins in a transmitted light image without needing fluorescence data. We call this class of technique Fluorescence Reconstruction Microscopy (FRM) and work here to make it more accessible to the end-users in three key regards. First, we extend FRM to challenging low-magnification, low-resolution microscopy as is needed in increasingly popular high content screening. Second, we uniquely relate FRM performance to every-day metrics of value to the end-user, such as cell counts, size, and feature detection rather than to abstract performance metrics from computer vision. Third, we provide accessible software tools and characterizations of FRM intended to aid researchers in testing and incorporating FRM into their own research.