CASSPER is a semantic segmentation-based particle picking algorithm for single-particle cryo-electron microscopy

Abstract Particle identification and selection, which is a prerequisite for high-resolution structure determination of biological macromolecules via single-particle cryo-electron microscopy poses major bottleneck automating the steps determination. Here, we present generalized deep learning tool, CASSPER, automated detection isolation protein particles in transmission microscope images. This tool uses Semantic Segmentation collection visually prepared training samples to capture differences intensities protein, ice, carbon, other impurities found micrograph. CASSPER semantic segmentation based method that does pixel-level classification completely eliminates need manual particle picking. Integration Contrast Limited Adaptive Histogram Equalization (CLAHE) enables high-fidelity micrographs with variable ice thickness contrast. A model works high efficiency on unseen datasets can potentially pick on-the-fly, enabling data processing automation.