TomoScore: A Neural Network Approach for Quality Assessment of Cellular cryo-ET

Electron cryo-tomography (cryo-ET) is a powerful imaging tool that allows three-dimensional visualization of subcellular architecture. During morphological analysis, reliable tomogram segmentation can only be achieved through high-quality data. However, unlike single-particle analysis or subtomogram averaging, the field lacks a useful quantitative measurement of cellular tomogram quality. Currently, the most prevalent method to determine cellular tomogram resolvability is an empirical judgment by experts, which is time-consuming. Methods like FSC between split tilt series suffer from severe geometrical artifacts. We address this gap with a neural network model to predict per-slice resolvability that can apply to tomograms collected from various species and magnifications. We introduce a novel metric, "TomoScore", providing a single-value evaluation of cellular tomogram quality, which is a powerful tool for pre-screening tomograms for subsequent automatic segmentation. We further explore the relationship between accumulated electron dose and resulting quality, suggesting an optimum dose range for cryo-ET data collection. Overall, our study streamlines data processing and reduces the need for human involvement during pre-selection for tomogram segmentation.