ETSAM: effectively segmenting cell membranes in cryo-electron tomograms

Cryogenic Electron Tomography (cryo-ET) is an emerging experimental technique used to visualize cell structures and macromolecules in their native cellular environment. Accurate segmentation of cell structures in cryo-ET tomograms, such as cell membranes, supports the study of cellular organization and function. However, several inherent limitations in cryo-ET, including the very low signal-to-noise ratio, missing wedge artifacts from limited tilt angles, and other noise artifacts, collectively hinder the reliable identification and delineation of these structures. In this study, we present ETSAM, a two-stage Segment Anything Model 2 (SAM2)-based fine-tuned AI method that effectively segments cell membranes in tomograms. It is trained on a diverse dataset comprising 83 experimental tomograms from the CryoET Data Portal (CDP) database and 28 simulated tomograms generated using PolNet. On an independent test set of 17 experimental tomograms for which ground-truth annotations are available, ETSAM outperforms other deep learning based methods by robustly identifying cell membranes with high sensitivity while also maintaining better precision. Cell membranes are hard to distinguish in the noisy 3D volumes from cryo-electron tomography. Here, authors adapt a video segmentation foundation model to the task, segmenting membranes across diverse organisms more accurately than existing tools.