Tomogram exploration through template matching and deep learning

Template matching (TM) is a long-standing tool in cryo-electron tomography (cryo-ET), where it has been applied to localize macromolecular complexes across a wide range of sample types. Recent algorithmic and computational advances have substantially increased the speed, accuracy, and practical resolution of the technique, enabling TM with fine angular sampling and voxel sizes that allow extraction of structural information beyond particle detection and toward exploration of higher-order spatial organization. In addition, deep learning (DL) approaches have emerged as powerful alternatives for particle picking and segmentation. Together, TM and DL form complementary components of modern cryo-ET workflows. This review discusses recent methodological advances in both approaches and highlights how they are increasingly integrated into cutting-edge pipelines for tomography exploration.