Cryo-Electron Microscopy

Cryo-electron microscopy studies biological structures by imaging frozen samples and inferring three-dimensional shapes from noisy two-dimensional particle images or tilt-series volumes. Work spans particle picking, pose estimation, denoising, 3D reconstruction, tomography segmentation, and refinement of atomic models, using statistical methods alongside deep learning, generative models, and geometric or topological constraints. Methods are evaluated on simulated and experimental particles, density maps, and tomograms, with attention to reconstruction resolution, structural accuracy, and downstream tasks such as localization and filament tracing. A central challenge is preserving weak signals and conformational variation despite noise and imaging artifacts, while ensuring that gains on synthetic data or intermediate metrics translate to reliable real-world structures.