3D Medical Imaging

3D medical imaging studies how to extract anatomical and clinical information from scans while preserving relationships across slices. CNNs, transformers, and self-supervised foundation models support tasks such as segmentation, lesion detection, registration, and report generation, evaluated on settings including BraTS brain MRI and LIDC-IDRI lung CT. For reconstruction and image enhancement, methods combine measured scan data and physical models with diffusion or flow models, neural fields, and Gaussian splatting. A central tension is that full-volume models capture spatial context but demand substantial computation, while slice-based approaches are cheaper; generalization across scanners and reliable clinical evaluation remain open challenges.