Official PyTorch implementation of "Latent Diffusion Autoencoders: Toward Efficient and Meaningful Unsupervised Representation Learning in Medical Imaging". LDAE is a novel unsupervised framework for 3D medical imaging that combines a latent diffusion model with semantic controls.

Deep learning-based tool for automated segmentation of sclerosis lesions in 3D MRI brain scans

Quantitative analysis of 3D vascular tree images using texture features. Implementation of methods for classifying blood vessel structures and monitoring physical parameters using 3D co-occurrence matrices, run-length matrices, and gradient-based features. Includes simulated vessel tree models and analysis of real confocal microscopy data.

Knee bone and cartilage segmentation in 3D MRI