Self-Supervised Masked Autoencoders with Dense-Unet for Coronary Calcium Removal in limited CT Data
This addresses the challenge of limited labeled data for medical image analysis, offering an incremental improvement for clinical diagnosis of coronary artery disease.
The paper tackled the problem of coronary calcification artifacts in CT angiography by proposing a self-supervised learning framework, Dense-MAE, which improved inpainting accuracy and stenosis estimation in few-shot scenarios compared to training from scratch.
Coronary calcification creates blooming artifacts in Computed Tomography Angiography (CTA), severely hampering the diagnosis of lumen stenosis. While Deep Convolutional Neural Networks (DCNNs) like Dense-Unet have shown promise in removing these artifacts via inpainting, they often require large labeled datasets which are scarce in the medical domain. Inspired by recent advancements in Masked Autoencoders (MAE) for 3D point clouds, we propose \textbf{Dense-MAE}, a novel self-supervised learning framework for volumetric medical data. We introduce a pre-training strategy that randomly masks 3D patches of the vessel lumen and trains the Dense-Unet to reconstruct the missing geometry. This forces the encoder to learn high-level latent features of arterial topology without human annotation. Experimental results on clinical CTA datasets demonstrate that initializing the Calcium Removal network with our MAE-based weights significantly improves inpainting accuracy and stenosis estimation compared to training from scratch, specifically in few-shot scenarios.