DALE-CT: Depth-Aware Foundation Models for Computed Tomography
Researchers have developed DALE-CT, a new family of 2D foundation models for processing computed tomography (CT) data. Built from scratch using a self-supervised learning approach called LeJEPA, DALE-CT incorporates a novel 3D depth-aware pre-training strategy with both automated and human-annotated supervision. This model achieved a Macro AUROC of 0.833 on the CT-RATE dataset for multi-abnormality detection, nearing the performance of state-of-the-art 3D vision-language models with less data and no textual supervision. AI
IMPACT Introduces a novel, data-efficient approach for medical image analysis, potentially improving diagnostic accuracy in CT scans.