KG-SAM: Injecting Anatomical Knowledge into Segment Anything Models via Conditional Random Fields
This addresses medical image segmentation problems for clinical applications by providing a more robust and generalizable approach, though it appears incremental as it builds on existing SAM models with domain-specific enhancements.
The paper tackles the challenge of applying Segment Anything Models to medical imaging by introducing KG-SAM, a framework that integrates anatomical knowledge graphs, conditional random fields, and uncertainty estimation, achieving average Dice scores of 82.69% on prostate segmentation and 78.05-79.68% on abdominal segmentation.
While the Segment Anything Model (SAM) has achieved remarkable success in image segmentation, its direct application to medical imaging remains hindered by fundamental challenges, including ambiguous boundaries, insufficient modeling of anatomical relationships, and the absence of uncertainty quantification. To address these limitations, we introduce KG-SAM, a knowledge-guided framework that synergistically integrates anatomical priors with boundary refinement and uncertainty estimation. Specifically, KG-SAM incorporates (i) a medical knowledge graph to encode fine-grained anatomical relationships, (ii) an energy-based Conditional Random Field (CRF) to enforce anatomically consistent predictions, and (iii) an uncertainty-aware fusion module to enhance reliability in high-stakes clinical scenarios. Extensive experiments across multi-center medical datasets demonstrate the effectiveness of our approach: KG-SAM achieves an average Dice score of 82.69% on prostate segmentation and delivers substantial gains in abdominal segmentation, reaching 78.05% on MRI and 79.68% on CT. These results establish KG-SAM as a robust and generalizable framework for advancing medical image segmentation.