Peter Soar

BoneGraph combines a SPECTER2-indexed bone corpus, knowledge graph, deterministic physics checks, and user-taught rules in a local five-tab system; oracle passages raise accuracy 42% to 78%.

The paper evaluates a direct strain prediction method for bone using XCT images, showing improved accuracy and reduced noise compared to displacement-derived strain.

This paper introduces D2IM, a deep learning model that predicts displacement and strain fields from undeformed X-ray images of bone using data from digital volume correlation.