Robust workflow for diaphyseal cortical bone thickness calculation.

Abstract

Cortical bone, prominently found in the diaphyseal region of long bones, can resist higher ultimate stresses than trabecular bone and serves as the primary load-bearing compartment of the skeleton. The importance of cortical bone in determining mechanical strength and assessing fracture risk has been highlighted in both experimental and computational studies, motivating the need for better understanding through large-scale analysis. To support large data processing, an automated technique for measuring cortical thickness from clinical CT scans with sub-millimetre accuracy was introduced by Treece et al. (2012). However, this method struggles to reconstruct and calculate cortical bone thickness across diverse long bone morphologies. In this study, we present an adapted version of the technique with improved robustness. When the 2012 published method is evaluated on 240 long bones across six types, it resulted in failures across all test cases, with mean failure rates of 2.9%, 8.0%, 10.5%, 13.7%, 17.9%, and 24.8% in humerus, femur, radius, ulna, tibia and fibula, respectively. In contrast, the proposed new method eliminated failures in all bones except for the fibula, where 9 out of 40 test cases failed with a reduced mean failure rate of 1.9%. These results demonstrate that the new method broadens the applicability of the previous approach by robustly handling morphological variation, making it more suitable for large-scale studies. We anticipate the proposed workflow will serve as a valuable resource for analysing datasets with population-level variability and improving our understanding of osteogenic phenomena in clinically meaningful contexts.