A structure-based algorithm for automated separation of subchondral bone in micro-computed tomography data

Structural measurements of subchondral and trabecular bone are of interest for a wide variety of communities ranging from anthropology to biomechanical engi- neering, yet continues to be a challenge partly because of the lack of automated techniques for use with high resolution data. Here we present a structure-based algorithm for separating cortical compartments from trabecular bone in binarized images. Using the thickness of the cortex as a seed value, bone connected to the cortex within a spatially local threshold value is identified and separated from the remaining bone. The algorithm was tested on biological images from human, chim- panzee, and gorilla datasets and compared to manual measurements. The average error was 2-3 voxel differences in thickness and total area errors were less than ten percent. The algorithm is repeatable, efficient, and requires few user inputs, provid- ing a means of separating cortical from trabecular bone. The Matlab code, example images, and datasets can be downloaded from uitbl.mechse.illinois.edu.