Bone MRI segmentation assessment based on synchrotron radiation computed microtomography

In this paper, we propose an original scheme to assess the efficiency of segmentation algorithms requested to quantify bone architecture from three-dimensional (3-D) micromagnetic resonance imaging ([mu]MRI) techniques. Our scheme takes advantage of the high-quality images provided by synchrotron radiation computed microtomography (SR [mu] CT). To perform this assessment, calcaneus samples were acquired from both SR [mu]CT and MRI devices. The segmentation process is evaluated by comparing morphometric and topologic parameters computed from the SR [mu]CT reference images and those computed from the MRI segmented images. Both an adaptive thresholding algorithm and a region-growing algorithm are tested within this framework. Index Terms--Bones quantization, image segmentation, magnetic resonance imaging (MRI), X-ray tomography.