A new method of segmentation of compact-appearing, transitional and trabecular compartments and quantification of cortical porosity from high resolution peripheral quantitative computed tomographic images.

Abstract

A transitional or cortico-trabecular junctional zone exists at any location composed of both cortical and trabecular bones such as the metaphyses of tubular bones and short bones like the femoral neck. The transitional zone comprises the inner cortex adjacent to the medullary canal and trabeculae abutting against the cortex contiguous with the endocortical surface. This is a site of vigorous remodeling. Intracortical remodeling cavitates the inner cortex expanding this transitional zone at the price of compact-appearing cortex so that it contains porosity, cortical fragments that resemble trabeculae, and trabeculae abutting the eroding cortex. The porosity of the transitional zone is an important source of bone loss. It reduces bone strength exponentially and is a quantifiable `fingerprint' of structural deterioration. A new automated method of segmentation of bone from background and bone into its compact-appearing cortex, transitional zone, and trabecular compartment is described, with a new approach to quantification of cortical porosity. Segmentation is achieved by automatically selecting attenuation profile curves perpendicular to the periosteal surface. Local bone edges are identified as the beginning and the end of the rising and falling S-shaped portions of the curve enabling the delineation of the compartments. Analyzing ~3600 consecutive overlapping profiles around the perimeter of each cross-sectional slice segments the compartments. Porosity is quantified as the average void volume fraction of all voxels within each compartment. To assess accuracy at the distal radius and tibia, μCT images of cadaveric specimens imaged at 19 μm voxel size served as the gold standard. To assess accuracy at the proximal femur, scanning electron microscopy (SEM) images of specimens collected at 2.5 μm resolution served as the gold standard. Agreement between HRpQCT and the gold standards for segmentation and quantification of porosity at the distal radius and tibia ranged from R(2)=0.87 to 0.99, and for the proximal femur ranged from 0.93 to 0.99. The precision error in vivo for segmentation and quantification of porosity in HRpQCT images at the distal radius, given by the root mean square error of the coefficient of variation, ranged from 0.54% for porosity of the transitional zone to 3.98% for area of the compact-appearing cortex. Segmentation of the transitional zone minimizes errors in apportioning cortical fragments and cortical porosity to the medullary compartment and so is likely to allow accurate assessment of fracture risk and the morphological effects of growth, aging, diseases and therapies.