Bone volume measurement of human femur head specimens by modeling the histograms of micro-CT images

Micro-CT analysis is a powerful technique for bone characterization. It allows to obtain histomorphometric parameters of bone sample. To do this, some different levels of data processing are required, a binarization of the images is usually performed as first step. Binarization consists in establishing a gray-level threshold in order to assign each pixel to bone or void spaces. A wrong choice of this threshold induces an overestimation or underestimation of the parameters. The main problem is represented by the quantization error due to spatial sampling, causing partial volume artifacts. Each voxel which samples the external surface of the tissue represents an average of both air and bone and the simple thresholding process operates an incorrect discrimination of the two materials. To overcome this limitation, the aim of this study is the extraction of bone volumetric information by fitting and reconstruction of the grey-levels histogram with a suitable set of functions. © 2012 Taylor & Francis Group.