Tensor controlled local structure enhancement of CT images for bone segmentation

Abstract

This paper addresses the problem of segmenting bone from Computed Tomography (CT) data. In clinical practice, identification of bone is done by thresholding, a method which is simple and fast. Unfortunately, thresholding alone has significant limitations. In particular, segmentation of thin bone structures and of joint spaces is problematic. This problem is particularly severe for thin bones such as in the skull (the paranasal sinus and around the orbit). Another area where current techniques often fail is automatic, reliable and robust identification of individual bones, which requires precise separation of the joint spaces. This paper presents a novel solution to these problems based on three-dimensional filtering techniques. Improvement of the segmentation results in more accurate 3D models for the purpose of surgical planning and intraoperative navigation.