Three Dimensional Reconstruction of the Inner Ear from CT Images

Abstract

Observation of the inner ear is necessary for the diagnosis of dizziness and hearing difficulties. It is, however, hard to judge its shape from two dimensional computer tomography (CT) images. Since the inner ear is a small and anatomically complex shape that is conjugated by three semicircular canals, it is difficult to confirm its shape from one fault surface of the human head. Many researchers address this problem using image processing methods. Recently, the usefulness of the region growing method has been reported for 3D reconstruction of human tissue. Since the semicircular canals of the inner ear area partly composed of thin tissue, it is difficult to extract the whole shape of the inner ear from a general region growing method. In this paper, we propose a modified region growing method for 3D reconstruction of the inner ear. To facilitate region growing, removal of regions adjacent to the inner ear is performed as a preprocessing process. As a result, we can make an accurate three dimensional model of the inner ear and calculate the angle of the semicircular canals.