Semiautomatic Segmentation of Nasal Airway Based on Collaborative Environment

Abstract

Nasal airway is studied to understand physiological and pathological characteristics of nasal breathing. It is difficult to segment nasal airway from Computed Tomography (CT) images because of its anatomical complexity. In this paper, we propose a software tool for semi-automatic segmentation of nasal airway. The segmentation result can be visualized as its three dimensional (3D) model. For the segmentation, 3D Region Growing is applied because nasal airway can be recognized by grouping similar pixel values of respiratory paths from CT images. The volume rendering method is used for the 3D visualization of nasal airway. The validation of the segmentation software tool was exercised by otolaryngologists. The accuracy of the segmentation was as good as manual segmentation by otolaryngologists. The processing time was greatly reduced from 5 hours to 3 minutes when 181 slices of CT images with the resolution of 512 by 512 were used to segment and visualize a 3D nasal airway model.