2003 年 41 巻 3 号 p. 187-196
This paper presents a method for the automated extraction of emphysematous lesions and quantitative evaluation of their distribution patterns using three-dimensional (3-D) CT images. Multi-detector row CT scanners enabled us to obtain volumetric CT images at high resolution. Most of studies of computer-aided diagnosis for pulmonary emphysema process and analyze low attenuation areas in CT images using two-dimensional image processing techniques. Few studies have used three-dimensional image processing techniques. It is possible to obtain the spatial distribution of emphysematous lesions by 3-D image processing techniques. We studied here the application of computer-aided diagnosis for pulmonary emphysema using 3-D image processing techniques. We extracted emphysematous lesions automatically using a region growing method. In order to analyze the distribution of the low attenuation areas detected, we examined the relation between emphysematous lesions and bronchi or blood vessels using Euclidean distance transformation or lung dividing. We applied the proposed method to real 3-D X-ray CT images. Experimental results showed that the method could extract emphysematous lesions and quantitatively evaluate the distribution patterns of the lesions.