微分幾何学に基づいた胸部thin-section CT像の特徴解析

概要

論文の詳細を見る
Lung cancer is known as one of the most difiicult cancer to cure, and the number of deaths that it causes is gradually increasing in Japan. It is important to detect the tumor at an early stage. In this paper, we describe an analysis method of the thoracic thin-section CT images to support a differential diagnosis of lung cancer. As the typical feature of the malignant lung cancer, we can observe the spicula around the abnormal tissue, and the convergence of the peripheral pulmonary blood vessels. We analyze these malignant features based on the differential calculus. As the expression of the spicula, we extract the ridges of the intensity field. The ridges are generalizations of local maxima for smooth function of independent variables. And to express the convergence of the peripheral pulmonary blood vessels, we use the intensity curve segmentation. The intensity curve segmentation are based on Gaussian curvature and mean curvature. We will show the possibility of their analysis effectiveness to differential diagnosis of lung cancer.
神戸市立工業高等専門学校の論文