In recent years, we can get high quality images in the short time for the progress of X-ray CT scanner. And the three dimensional (a-D) analysis of pulmonary organs using multidetector-row CT (MDCT) images, is expected. This paper presents a method for classifying lung field into each lobe using MDCT images of the whole lung field. It is possible to recognize the position of nodule by classifying lung field into these domains. The lung structures differ on the right one and left one. The right lung is divided into three regions (upper lobe, middle lobe, lower lobe) by major fissure and minor fissure. And, the left lung is divided into two regions (upper lobe, lower lobe) by major fissure. Watching MDCT images carefully, we find that the surroundings of fissures have few blood vessels. Therefore, lung field is classified by extraction of the domain where the distance from pulmonary blood vessels is large and connective search of these extracted domains. These extraction and search are realized by 3-D inverse distance transform and 3-D Hough transform. And also blood vessels are divided using information that classified lung field.
展开▼