Lung metastasis detection and visualization on CT images: a knowledge-based method

A solution to the problem of lung metastasis detection oil computed tomography (CT) scans of the thorax is presented. A knowledge-based top-down approach for image interpretation is used. The method is inspired by the manner in which a radiologist and radiotherapist interpret CT images before radiotherapy is planned. A two-dimensional followed by a three-dimensional analysis is performed. The algorithm first detects the thorax contour, the lungs and the ribs, which further help the detection of metastases. Thus, two types of tumors are detected: nodules and metastases located at the lung extremities. A method to visualize the anatomical structures segmented is also presented. The system Was tested on 20 patients (988 total images) from the Oncology Department of La Chaux-de-Fonds Hospital and the results show that the method is reliable as a computer-aided diagnostic tool for clinical purpose in all oncology department. Copyright (C) 2002 John Wiley Sons, Ltd.