Newswise — Lung cancers continue to kill more people than any other type of cancer, with current survival rated at only 12-15%. Early detection and treatment greatly improves prognosis. For example, detecting and treating tumours less than 3cm in size increases survival to 75%. Lung cancer can be detected at a sufficiently early stage using spiral CT. A single thoracic CT scan may comprise some 500 separate "slice" images. The improved resolution of modern CT scanners enables earlier detection of abnormalities, but the interpretation times become impractical and errors increase. This problem could theoretically be solved by using computer-aided detection (CAD) to help radiologists manage the demand for more numerous and improved image analysis. However, the computer-based systems that have been proposed to date have demonstrated low sensitivity and a large number of false positive results.
THE OXFORD INVENTIONResearchers at the University of Oxford Engineering Science department have developed a new method incorporating a process called Visual Moving Features (VMF) for detecting pulmonary nodules. VMF detects a predefined structure in a subject by image analysis based on a representation of the structure constructed across a number of parallel image planes. The method has been tested on results from 12 clinical cases involving a total of 3875 sectional images and 106 nodules. Every nodule recognised by a skilled radiologist was detected by the Oxford method, with far fewer false-positive detections than competitive methods.
RSS