Quantitative analysis of cancer candidate regions in chest X-ray CT images

Abstract

In this paper, we improved our reduction method of false positive rate in computer aided diagnosis of Lung cancer Screening system by CT(LSCT). We had changed a target size of detective cancer’s diameter from more than 10mm to more than 5mm. By this change, many false positives whose diameters were under 10mm were newly detected. In order to reduce them, we proposed a method of feature analysis of cancer candidate regions using new feature parameters. This reduction method was applied to LSCT images of 23 samples. Consequently, the average number of false positives per sample was 6.5 without any additional false positives. These results showed the effectiveness of the proposed method.