Automatic lung cancer detection using color histogram calculation

Abstract

Lung cancer is a disease that caused by uncontrolled cell growth in lung. Lung cancer is still the first worldwide killer. CT Scan Thorax is a method for early detection of lung cancer patients. However, cancer detection in lung CT-Scan image still done manually. In this paper, the segmentation of lung image is proposed. Cancer segmentation will process the lung CT-Scan as an image input with watershed process to cut off cavity area. The result will be processed by color histogram calculation to obtain mean and standard deviation value. This value is useful for evaluate non-cancer area and produce cancer image. Segmentation process will be followed by measurement of cancer and cavity area. The overall output is percentage between the large of cancer area and cavity area. The experiment represented that this method is able to detect lung cancer automatically. The performance segmentation for assessment errors obtained an average cavity area segmentation 12.75% and cancer area segmentation 31.74%.