Automatic Detection and Segmentation of Lung Nodule on CT Images

Abstract

Lung nodule detection and segmentation is important for clinical diagnosis. This paper proposes a lung nodule detection and segmentation method based on a fully convolutional network (FCN), the level set method and other image processing techniques. Firstly, lung CT images are put into the FCN for lung segmentation. Secondly, lung nodules are detected inside the lung area using the threshold method and other image processing techniques. Finally, the detected lung nodules and their spiculation are segmented by the level set method and threshold method based on the coordinate system transformation. Experimental result shows that the proposed method can effectively detect and segment lung nodules with the detection accuracy of 100% and the dice overlap index of segmentation of 0.9. Therefore, this method can provide helpful references for clinical diagnosis on lung cancers.