Automated Lung Nodule Segmentation Using an Active Contour Model Based on PET/CT Images

Most of the traditional segmentation methods segment not only malignant nodules but also some blood vessels and benign nodules, which increases the workload of nodule recognition and complicates the assessment of benign versus malignant nodules. Here, we propose an automated method of lung-nodule segmentation using an active contour model based on positron emission tomography/computed tomography (PET/CT) images. The method is divided into the following three steps. (1) Threshold segmentation and regional growth segmentation are combined to achieve lung parenchyma segmentation on CT. (2) Template matching is used to segment the lung nodule on PET. (3) An active contour model is used to accurately segment the lung nodule on CT. The experimental results showed that this method can effectively segment lung nodules on PET/CT and that it achieves higher segmentation accuracy than other commonly used methods.