Auto-Segmentation of lung in CT image series based on level set method with prior knowledge
Rui Shi Liang, Xue Jiao Chen, Jian Xun Zhang · 2017
With development of medical image processing, more and more cancers are diagnosed and treated with assistance of computer, especially lung cancer. Extracting lung from CT image series is usually the first step. Hundreds of CT images slow down the processing. As to clinical applications, the accuracy and efficiency is crucial. We proposed and realized an auto-segmentation of lung in CT image series based on level set method with prior knowledge. Firstly, we auto-segmented one CT image with LBF method. Secondly, we extracted the lung contour from the segmentation with region grow and feature selection. In the end, we extracted the lung contour of the next CT image with DRLSE method. The previous lung contour was set to be the initial contour to ensure the accuracy and efficiency. The experiment results proved that our approach is more stable and faster. It spent only about 67% of the time that LBF and DRLSE did.