基于CT图像的肺实质细分割综合方法

肺实质的分割是肺结节检测和诊断的基础，是肺部疾病计算机辅助诊断的关键步骤之一。针对肺部计算机断层扫描（computer tomography,CT)图像，采用最佳阈值法进行初步分割，去除背景，用结合上下文分析的区域生长法去掉气管、支气管，对左右肺连结的情况进行像素分析，分开左右肺，对提取出来的肺区用滚球法进行修复，得到肺实质图像。去除气管和分割左右肺的算法是针对肺部CT图像的特征提出的，具有简单高效的特点。实验表明，该综合方法的准确性和可靠性较高，有较好的应用前景。

A method of careful lung segmentation based on CT images

The segmentation of lung parenchyma is the foundation of lung nodule detection and diagnosis. It is also one of the key stages in computer aided diagnosis systems to detect lung diseases. Therefore, this article made rough segmentation to get rid of the background with optimal threshold algorithm, and used the region growth algorithm and contextual analysis to reject the trachea and main bronchi. Then, pixel analysis were used to make the left and right lung segmentation. At last, the lung fields were gained and repaired with falling-ball algorithm. The algorithm used to reject the trachea and segment the left and right lung was especially designed according to the feature of the lung CT images, which were simple and efficient. The experiment shows that this method has good accuracy, robustness and future applications.