论文部分内容阅读
为了提高肺结节计算机辅助检测、辅助诊断的准确性,研究精确分割肺实质方法。首先,采用Otsu法实现阈值法粗分割;其次,设计三维区域增长方法有效去除粘连在肺实质上的主气管和支气管树;再次,采用灰度积分投影法以及行扫描方法对左右肺粘连的情况进行判别和处理,实现左右肺分离;最后,用滚球法修复肺区边界,实现肺实质分割。该方法有效解决了肺实质分割中粘连肺实质的气管、支气管树去除问题,高密度的肺部组织易丢失、粘连肺壁的肺结节容易被遗漏等问题。通过对多组LIDC-IDRI肺部图像数据集的实验表明,该方法能够精确高效地实现对于肺实质分割。
In order to improve the computer-aided detection of pulmonary nodules, the accuracy of diagnostic assistance, the study of accurate segmentation of pulmonary parenchyma method. First, Otsu method is used to implement threshold segmentation. Second, the three-dimensional regional growth method is designed to effectively remove the main trachea and bronchial tree attached to the lung. Thirdly, the gray-scale integral projection method and the line scan method are used to detect the left and right lung adhesion Discrimination and treatment, to achieve left and right lung separation; Finally, the use of ball method to repair the boundary of lung area, to achieve real lung segmentation. The method effectively solves the problems of removal of tracheal and bronchial tree in parenchymatous pulmonary parenchyma, high loss of lung tissue in high density and easily missed pulmonary nodules in the wall of adhesion. Experiments on multiple sets of LIDC-IDRI lung image datasets show that the proposed method can accurately and efficiently slice the lungs.