基于高清晰度CT图像的孤立性肺结节计算机辅助诊断系统  

Computer- aided diagnosis system for solitary pulmonary nodules based on high definition CT images

作　　者：张泽文[1] 张才擎[2] 张成琪[3] 

机构地区：[1]山东中医药大学第一临床医学院,山东济南250014 [2]山东省千佛山医院呼吸科,山东济南250014 [3]山东省千佛山医院影像科,山东济南250014

出　　处：《中国医学物理学杂志》2015年第5期669-673,共5页Chinese Journal of Medical Physics

基　　金：国家自然科学基金(61272245)

摘　　要：本文介绍了一款基于高清晰度CT(High Definition CT,HDCT)图像的孤立性肺结节(Solitary Pulmonary Nodule,SPN)计算机辅助诊断系统(Computer-Aided Diagnosis System,CADS),该系统对SPN良恶性诊断具有一定的临床使用价值。收集经临床病理证实SPN患者110例,包括恶性肿瘤、良性肿瘤、结核和炎性假瘤,随机抽取60例作为实验集,50例作为验证集。实验集HDCT图像经图像预处理、感兴趣区域(Region of Interest,ROI)基于标记的分水岭算法分割和ROI纹理特征参数提取,对获得的5项纹理特征参数做统计处理,将统计结果应用于系统以对SPN做良恶性分析并给出提示信息。验证集HDCT图像输入系统,由运行结果可知本系统对SPN诊断能获得较为满意的效果。本系统运用基于标记的分水岭算法,对于胸壁或纵隔粘连的结节及磨玻璃病变等均可以较好地将其分割提取出来,同时自动分析后给医师提示SPN良恶性,辅助临床医师对SPN诊断,提高恶性SPN的检出率,使诊断更加客观、科学。A computer-aided diagnostic system （CADS） for solitary pulmonary nodule （SPN） based on high definition CT （HDCT） images is introduced in this paper. And the CADS has certain clinical values in the diagnosis of malignant and benign SPN. Confirrned by clinical pathology, 110 cases of SPN were selected, including malignant tumors, benign tumors, tuberculosis and inflammatory pseudotumors. Among these cases, 60 cases were randomly selected as the experimental set, while the other 50 cases were in the validation set. HDCT images of the experimental set were preprocessed, and the region of interest （ROI） was segmented by marker-based watershed algorithm, and ROI texture feature parameters were extracted to statistically analyze the acquired five texture feature parameters. The statistical results were applied in the CADS to diagnose malignant and benign SPN and provide related messages. HDCT images of the validation set were inputted in the CADS. The system operation performance showed the CADS could achieve a satisfactory effect in the diagnosis of SPN. The CADS, applying marker-based watershed algorithm, can effectively segment and extract the nodules adjoined with the chest wall or mediastinum and the ground glass lesions. Meanwhile, the automatic analysis of CADS can provide some diagnostic messages of SPN to assist clinicians in the clinical diagnosis, improving the detection rate of malignant SPN and making the diagnosis more objective and scientific.

关 键 词：计算机辅助诊断 孤立性肺结节 高清晰度CT 分水岭算法 纹理特征提取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] R318[自动化与计算机技术—计算机科学与技术]