基于ROC理论评价单色显示器对肺结节检出效能的影响

Evaluate the Influence of Monochrome Display on the Detection of Pulmonary Nodules Based on ROC Theory

目的评价不同分辨率的医用单色液晶显示器对胸部DR影像图上肺结节检出效能的影响。方法从图像存储与传输系统(PACS)中在线选取胸部直接数字化X线摄影(DR)影像图93幅,其中确诊图38幅、疑诊图32幅、正常图23幅(均由CT证实),将阳性病例按结节直径大小分为A、B两组,请3名高年资医师在3种不同分辨率的显示器上集中进行3次独立读图,对结节有无的评判采用5等分法:肯定有、可能有、不确定、可能无、肯定无,每名医师针对特定显示器上的每幅图像给出自己的信任等级。采用SPSS13.0绘制受试者操作特性(ROC)曲线,并对曲线下的面积(AUC)做Z检验。结果使用2MP、3MP、5MP显示器识读A组阳性病例的影像资料时ROC曲线AUC分别为0.643、0.686、0.739,识读B组阳性病例的影像资料时分别为0.813、0.832、0.846,Z检验表明不同放射系统诊断效能比较差异无统计学意义(P>0.05)。结论在不限制图像后处理工具的情况下,使用不同分辨率的医用单色液晶显示器识读不同尺寸肺结节时诊断效能相当。

Objective Evaluate the influence of medical monochrome liquid crystal display（LCD）with different resolutions on the detection of pulmonary nodules. Methods Ninety-three direct digital radiography（DR）chest images were selected on-line from picture archiving and communication system （PACS）,including 38 positive,32 suspected and 23 normal cases. All of the images were confirmed by CT. These positive images were divided into group A and B according to the diameter of nodules. Three experienced radiologists blindly read the 93 images using three types of displays with different resolutions. Regarding the presence of nodule,five-point confidence system was used：definitely absent,probably absent,indefinite,probably present and definitely present. All observers marked their confidence levels of each image according to the presence of pulmonary nodule on different displays. Software SPSS 13.0 was employed for statistical analysis. Results In the group-A cases,the areas under ROC curves were 0.643,0.686 and 0.739,and in the group-B cases,those were 0.813,0.832 and 0.846 respectively. There was no difference in the detection efficacy among different radiodiagnostic systems. Conclusion Regardless of post-processing tools,it was equivalent to detect pulmonary nodules with different diameters of Group A or B among medical monochrome LCDs with different resolutions.