计算机辅助检测系统在数字化X线胸片肺结节筛查的临床应用

发布时间：2018-08-15 15:12

【摘要】：【目的】探讨计算机辅助检测(CAD)系统在数字化X线胸片肺结节筛查中对不同年资放射医师的结节检测敏感性及观察者间变异的影响。【方法】由两名高年资医生筛选经CT证实的100例正常及200例含结节的后前位数字化胸片,并共同制定结节组结节的大小、位置的金标准。由另外的四名高年资及四名低年资医生独立对两组的胸片分别不用及应用CAD辅助阅片并保存结果。使用配对t检验评估两组医生在应用CAD辅助阅片之前与之后的平均结节检测敏感性的差异;使用变异系数(CV)对两组医生应用CAD辅助阅片之前与之后的观察者间变异进行评估。【结果】两组医生应用CAD辅助阅片之前及之后平均结节检测敏感性分别为(41.1±2.0)%、(28.0±2.0)%及(45.0±1.8)%、(39.2±0.9)%,两组医师在应用CAD辅助阅片之前与之后的平均检测敏感性有统计学差异;应用CAD辅助阅片之前及之后所有医师的平均检测敏感性变异系数分别为20.9%及8.1%。【结论】应用CAD能帮助放射科医师提高结节检出率,并能缩小不同年资医师间结节检测能力的差异。
[Abstract]:[objective] to investigate the effect of computer-aided detection (CAD) system on the sensitivity and interobserver variation of lung nodules in radiologists of different years. [methods] two senior doctors were employed in this study. To screen 100 cases of normal and 200 cases of anterior digital chest radiographs with nodule confirmed by CT. A gold standard for the size and location of nodules was established. The other four senior and four junior doctors independently examined the chest radiographs in both groups without and using CAD assisted radiographs to save the results. A paired t test was used to evaluate the difference in the sensitivity of the mean nodule detection between the two groups before and after the use of CAD assisted radiography. The variability coefficient (CV) was used to evaluate the variation between the two groups before and after the CAD assisted reading. [results] the sensitivity of the mean nodule detection before and after the application of CAD was (41.1 卤2.0)%, (28.0 卤2.0)% and (28.0 卤2.0)%, respectively. (45.0 卤1.8) and (39.2 卤0.9). The sensitivity of the two groups before and after the application of CAD was significantly different. The average sensitivity coefficient of variation before and after the application of CAD was 20.9% and 8.1% respectively. [conclusion] the application of CAD can help radiologists to improve the detection rate of nodules. And it can reduce the difference of nodule detection ability among different years doctors.
【作者单位】：中山大学附属第一医院放射科;
【基金】：国家自然科学基金面上项目(8161792)
【分类号】：R563;R816.41

【参考文献】