论文部分内容阅读
目的对R.E.N.A.L肾肿瘤评分(RNS)系统评价一致性进行分析,探讨此系统的可重复性。方法收集64例经病理证实为肾细胞癌、行肾部分切除术患者的增强CT图像资料,由两位影像学医师分别采用RNS系统进行评分,记录指标R、E、N、A和L得分及其总分。以Kappa分析法评估同一观察者对RNS各指标及其总分的一致性,用Kendall tau-b分析法评估不同观察者间RNS各指标以及RNS总分的一致性、相关性。结果同一观察者R、E、N、A、L的符合率分别为96.9%、95.3%、84.4%、89.1%和81.3%,RNS总分符合率为86.0%;各指标的Kappa值分别为0.949、0.915、0.708、0.830和0.677,RNS总分的Kappa值为0.716。不同观察者间RNS各指标的符合率分别为95.3%、93.8%、86.0%、93.8%和73.4%,RNS总分的符合率为82.4%;各个指标的相关系数分别为0.924、0.917、0.798、0.929和0.480,RNS总分的相关系数为0.836,P均<0.001。结论 RNS系统评价在同一观察者及不同观察者间一致性好,具有良好的可重复性;其中指标L变异性较大,在5个指标中可靠性相对较低。
Objective To analyze the consistency of systematic reviews of R.E.N.A.L renal tumor score (RNS) and to explore the reproducibility of this system. Methods 64 cases of pathologically confirmed renal cell carcinoma underwent partial nephrectomy enhanced CT image data by the two imaging physicians were scored using RNS system, recording indicators R, E, N, A and L scores and Its total score. The Kappa analysis was used to evaluate the consistency of RNS indexes and their total scores by the same observer. The consistency and correlation of RNS indexes and RNS total scores among different observers were evaluated by Kendall tau-b analysis. Results The coincidence rates of R, E, N, A and L in the same observer were 96.9%, 95.3%, 84.4%, 89.1% and 81.3% respectively, and the coincidence rate of RNS was 86.0%. The Kappa values of each index were 0.949 , 0.915, 0.708, 0.830 and 0.677, respectively. The Kappa value of RNS score was 0.716. The coincidence rates of RNS indicators among different observers were 95.3%, 93.8%, 86.0%, 93.8% and 73.4% respectively, and the coincidence rate of RNS score was 82.4%. The correlation coefficients of each index were 0.924, 0.917 and 0.798, 0.929 and 0.480 respectively. The correlation coefficient of RNS score was 0.836, P <0.001. Conclusions The RNS systematic evaluation has good consistency and good repeatability among the same observer and different observers. Among them, the index L has a large variability, and the reliability of the five indexes is relatively low.