Enhancing component on CT to predict malignancy in cystic renal masses and interobserver agreement of different CT features

Enhancing component on CT to predict malignancy in cystic renal masses and interobserver agreement of different CT features
复制标题

DOI:
10.2214/ajr.04.0372
复制
发表时间:
2006-03-01
影响因子:
5
通讯作者:
Tomlinson, G
Tomlinson, G
中科院分区:
医学2区
文献类型:
--
作者:
Benjaminov, O;Atri, M;Tomlinson, G

文献摘要

被引文献

相似文献

OBJECTIVE.本研究的目的是确定复杂囊性肾肿块的CT特征,这些特征是最能预测恶性肿瘤的,并评估观察者间在解释这些特征时的差异。两名对病理结果不知情的放射科医生回顾性分析了30例接受手术的患者(19例男性和11例女性;年龄范围,28-76岁;平均年龄,59 +/- 13岁)的36个连续囊性肾脏肿块的CT扫描。研究人群仅包括大体病理学和成像上具有囊性成分的肿块。所有患者均行CT增强扫描。评审员记录了每个囊性肿块的CT特征,包括增强成分的存在。计算每个CT特征预测恶性肿瘤的准确度值和比值比。加权kappa用于测量观察者间的一致性。囊性肾癌21例,良性囊性病变11例。所有囊性肾细胞癌均显示增强的间隔或结节成分。两个评价者预测恶性肿瘤间隔增强的平均敏感性和特异性为83%(95%置信区间[CI],65-93%)和82%(95% CI,56-94%);结节性增强,67%(95% CI,49-81%)和96%(95% CI,75-99%);间隔或结节增强分别为100%(95% CI,86-100%)和86%(95% CI,67-95%)。观察者间对间隔强化和结节强化的一致性分别为良好(K = 0.67)和中度(K = 0.57)。结节状或间隔强化对预测恶性肿瘤的敏感性最高,观察者之间的一致性为中等至良好。
OBJECTIVE. The objective of our study was to determine the CT features of complex cystic renal masses that are the most predictive of malignancy and to assess interobserver variability when interpreting these features.MATERIALS AND METHODS. Two radiologists blinded to pathology results retrospectively reviewed CT scans of 36 consecutive cystic renal masses in 30 patients (19 men and 11 women; age range, 28-76 years; mean age, 59 +/- 13 years) who had undergone surgery. The study population included only masses with a cystic component on gross pathology and imaging. All patients underwent contrast-enhanced CT. The reviewers recorded the CT features of each cystic mass, including the presence of enhancing components. Accuracy values and odds ratio to predict malignancy were calculated for each CT feature. Weighted kappa was used to measure interobserver agreement.RESULTS. There were 21 cystic renal cell cancers and 11 benign cystic lesions. All cystic renal cell carcinomas showed an enhancing septal or nodular component. The mean sensitivity and specificity of the two reviewers in predicting malignancy for the presence of septal enhancement were 83% (95% confidence interval [CI], 65-93%) and 82% (95% CI, 56-94%); for nodular enhancement, 67% (95% CI, 49-81%) and 96% (95% CI, 75-99%); and for either septal or nodular enhancement, 100% (95% CI, 86-100%) and 86% (95% CI, 67-95%), respectively. The interobserver agreements for septal and nodular enhancement were good (K = 0.67) and moderate (K = 0.57), respectively.CONCLUSION. The presence of either nodular or septal enhancement shows the highest sensitivity for predicting malignancy with moderate to good interobserver agreement.