Utility of conventional and diffusion-weighted MRI features in distinguishing benign from malignant endometrial lesions.

European Journal of Radiology(2014)

引用 18|浏览2
暂无评分
摘要
To evaluate the utility of conventional MRI and diffusion-weighted imaging (DWI) in differentiating benign from malignant endometrial lesions.52 patients with an abnormal endometrium on MRI and subsequent pathologic evaluation (35 benign, 17 malignant) were included. Two radiologists (R1, R2) independently evaluated endometrial abnormalities for characteristics on conventional MRI and DWI. Findings were assessed using unpaired t-tests, Fisher's exact test, and multi-variate logistic regression.Findings with significantly higher frequency in malignant abnormalities were: presence of irregularly marginated endometrial lesion (R1: 71% vs. 34%, R2: 94% vs. 26%), irregular endo-myometrial interface on T2WI (R1: 77% vs. 26%, R2: 94% vs. 29%), irregular endo-myometrial interface on post-contrast T1WI (R1: 82% vs. 23%, R2: 88% vs. 20%), increased signal on high b-value DWI (R1: 82% vs. 20%, R2: 94% vs. 20%), decreased ADC (R1: 88 vs. 40%, R2: 94% vs. 20%) (all p<0.001, both readers). Endometrial thickness, presence of any focal endometrial lesion regardless of contour, diameter of endometrial lesion, endometrial heterogeneity on T2WI, decreased T2 signal, and increased endometrial enhancement, failed to show significant differences between groups (all p≥0.159, both readers). At multivariate analysis, for R1, irregular endo-myometrial interface on post-contrast T1WI and increased DWI signal were significant independent predictors of malignancy (AUC=0.89); for R2, only increased DWI signal was a significant independent predictor of malignancy (AUC=0.87).Abnormal signal on DWI and irregularity of either the endo-myometrial interface or focal endometrial lesion were the most helpful MRI features in differentiating benign from malignant endometrial abnormalities.
更多
查看译文
关键词
Endometrial abnormality,Diffusion-weighted,MRI
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要