摘要
目的:探讨基于 MRI 征象及临床因素的评分系统预测胎盘植入性疾病(PAS)的应用价值.方法:回顾性分析 132 例孕妇的临床及MRI资料,其中非PAS组 80 例,PAS组 52 例(其中黏连型胎盘植入(PA)亚组 25 例、植入型胎盘植入(PI)/穿透型胎盘植入(PP)亚组 27 例).采用 5 分评分法对MRI征象及临床因素进行评分.统计分析各参数的组间差异性以及与 PAS的相关性,采用 ROC曲线分析各个相关参数对PAS的诊断效能.利用二元 Logistics 回归分析构建 MRI+临床诊断模型,并运用ROC曲线分析该模型的诊断效能.结果:两位医师对 MRI 征象评分的一致性极好(ICC＞0.9,P＜0.05).孕次评分、产次评分、子宫手术史评分、胎盘位置评分、11 个征象总分及 7 个共识征象总分在非PAS组与PAS组间差异具有统计学意义并均与PAS呈正相关性(P＜0.05),其中 7 个共识征象总分相关性最高(r＝0.635).孕次评分、产次评分、子宫手术史评分、胎盘位置评分、11 个征象总分及 7 个共识征象总分预测PAS的ROC曲线下面积(AUC)分别为 0.625、0.684、0.778、0.741、0.868、0.875(P＜0.05).ROC曲线分析结果显示,MRI+临床诊断模型预测PAS组与非PAS组、非PAS组与PA亚组、PA亚组与PI/PP亚组的 AUC 分别为 0.892、0.795、0.871(P＜0.05).结论:基于 MRI 及临床相关因素的评分系统可以较准确地预测PAS及其植入深度,具有较高的应用价值.
Abstract
Objective:This study aims to explore the application value of a scoring system based on MRI features and clinical factors in predicting placenta accreta spectrum disorder(PAS).Methods:The MRI and clinical data of 132 pregnant women were retrospectively analyzed,including 80 cases in the non-PAS group and 52 cases in the PAS group,with 25 cases in the placenta accreta(PA)sub-group and 27 cases in the placenta increta(PI)/placenta percreta(PP)subgroup.MRI features and clinical factors were scored using a 5-point scoring method.The difference between groups of each pa-rameter and the correlation with PAS were statistically analyzed.ROC curve analysis was used to ana-lyze the diagnostic efficacy of each related parameter on PAS.The MRI+clinical diagnosis model was constructed by binary logistic regression analysis,and the diagnostic efficacy of the model was analyzed by ROC curve.Results:The scores of MRI features between the two doctors were highly consistent(ICC＞0.9,P＜0.05).The pregnancy scores,parity scores,uterine surgery history scores,placental po-sition scores,total scores of 11 features,and total scores of 7 consensus features were significantly dif-ferent between the non-PAS group and PAS group,and were all positively correlated with PAS(P＜0.05).Among them,the total score of 7 consensus features had the highest correlation(r＝0.635).The area under ROC curve(AUC)predicting PAS was 0.625,0.684,0.778,0.741,0.868,0.875,respectively(P＜0.05).Furthermore,the ROC curve analysis showed that the AUC of the MRI+clinical diagnosis model in predicting PAS group and non-PAS group,non-PAS group and PA subgroup,and PA sub-group and PI/PP subgroup were 0.892,0.795,and 0.871,respectively(P＜0.05).Conclusion:The sco-ring system based on MRI features and clinical factors can accurately predict PAS and its implantation depth,which has high application value.
NETL
NSTL
维普
万方数据