Department of Gynecology and Obstetrics, Division of Prenatal Medicine, University Hospital of Schleswig-Holstein, Campus Luebeck, Luebeck, Germany.
J Ultrasound Med. 2020 Feb;39(2):301-309. doi: 10.1002/jum.15105. Epub 2019 Aug 14.
To investigate the interobserver and intraobserver variability and corresponding learning curve in a semiautomatic approach for a standardized assessment of the fetal heart (fetal intelligent navigation echocardiography [FINE]).
A total of 30 stored spatiotemporal image correlation volume data sets of second-trimester fetuses were evaluated by 3 physicians with different levels of expertise in fetal echocardiography by using the FINE approach. Data were analyzed regarding the examination time and proper reconstruction of the diagnostic cardiac planes. The completions and numbers of correct depictions of all diagnostic planes were evaluated by a blinded expert (time t0). To determine interobserver and intraobserver variability, the volumes were reassessed after a 4-week training interval (time t1).
All operators were able to perform the investigation on all 30 volumes. At t0, the interobserver variability between the beginner and both the advanced (P = .0013) and expert (P < .0001) examiners was high. Focusing on intraobserver variability at t1, the beginner showed a marked improvement (P = .0087), whereas in advanced and expert hands, no further improvement regarding proper achievement of all diagnostic planes could be noticed (P > .999; P = .8383). The beginner also showed improvement in the mean investigation time (t0, 82.8 seconds; t1, 73.4 seconds; P = .0895); nevertheless, the advanced and expert examiners were faster in completing the examination (t1, advanced, 20.9 seconds; expert, 28.3 seconds; each P < .0001).
Based on our results, the FINE technique is a reliable and easily learned method. The use of this semiautomatic work flow-based approach supports evaluation of the fetal heart in a standardized and time-saving manner. A semiautomatic evaluation of the fetal heart might be useful in facilitating the detection of fetal cardiac anomalies.
研究半自动胎儿心脏标准化评估方法(胎儿智能导航超声心动图[FINE])的观察者间和观察者内可变性及其相应的学习曲线。
共 30 例储存的时空关联容积数据由 3 名具有不同胎儿超声心动图专业知识水平的医师使用 FINE 方法进行评估。分析检查时间和诊断心脏平面的适当重建。由一位盲法专家(时间 t0)评估所有诊断平面的完成情况和正确描绘次数。为了确定观察者间和观察者内的可变性,在 4 周的培训间隔后重新评估体积(时间 t1)。
所有操作者均能对所有 30 个容积进行检查。在 t0,初学者与高级(P=.0013)和专家(P < .0001)检查者之间的观察者间变异性较高。关注 t1 时的观察者内变异性,初学者有明显改善(P=.0087),而高级和专家手中,所有诊断平面的适当完成率均无进一步改善(P > .999;P=.8383)。初学者的平均检查时间也有所改善(t0,82.8 秒;t1,73.4 秒;P=.0895);然而,高级和专家检查者完成检查的速度更快(t1,高级,20.9 秒;专家,28.3 秒;均 P < .0001)。
根据我们的结果,FINE 技术是一种可靠且易于学习的方法。使用这种半自动基于工作流程的方法可以支持以标准化和节省时间的方式评估胎儿心脏。半自动评估胎儿心脏可能有助于发现胎儿心脏畸形。