Georgiadis D, Kaps M, Siebler M, Hill M, König M, Berg J, Kahl M, Zunker P, Diehl B, Ringelstein E B
University Department of Medicine and Therapeutics, Western Infirmary, Glasgow, UK.
Stroke. 1995 Mar;26(3):439-43.
The purpose of this study was the evaluation of intraobserver, interobserver, and intrasubject variability in the Doppler detection of microembolic signals in patients with mechanical prosthetic valves. Simultaneously, the feasibility of automated embolus detection by means of a neuronal network was investigated.
From 25 patients with mechanical prosthetic heart valves, single transcranial Doppler monitoring sessions of 30 minutes' duration were recorded on videotape, randomized, and subsequently analyzed by eight independent trained observers from three centers. Three observers evaluated these tapes on three separate occasions, blinded to their previous results. An additional 48 patients with prosthetic heart valves were repetitively monitored with transcranial Doppler ultrasonography for 30 minutes three times within 1 year to examine the long-term variability in the occurrence of microembolic signals. Finally, in an effort to assess the short-term intrasubject variability, 20 patients were examined for 90 minutes, and the results of the three 30-minute periods were compared. The interobserver, intraobserver, and intrasubject (both short- and long-term) variability was evaluated.
No significant differences in microembolic signal counts were found among the different observers, between the human observers and the neuronal network, or among the three separate evaluations of stored data by the same observer. The same was true for repeat examinations of the same patient (P > .05, Student's two-paired t test and Friedman's test).
The detection of microembolic signals in patients with prosthetic cardiac valves is a reproducible technique. The reliable performance of the neuronal network argues for a broader use of this device. The intrasubject stability of the microembolic rate over 1 year supports the concept that the underlying emboligenic process is associated with intrinsic mechanical properties of the valve implant and not due to a thromboembolic process in the heart.
本研究旨在评估机械人工心脏瓣膜患者微栓子信号多普勒检测中的观察者内、观察者间及个体内变异性。同时，研究了利用神经网络进行自动栓子检测的可行性。
对25例机械人工心脏瓣膜患者进行持续30分钟的经颅多普勒单次监测，并录制在录像带上，随机分组，随后由来自三个中心的八名独立训练有素的观察者进行分析。三名观察者在三个不同时间对这些录像带进行评估，且对他们之前的结果不知情。另外48例人工心脏瓣膜患者在1年内重复进行三次经颅多普勒超声检查，每次30分钟，以检查微栓子信号出现的长期变异性。最后，为评估个体内短期变异性，对20例患者进行90分钟检查，并比较三个30分钟时间段的结果。评估了观察者间、观察者内及个体内（短期和长期）变异性。
不同观察者之间、人工观察者与神经网络之间以及同一观察者对存储数据的三次单独评估之间，微栓子信号计数均无显著差异。同一患者的重复检查结果也是如此（P>.05，配对t检验和弗里德曼检验）。
人工心脏瓣膜患者微栓子信号的检测是一种可重复的技术。神经网络的可靠性能支持更广泛地使用该设备。微栓子发生率在1年内的个体内稳定性支持了以下观点，即潜在的栓子形成过程与瓣膜植入物的内在机械特性有关，而非心脏内的血栓栓塞过程所致。
