Higuchi Satoshi, Li Roland, Gerstenfeld Edward P, Liem L Bing, Im Sung Il, Kalantarian Shadi, Ansari Minhaj, Abreau Sean, Barrios Joshua, Scheinman Melvin M, Tison Geoffrey H
Section of Cardiac Electrophysiology, Division of Cardiology, University of California, San Francisco, San Francisco, California.
Division of Cardiology, Department of Medicine, University of California, San Francisco, San Francisco, California.
Heart Rhythm O2. 2023 Jul 13;4(8):491-499. doi: 10.1016/j.hroo.2023.07.004. eCollection 2023 Aug.
It remains difficult to definitively distinguish supraventricular tachycardia (SVT) mechanisms using a 12-lead electrocardiogram (ECG) alone. Machine learning may identify visually imperceptible changes on 12-lead ECGs and may improve ability to determine SVT mechanisms.
We sought to develop a convolutional neural network (CNN) that identifies the SVT mechanism according to the gold standard of SVT ablation and to compare CNN performance against experienced electrophysiologists among patients with atrioventricular nodal re-entrant tachycardia (AVNRT), atrioventricular reciprocating tachycardia (AVRT), and atrial tachycardia (AT).
All patients with 12-lead surface ECG during sinus rhythm and SVT and had successful SVT ablation from 2013 to 2020 were included. A CNN was trained using data from 1505 surface ECGs that were split into 1287 training and 218 test ECG datasets. We compared the CNN performance against independent adjudication by 2 experienced cardiac electrophysiologists on the test dataset.
Our dataset comprised 1505 ECGs (368 AVNRT, 304 AVRT, 95 AT, and 738 sinus rhythm) from 725 patients. The CNN areas under the receiver-operating characteristic curve for AVNRT, AVRT, and AT were 0.909, 0.867, and 0.817, respectively. When fixing the specificity of the CNN to the electrophysiologist adjudicators' specificity, the CNN identified all SVT classes with higher sensitivity: (1) AVNRT (91.7% vs 65.9%), (2) AVRT (78.4% vs 63.6%), and (3) AT (61.5% vs 50.0%).
A CNN can be trained to differentiate SVT mechanisms from surface 12-lead ECGs with high overall performance, achieving similar performance to experienced electrophysiologists at fixed specificities.
仅使用12导联心电图（ECG）来明确区分室上性心动过速（SVT）机制仍然很困难。机器学习可以识别12导联心电图上肉眼难以察觉的变化，并可能提高确定SVT机制的能力。
我们试图开发一种卷积神经网络（CNN），根据SVT消融的金标准来识别SVT机制，并在房室结折返性心动过速（AVNRT）、房室折返性心动过速（AVRT）和房性心动过速（AT）患者中，将CNN的性能与经验丰富的电生理学家的性能进行比较。
纳入2013年至2020年期间在窦性心律和SVT期间有12导联体表心电图且SVT消融成功的所有患者。使用来自1505份体表心电图的数据训练CNN，这些数据被分为1287份训练心电图数据集和218份测试心电图数据集。我们在测试数据集上比较了CNN的性能与2名经验丰富的心脏电生理学家的独立判定结果。
我们的数据集包括来自725名患者的1505份心电图（368份AVNRT、304份AVRT、95份AT和738份窦性心律）。AVNRT、AVRT和AT在接收者操作特征曲线下的CNN面积分别为0.909、0.867和0.817。当将CNN的特异性固定为电生理学家判定者的特异性时，CNN以更高的敏感性识别所有SVT类别：（1）AVNRT（91.7%对65.9%），（2）AVRT（78.4%对63.6%），以及（3）AT（61.5%对50.0%）。
可以训练CNN从体表12导联心电图中区分SVT机制，总体性能较高，在固定特异性时达到与经验丰富的电生理学家相似的性能。
