Skip to main content
QUICK REVIEW

[论文解读] Generalizing electrocardiogram delineation -- Training convolutional neural networks with synthetic data augmentation

Guillermo Jiménez-Pérez, Juan Acosta|arXiv (Cornell University)|Nov 25, 2021
ECG Monitoring and Analysis被引用 4
一句话总结

该论文提出了一种基于合成数据增强和新型损失函数的深度学习框架,用于心电图(ECG)波形界定,实现了在多种心电图数据库上99.38%的F1分数和亚20毫秒的界定误差。该方法结合了伪合成心电信号生成与形状正则化训练,提升了在心 beat 级别波形标志点检测中的泛化能力与鲁棒性。

ABSTRACT

Obtaining per-beat information is a key task in the analysis of cardiac electrocardiograms (ECG), as many downstream diagnosis tasks are dependent on ECG-based measurements. Those measurements, however, are costly to produce, especially in recordings that change throughout long periods of time. However, existing annotated databases for ECG delineation are small, being insufficient in size and in the array of pathological conditions they represent. This article delves has two main contributions. First, a pseudo-synthetic data generation algorithm was developed, based in probabilistically composing ECG traces given "pools" of fundamental segments, as cropped from the original databases, and a set of rules for their arrangement into coherent synthetic traces. The generation of conditions is controlled by imposing expert knowledge on the generated trace, which increases the input variability for training the model. Second, two novel segmentation-based loss functions have been developed, which attempt at enforcing the prediction of an exact number of independent structures and at producing closer segmentation boundaries by focusing on a reduced number of samples. The best performing model obtained an $F_1$-score of 99.38\\% and a delineation error of $2.19 \\pm 17.73$ ms and $4.45 \\pm 18.32$ ms for all wave's fiducials (onsets and offsets, respectively), as averaged across the P, QRS and T waves for three distinct freely available databases. The excellent results were obtained despite the heterogeneous characteristics of the tested databases, in terms of lead configurations (Holter, 12-lead), sampling frequencies ($250$, $500$ and $2,000$ Hz) and represented pathophysiologies (e.g., different types of arrhythmias, sinus rhythm with structural heart disease), hinting at its generalization capabilities, while outperforming current state-of-the-art delineation approaches.

研究动机与目标

  • 解决用于训练深度学习模型的心电图数据库标注数据稀缺且多样性有限的问题。
  • 提升心电图波形界定模型在不同导联配置、采样频率和病理状态下的泛化能力。
  • 通过基于基础心电图波段的伪合成数据生成方案,减少对大规模标注数据集的依赖。
  • 通过引入新型基于分割的损失函数,增强模型对边界预测准确性和固定结构数量的约束,从而提升鲁棒性。
  • 开发一种临床可用的端到端深度学习流水线,实现高精度与强泛化能力的心电图标志点自动检测。

提出的方法

  • 通过基于专家定义规则的概率性组合真实数据库中的基础心电图波段(P波、QRS波群、T波),构建了伪合成数据生成方案,实现连贯心电信号的组装。
  • 该合成数据生成器通过施加形态学修改和病理生理状态(如心律失常和传导异常)引入可控的变异。
  • 采用全卷积的U-Net架构,用于在多导联和不同采样率下预测P波、QRS波群和T波的标志点(起始点与终止点)。
  • 引入两种新型损失函数:一种强制模型对每类心脏结构仅预测一个实例;另一种则聚焦于关键样本子集,以提升边界定位的精确性。
  • 通过使用基于中位数的滑动窗口(256个样本)对输入信号进行归一化,提升对不同心律下振幅变化的鲁棒性。
  • 模型在三个开源数据库(QT、LU和浙江大学附属医院数据库)上进行训练与评估,涵盖12导联和Holter记录,采样频率为250、500和2000 Hz。

实验结果

研究问题

  • RQ1基于真实心电图波段的合成数据增强是否能提升深度学习模型在多样化心电图数据库中的波形界定泛化能力?
  • RQ2形状正则化损失函数在强制模型准确检测固定数量的心脏波形及实现精确边界定位方面效果如何?
  • RQ3在合成数据上训练的深度学习模型,能在具有不同导联配置和病理状态的真实世界心电图数据库上达到怎样的最先进性能?
  • RQ4将专家知识融入合成数据生成过程,是否能增强模型的鲁棒性并减少对特定波形形态的过拟合?
  • RQ5该模型在快速室性心动过速或细微波形等复杂病例中的表现如何?

主要发现

  • 该模型在三个公开心电图数据库的联合测试集上实现了99.38%的F1分数,表明标志点检测接近完美。
  • P波、QRS波群和T波的平均界定误差分别为2.19 ± 17.73 ms(起始点)和4.45 ± 18.32 ms(终止点),展现出极高的时间精度。
  • 模型在不同导联配置(12导联 vs. Holter)、采样频率(250、500、2000 Hz)以及病理生理状态(包括多种心律失常和结构性心脏病)的数据库中均表现出良好泛化能力。
  • 该模型优于现有最先进方法,尤其在减少误检方面表现突出,如波形误标、标志点错位或噪声激活。
  • 尽管模型整体鲁棒性较强,但对输入归一化方式仍表现出一定敏感性,使用256样本的中位数滑动窗口归一化后性能进一步提升。
  • 在极快的室性心动过速情况下,模型表现略有下降,这同样是专家心内科医生面临的挑战,表明该领域仍存在临床待突破的前沿。

更好的研究,从现在开始

从阅读论文到最终审阅,大幅缩短您的研究时间。

无需绑定信用卡

本解读由 AI 生成,并经人工编辑审核。