Annu Int Conf IEEE Eng Med Biol Soc. 2022 Jul;2022:3582-3585. doi: 10.1109/EMBC48229.2022.9871252.
The electrocardiogram (ECG) is a common source of electrical artifact in electroencephalogram (EEG). Here, we present a novel method for removing ECG artifact that requires neither simultaneous ECG nor transformation of the EEG signals. The approach relies upon processing a subset of EEG channels that contain ECG artifact to identify the times of each R-wave of the ECG. Within selected brief epochs, data in each EEG channel is signal-averaged ± 60 ms around each R-wave to derive an ECG template specific to each channel. This template is subtracted from each EEG channel which are aligned with the R-waves. The methodology was developed using two cohorts of infants: one with 128-lead EEG including an ECG reference and another with 32-lead EEG without ECG reference. The results for the first cohort validated the methodology the ECG reference and the second demonstrated its feasibility when ECG was not recorded. This method does not require independent, simultaneous recording of ECG, nor does it involve creation of an artifact template based on a mixture of EEG channel data as required by other methods such as Independent Component Analysis (ICA). Spectral analysis confirms that the method compares favorably to results using simultaneous recordings of ECG. The method removes ECG artifact on an epoch by epoch level and does not require stationarity of the artifact. Clinical Relevance - This approach facilitates the removal of ECG noise in frequency bands known to play a central role in brain mechanisms underlying cognitive processes.
心电图（ECG）是脑电图（EEG）中常见的电伪迹来源。在这里，我们提出了一种新的去除 ECG 伪迹的方法，该方法既不需要同时记录 ECG，也不需要对 EEG 信号进行转换。该方法依赖于处理包含 ECG 伪迹的 EEG 通道子集，以识别 ECG 中每个 R 波的时间。在选定的短暂时段内，每个 EEG 通道中的数据在每个 R 波周围 ± 60 ms 处进行信号平均，以得出每个通道特有的 ECG 模板。将该模板从与 R 波对齐的每个 EEG 通道中减去。该方法是使用两个婴儿队列开发的：一个具有包含 ECG 参考的 128 导联 EEG，另一个具有没有 ECG 参考的 32 导联 EEG。第一个队列的结果验证了该方法和 ECG 参考，第二个队列证明了在没有记录 ECG 时该方法的可行性。该方法不需要独立、同时记录 ECG，也不需要像独立成分分析（ICA）等其他方法那样基于 EEG 通道数据的混合来创建伪迹模板。频谱分析证实，该方法与使用同时记录的 ECG 的结果相比具有优势。该方法在逐个时段去除 ECG 伪迹，并且不需要伪迹的稳定性。临床相关性 - 这种方法有助于去除已知在认知过程的大脑机制中起核心作用的频带中的 ECG 噪声。
