Svensson O, Sörnmo L, Pahlm O
Department of Signal Processing, Lund University, Sweden.
Med Biol Eng Comput. 1994 Jul;32(4 Suppl):S9-15. doi: 10.1007/BF02523321.
The effects of amplitude resolution in the signal-averaged ECG are studied in relation to the analysis of cardiac late potentials. The statistical properties of ECG signals from 22 patients after myocardial infarction were investigated in terms of amplitude distribution and noise level for the material. It was found that unbiased averaging could be achieved using a resolution greater than 10 microV. The noise levels of the bandpass-filtered individual X, Y and Z leads (range of 1-10 microV) and the vector magnitude were also investigated. Marked intra-lead differences in noise level were found, indicating that analysis based on individual leads is preferable. The effects of quantisation noise on the vector magnitude were negligible for amplitude resolutions below 5 microV. However, the additional noise contribution at 5 microV could be compensated by a moderate increase in acquisition time. The sensitivity to noise was considered when determining the endpoint of the filtered QRS complex using the vector magnitude.
研究了信号平均心电图中幅度分辨率对心脏晚电位分析的影响。根据材料的幅度分布和噪声水平，对22例心肌梗死后患者的心电图信号的统计特性进行了研究。发现使用大于10微伏的分辨率可以实现无偏平均。还研究了带通滤波后的单个X、Y和Z导联（范围为1 - 10微伏）以及矢量大小的噪声水平。发现导联内噪声水平存在显著差异，表明基于单个导联的分析更可取。对于低于5微伏的幅度分辨率，量化噪声对矢量大小的影响可以忽略不计。然而，5微伏时的额外噪声贡献可以通过适度增加采集时间来补偿。在使用矢量大小确定滤波后的QRS复合波终点时考虑了对噪声的敏感性。
