Abstract: Multidimensional Independent Component Analysis（MICA） is suitable for multi-dimensional multi-
component signal analysis and component extraction, can be used in the actual extraction of fetal ECG. To illustrate the reliability of the results of MICA separation algorithm, it is verified by using synthetic model generated simulations Holter ECG data. With vkMICA, cfMICA, SJADE, MSOBI four kinds of MICA algorithm, pregnant women actually collected ECG data sets are processed. The results show the four MICA algorithm can extract the fetal ECG signal more pure, and can better reflect the actual characteristics of fetal ECG.

Key words: multidimensional independent component analysis, technology of joint block diagonalization, fetal Electrocardiogram（ECG） signal, algorithm

摘要： 多维独立分析（Multidimensional Independent Component Analysis，MICA）方法适用于实现多分量信号分析和分量提取，可实际应用于对胎儿心电信号的提取。为了说明MICA算法分离结果的可靠性，利用动态心电合成模型生成仿真的心电信号数据进行了验证，用vkMICA、cfMICA、SJADE、MSOBI四种MICA算法对实际采集到孕妇心电数据集进行了处理。实验结果显示，四种MICA算法可提取出较纯净的胎儿心电信号，并能较好地反映出胎儿心电信号的实际特征。

关键词: 多维独立成分分析, 联合块对角化技术, 胎儿心电信号, 算法