Abstract: In this paper，a novel method for extracting fetal electrocardiogram（FECG） from abdominal composite signal of a pregnant is proposed.The maternal component in the abdominal electrocardiogram（ECG） signal is a nonlinearly transformed version of the maternal ECG（MECG） and this nonlinearly relation is identified by the General Regression Neural Network （GRNN）.The FECG is extracted by subtracting the nonlinearly transformed version of MECG from the abdominal ECG signal.The baseline shift and noise in FECG are suppressing by wavelet packet denoise technique.Visual test results obtain from synthetic ECG signals and real ECG signals demonstrate the effectiveness of the proposed method in extracting the FECG even when it is totally embedded within the maternal QRS complex.

Key words: fetal electrocardiogram（FECG）, General Regression Neural Network（GRNN）, wavelet packet denoise

摘要： 针对胎儿心电难以提取问题，提出一种从母体腹壁混合信号中提取胎儿心电的方法。利用广义回归神经网络（GRNN）估计母体心电信号传导至腹壁的非线性变换，将非线性变换后的母体心电信号从腹壁混合信号中减去，再通过小波包去噪技术抑制胎儿心电的基线漂移和噪声，得到清晰的胎儿心电。应用合成心电信号和临床心电信号完成实验，在胎儿心电和母体心电QRS波完全重叠情况下，提取出清晰的胎儿心电。实验结果验证了方法的有效性。

关键词: 胎儿心电, 广义回归神经网络, 小波包去噪