基于多重分形理论的心电诊断系统设计与应用

Design and application of electrocardiograph diagnosis system based on multifractal theory

设计实现了一种利用差分阈值法进行心电数据自动分段算法,该算法可识别连续心电数据的各个心电周期;基于多重分形理论,获取心电数据的多重分形半谱特征和广义Hurst指数特征,用于神经网络模型的训练,以实现心电数据分类,其分类的准确率为97%。实现了心电诊断系统并用于实际应用,该系统能自动识别包含多个周期的心电序列,忽略该心电序列中首尾不完整心电周期数据,并可对心电数据各个周期进行分类标注。

An automatic segmentation algorithm for ECG data using differential threshold method was designed, which could identify the various ECG cycles of continuous ECG data. And it could obtain the multifractal features of multiple fractal and generalize hurst index feature of ECG data, these features were used to train artificial neural network in order to classify ECG data, the accuracy of the classifier could reach 97%. An ECG diagnosis system was implemented, which can automatically identify ECG sequences that contain multiple ECG cycles, and can automatically ignore the incomplete ECG cycle data, and could annotate every cycle of ECG data.