基于BP神经网络的心电信号智能诊断设计

Design of electrocardiosignal intelligent diagnosis based on BP neural network

结合差分阈值法、多重神经网络心电诊断方法提出一种新型的递进型双层BP神经网络心电信号智能诊断设计。采用差分阈值法检测QRS波特征点,通过心电采集电路板对该信号进行放大滤波处理后进行心电诊断。通过对比单一BP神经网络与递进型双层BP神经网络识别结果,得出递进型双层BP神经网络识别室性早搏、房性早搏、心房颤动时识别率更高(达到97%以上)的结论。测试结果表明,该设计有助于诊测和判断心血管疾病,提高了对治疗方法的有效性判断。

In comparison with the difference threshold method and multiple neural network ECG diagnosis method, a new electrocardiosignal intelligent diagnosis design of progressive dual-layer BP neural network is proposed. The difference threshold method is used to detect the feature points of QRS wave. The signal is amplified and filtered in ECG acquisition circuit board for ECG diagnosis. The recognition results of single BP neural network and progressive dual-layer BP neural network are compared to obtain the conclusion that the progressive dual-layer BP neural network can improve the recognition ratio of ventricular premature beat, atrial premature beat and atrial fibrillation by 97%. The test results show that the design can diagnose and judge the cardiovascular disease, and improve the judgment accuracy of treatment methods.