Noninvasive Blood Pressure Waveform Measurement Method Based on CNN-LSTM

• Published in: Biometric Recognition pp. 664 - 675
• Main Authors: Wang, Zheng, Lin, Dongmei, Zhang, Aihua, Ma, Yurun, Chen, Xiaolei
• Format: Book Chapter
• Language: English
• Published: Cham Springer Nature Switzerland
• Series: Lecture Notes in Computer Science
• Subjects: Blood pressure wave; Convolutional neural networks; Long Short Term Memory Networks; Pulse wave
• ISBN: 3031202325; 9783031202322
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-031-20233-9_67

Cardiovascular disease is a serious threat to human health. Continuous blood pressure (BP) waveform measurement is of great significance for the prevention of cardiovascular disease. Therefore, convenient and accurate BP measurement is a vital problem. This paper intends to visualize the weak pulsation of human radial artery pulse, combining the advantages of convolutional neural networks (CNN) and Long Short-Term Memory Networks (LSTM). A CNN-LSTM blood pressure measurement method based on pulse wave and blood pressure wave data is proposed. Experiments show that the six blood pressure correlation coefficients of the non-invasive blood pressure measurement method based on CNN-LSTM all exceed 0.99, and the average MSE loss is only around 0.004. This network is superior to CNN and LSTM networks and is expected to be used for blood pressure wave measurement in humans in the future.