Abstract

Info

Title:: Acquisition, feature analysis and application of ballistocardiogram signals

Author(s):: LI Qian; WANG Fei; LIU Qian; HUANG Qijun; CHANG Sheng; HE Jin; WANG Hao; School of Physics and Technology, Wuhan University, Wuhan 430072, China

Keywords:: Keywords: ballistocardiogram signal; electrocardiogram signal; cardiac cycle; correlation analysis

Abstract:: Abstract: Objective To explore the intrinsic relationship between ballistocardiogram (BCG) signals and electrocardiogram (ECG) signals for improving the value of BCG signals in clinical application. Methods BCG signals and ECG signals were simultaneously acquired, and the J wave of BCG signals was extracted by wavelet preprocessing and adaptive template matching for obtaining the cardiac cycle interval duration of BCG signals which were then compared with the cardiac cycle interval duration of the ECG signals acquired at the same time. One-dimensional convolutional neural network was used to establish a reconstructed mapping of BCG signals to ECG signals, thereby obtaining the morphology of the original ECG signals. Results The correlation between the cardiac cycle interval duration of BCG signals and that of ECG signals acquired at the same time was 98.16%, and the morphology correlation was 95.44%. Several features such as RR interval, QRS band width and ST band width of the original ECG signals could be obtained from BCG signals. Conclusion Effective ECG signal characteristics can be found in BCG signals, which can improve the clinical significance of BCG signals.