Abstract

Info

Title:: Deep learning-based method for finger-radial PPG signal transferring

Author(s):: ZHOU Weiding; CHEN Zhaoxue; School of Health Science and Engineering, University of Shanghai for Science and Technology, Shanghai 200093, China

Keywords:: Keywords: photoplethysmogram signal pulse wave pulse diagnosis deep learning

Abstract:: Abstract: In view of the fact that most pulse wave detection instruments currently exist on the market detect the photoplethysmography (PPG) signal from the fingertip, a convolutional neural network based finger-radial PPG signal transferring method is proposed for obtaining the corresponding radial PPG signal under the condition that only the finger PPG signal is available. The method is mainly realized by one-dimension convolutional neural network through end-to-end training, and the model is composed of 3 parts, namely encoder, decoder and skip connection. The features of the fingertip pulse wave signal are extracted through the encoder network, and then the feature maps are expanded through the decoder network while the feature maps are concatenated by skip connection. In the study, 60 finger and radial PPG signals are collected for experiments, and the proposed method is compared with transfer function model and windkessel model. The results show that the MAE and PRD of the radial PPG signal reconstructed by the proposed model reached 1.4%[±]0.3% and 3.6[%±]1.2%, respectively, indicating that the proposed method is superior to the other models. It is demonstrated that the proposed model can accurately transferring the finger PPG signal to the radial PPG signal.