Xin Tian

Xin Tian

A Ph.D. candidate in ECE, University of Maryland, College Park.

Research

A Cross-Domain Joint Dictionary Learning Approach for ECG Reconstruction from PPG

From March 01, 2019

Abstract

An emerging research direction considers the inverse problem of inferring electrocardiogram (ECG) from photoplethysmogram (PPG) to bring about the synergy between the easy measurability of PPG and the rich clinical knowledge of ECG to facilitate preventive healthcare. Previous reconstruction using a universal basis has limited accuracy due to the lack of rich representative power. This paper proposes a cross-domain joint dictionary learning (XDJDL) framework to maximize the expressive power for the two cross-domain signals. Building on K-SVD technique, XDJDL optimizes simultaneously the PPG and ECG signal representations and the transform between them, enabling the joint learning of a pair of signal dictionaries with a transform to characterize the relation between their sparse codes. The proposed model is evaluated with 34,000+ ECG/PPG cycle pairs containing a variety of ECG morphologies and cardiovascular diseases. Experimental results validate the accuracy and the generality of the proposed algorithm, suggesting an encouraging potential for disease screening using PPG measurement based on the proactive learned PPG-ECG relationship.

A Pilot Study: ECG Reconstruction from PPG

From October 01, 2018

Abstract

In this work, we study the relation between electrocardiogram (ECG) and photoplethysmogram (PPG) signals and infer the waveform of ECG using PPG measurement. As the first work to address this inverse problem, we propose and train a linear transform which maps the discrete cosine transform (DCT) coefficients of the PPG cycle to those of the corresponding ECG cycle. The resulting DCT coefficients of the ECG cycle are then inversely transformed to obtain the predicted ECG waveform. The proposed method is evaluated with the different morphologies of the PPG and ECG signals on three benchmark datasets with a variety of combinations of age, weight, and health conditions using different training modes. Experimental results show that the proposed method can achieve a high prediction accuracy greater than 0.92 in averaged correlation for each dataset when the model is trained subject-wise. With a signal processing and learning system that is designed synergistically, we are able to reconstruct ECG signal by exploiting the relation of these two types of cardiovascular measurement.