A generalised wavelet packet‐based anonymisation approach for ECG security application

Abstract

AbstractAn electrocardiogram (ECG) signal contains cardiovascular information, which is private in nature. Therefore, during the distribution and storage of an ECG in a public repository, it is necessary to anonymise the ECG data. An eavesdropper could record an unsecured ECG and use it as recognition data to gain access via an ECG biometric system. Anonymised ECG data also hide cardiovascular details of a patient upholding Health Information Protection and Privacy Act (1996) regulations. In this paper, a generalised wavelet packet‐based ECG anonymisation framework is proposed. The proposed method uses wavelet packet transform and reversible function and/or operation to conceal fiducial and non‐fiducial features from normal and abnormal ECG signals. At the receiver end, only authorised personnel possessing a secret key and know the reversible function will be able to reconstruct the original ECG from the anonymised ECG. Performance analysis using cross correlation, power spectral density and percentage residual difference methods show that the proposed anonymisation algorithm conceals fiducial and non‐fiducial features in normal and abnormal ECG signals. Moreover, the reconstructed ECG is highly correlated with the original ECG, which achieves a lossless reconstruction of the ECG data and proves the robustness of the proposed method. It is found from the performance analysis results that the proposed anonymisation scheme provides high‐security protection to ECG data and patient privacy. Copyright © 2017 John Wiley & Sons, Ltd.