ROSEmark: Robust semi-blind ECG watermarking scheme using SWT-DCT framework

Coronary heart disease is a major cause of fatality worldwide. Online access to a patient's ECG record on networked hand-held devices enables its timely assessment by a cardiologist. However, the transmission of the ECG data on an unsecured network leads to a major concern about its authenticity. As a solution to this problem, this paper proposes ROSEmark - a robust semi-blind scheme for watermarking the ECG records. Each ECG record from the MIT-BIH Arrhythmia database is decomposed into equal-sized segments, and a set of key segments is selected randomly for watermarking. Each key segment is transformed using a 3-level stationary wavelet transform (SWT), and the resultant approximation vector is subjected to discrete cosine transform (DCT). An extreme learning machine (ELM) is applied for modulating the low-frequency coefficients, and the watermark is embedded in the modulated coefficients. Further, to ensure the security of the algorithm, the watermark is XORed with a randomly generated bitmask sequence. Finally, ROSEmark's performance is evaluated for fidelity and robustness using a variety of interferences. It is found that ROSEmark yields an average PSNR and PRD values of 50.72 dB and 0.7636, respectively. The experiments demonstrate that ROSEmark outperforms its competitors in terms of the standard evaluation metrics and is also time-efficient, thus making it suitable for real-time applications such as telemedicine.(C) 2022 Elsevier Inc. All rights reserved.

Automated summary

Coronary heart disease is a leading cause of mortality worldwide. This paper proposes a robust semi-blind watermarking scheme, ROSEmark, for authenticating ECG records transmitted through networked devices. Experimental results demonstrate that ROSEmark outperforms competitors in terms of fidelity and robustness, making it suitable for real-time applications such as telemedicine.