Improving Prediction Error Expansion with Regression Using Mirror Embedding for Reversible ECG Steganography

Health developments have made it easier to transfer patient information although raising concerns about data vulnerability. In such scenarios, ensuring patient data security when transmitted and stored is essential. Steganography with the Prediction Error Expansion (PEE) can be a solution for storing critical patient data in ECG signals. However, PEE has shortcomings in balancing the quality and capacity of the embedding with the threshold used. Therefore, the method’s improvisation utilizes a mirror embedding scheme in PEE with a regression predictor to decrease the disparity of prediction results. To evaluate the method, the experiment used datasets from ECG MIT-BIH. The results show that the resulting ECG signal can be maintained in a manner that is as similar as possible to the original signal. The quality of the embedding results can also be maintained above 52.8 dB for SNR and below 0.252 for PRD at high bps, which is higher than that of other PEE-based or newest ECG steganographic methods. The resulting algorithm shows an increasing speed of up to 22.7 times from the existing method.