A contextual based double watermarking of PET images by patient ID and ECG signal

This paper presents a novel digital watermarking framework using electrocardiograph (ECG) and demographic text data as double watermarks. It protects patient medical information and prevents mismatching diagnostic information. The watermarks are embedded in selected texture regions of a PET image using multi-resolution wavelet decomposition. Experimental results show that modifications in these locations are visually imperceptible. The robustness of the watermarks is verified through measurement of peak signal to noise ratio (PSNR), cross-correlation (CC%), structural similarity measure (SSIM) and universal image quality index (UIQI). Their robustness is also computed using pixel-based metrics and human visual system metrics. Additionally, beta factor (β) as an edge preservation measure is used for degradation evaluation of the image boundaries throughout the watermarked PET image. Assessment of the extracted watermarks shows watermarking robustness to common attacks such as embedded zero-tree wavelet (EZW) compression and median filtering.