Beta Wavelet Neural Networks for Medical Image Watermarking

A Fast and Robust Approach

Abstract

Smart devices and modern communication technologies now connect medical equipment more easily, which helps improve diagnostic processes. These systems use medical images to support diagnosis and decision-making, so it’s important to protect those images. Digital watermarking offers an effective way to secure medical images by embedding information that can verify authenticity, protect copyright, and ensure traceability throughout the healthcare workflow. To address this issue, this paper presents a robust and efficient medical image watermarking scheme that integrates the fast Beta wavelet transform (FBWT) with wavelet neural networks (WNN). Firstly, we created a library of activation functions containing a newly introduced family of Beta wavelets. Secondly, leveraging multi-resolution analysis (MRA) and fast wavelet transform (FWT), the medical image was decomposed to obtain wavelet coefficients. Finally, this approach embeds a watermark within the least significant contributions of the host medical image using WNN while maintaining high imperceptibility and robustness. Experimental results demonstrate the effectiveness of the scheme in balancing invisibility and resilience against various attacks, making it a promising solution for securing medical images in telemedicine applications. For imperceptibility evaluation, we used the peak signal-to-noise ratio (PSNR) and the structural similarity index measure (SSIM), with values of PSNR = 81.49 and SSIM = 1.000. In robustness testing, we measured the normalized correlation (NC) and bit error rate (BER), obtaining values of NC = 1.000 and BER = 0.