Scholarlink

<p>The security of medical images is a growing concern in digital healthcare, especially with the increasing risks of unauthorized access and tampering. This paper proposes a robust multi-step watermarking model designed to ensure the integrity and confidentiality of medical image data. The model begins by converting patient metadata into binary format using ASCII encoding and enhances robustness through forward error correction with Hamming and BCH codes. A Convolutional Neural Network (CNN) is employed to detect the Region of Non-Interest (RONI), ensuring that the watermark is embedded only in diagnostically safe areas. The watermark is then embedded within the LL sub-band of a one-level Discrete Wavelet Transform (DWT), followed by Discrete Cosine Transform (DCT) and Singular Value Decomposition (SVD), achieving imperceptibility and resilience. Experimental evaluation on medical images demonstrated strong performance with PSNR values ranging from 35.2&#xa0;dB to 38.4&#xa0;dB, SSIM between 0.92 and 0.96, and Normalized Correlation (NC) values close to 1 under various image processing and malicious attacks. The proposed method ensures robustness against geometric and signal-based distortions while maintaining high image fidelity, making it highly applicable for secure telemedicine and electronic health record systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Securing medical images using multi-stepwatermarking model

  • 期刊论文

  • 发表日期：

    2025年7月22日

查看全文
  • Saima Kanwal,
  • Feng Tao,
  • Rizwan Taj,
  • Qamar Abbas,
  • Farhan Amin

摘要

The security of medical images is a growing concern in digital healthcare, especially with the increasing risks of unauthorized access and tampering. This paper proposes a robust multi-step watermarking model designed to ensure the integrity and confidentiality of medical image data. The model begins by converting patient metadata into binary format using ASCII encoding and enhances robustness through forward error correction with Hamming and BCH codes. A Convolutional Neural Network (CNN) is employed to detect the Region of Non-Interest (RONI), ensuring that the watermark is embedded only in diagnostically safe areas. The watermark is then embedded within the LL sub-band of a one-level Discrete Wavelet Transform (DWT), followed by Discrete Cosine Transform (DCT) and Singular Value Decomposition (SVD), achieving imperceptibility and resilience. Experimental evaluation on medical images demonstrated strong performance with PSNR values ranging from 35.2 dB to 38.4 dB, SSIM between 0.92 and 0.96, and Normalized Correlation (NC) values close to 1 under various image processing and malicious attacks. The proposed method ensures robustness against geometric and signal-based distortions while maintaining high image fidelity, making it highly applicable for secure telemedicine and electronic health record systems.