Scholarlink

<p>Protection of sensitive information in medical images is crucial for diagnosing health issues accurately. Unauthorized modification of such images may lead to misdiagnoses, potentially harming patient health. In this paper, we propose a novel fragile zero watermarking scheme for medical image tamper detection and Region of Interest (ROI) recovery, using the Chinese Remainder Theorem (CRT) to generate multiple compact shares of each ROI pixel. In the proposed method, the selected ROI is preserved unaltered as a non-embedding region through zero-watermarking technique, while the Least Significant Bits (LSBs) of the Region of Non-Interest (RONI) serve as the embedding region. CRT-based shares of the ROI are generated, scrambled, and merged to form a watermark. Unlike conventional approaches that separate authentication and recovery data, the proposed method uses the same CRT shares interchangeably to detect and recover tampering. In this way, even if one share is tampered, it does not affect the remaining shares, and ROI recovery remains feasible. Tamper detection and localization are performed via majority voting across the shares to losslessly reconstruct ROI pixels from any subset of untampered shares, ensuring accurate detection and ROI reversibility. Moreover, when recovery is not possible, authentication is still ensured. The experimental results on MRI, CT, ultrasound, and microscopic images demonstrate that the proposed scheme offers superior performance compared to state-of-the-art methods, achieving higher imperceptibility (PSNR &gt; 51 dB), larger ROI sizes (up to 22% of the image), and higher tamper detection accuracy. Moreover, it enables effective ROI recovery even when up to 44% of the image is altered, thereby underscoring the scheme’s potential for reliable and efficient medical image protection.</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

A novel CRT-based zero watermarking for medical image tamper detection and recovery

  • 期刊论文

  • 发表日期：

    2026年5月29日

查看全文
  • Wassila Belferdi,
  • Ali Behloul

摘要

Protection of sensitive information in medical images is crucial for diagnosing health issues accurately. Unauthorized modification of such images may lead to misdiagnoses, potentially harming patient health. In this paper, we propose a novel fragile zero watermarking scheme for medical image tamper detection and Region of Interest (ROI) recovery, using the Chinese Remainder Theorem (CRT) to generate multiple compact shares of each ROI pixel. In the proposed method, the selected ROI is preserved unaltered as a non-embedding region through zero-watermarking technique, while the Least Significant Bits (LSBs) of the Region of Non-Interest (RONI) serve as the embedding region. CRT-based shares of the ROI are generated, scrambled, and merged to form a watermark. Unlike conventional approaches that separate authentication and recovery data, the proposed method uses the same CRT shares interchangeably to detect and recover tampering. In this way, even if one share is tampered, it does not affect the remaining shares, and ROI recovery remains feasible. Tamper detection and localization are performed via majority voting across the shares to losslessly reconstruct ROI pixels from any subset of untampered shares, ensuring accurate detection and ROI reversibility. Moreover, when recovery is not possible, authentication is still ensured. The experimental results on MRI, CT, ultrasound, and microscopic images demonstrate that the proposed scheme offers superior performance compared to state-of-the-art methods, achieving higher imperceptibility (PSNR > 51 dB), larger ROI sizes (up to 22% of the image), and higher tamper detection accuracy. Moreover, it enables effective ROI recovery even when up to 44% of the image is altered, thereby underscoring the scheme’s potential for reliable and efficient medical image protection.