Scholarlink

<p>The increasing rise of e-healthcare and telemedicine needs secure medical image transmission and storage in order to protect patients' sensitive medical information. Traditional approaches for medical image tamper detection struggle to resist content-preserving changes such as resizing and compression. This paper presents an innovative, robust method for detecting&#xa0;tampered&#xa0;medical images while preserving crucial diagnostic data. The proposed method consists of the following stages: segmentation, hashing and embedding, extraction, and verification.&#xa0;Regions of interest (ROI) and regions of non-interest (RONI) are first determined in medical images employing a Hybrid Residual U-Net + + (HResUN) framework. Next, the unique hash values for the&#xa0;crucial regions&#xa0;are produced by Locality-Sensitive Hashing (LSH) and the non-critical region is divided into low-frequency and high-frequency sub-bands using the Discrete Wavelet Transform (DWT).Finally hashed ROI is&#xa0;embedded into the high non-critical regions using the method of&#xa0;Least Significant Bit (LSB).&#xa0;When transmitted, hash values are extracted, recalculated, and compared to ensure authenticity. The proposed approach uses measures such as PSNR, SSIM, MSE, NCC, and TAF to validate tamper detection accuracy and image quality. This technique is developed in Python and achieves a high PSNR of 38.79&#xa0;dB.</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

Robust medical image tamper detection using hybrid residual U-Net ++ and DWT-LSB based hash embedding

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月5日

查看全文
  • Pullaiahgari Sarika,
  • R. Shankar

摘要

The increasing rise of e-healthcare and telemedicine needs secure medical image transmission and storage in order to protect patients' sensitive medical information. Traditional approaches for medical image tamper detection struggle to resist content-preserving changes such as resizing and compression. This paper presents an innovative, robust method for detecting tampered medical images while preserving crucial diagnostic data. The proposed method consists of the following stages: segmentation, hashing and embedding, extraction, and verification. Regions of interest (ROI) and regions of non-interest (RONI) are first determined in medical images employing a Hybrid Residual U-Net + + (HResUN) framework. Next, the unique hash values for the crucial regions are produced by Locality-Sensitive Hashing (LSH) and the non-critical region is divided into low-frequency and high-frequency sub-bands using the Discrete Wavelet Transform (DWT).Finally hashed ROI is embedded into the high non-critical regions using the method of Least Significant Bit (LSB). When transmitted, hash values are extracted, recalculated, and compared to ensure authenticity. The proposed approach uses measures such as PSNR, SSIM, MSE, NCC, and TAF to validate tamper detection accuracy and image quality. This technique is developed in Python and achieves a high PSNR of 38.79 dB.