Scholarlink

<p>Medical images contain sensitive and private health information of patients, which is crucial to be safeguarded from unauthorized access. Various encryption schemes are used for improving security; however, they are subject to adversaries associated with higher complexity, ineffective compression ratios, and slower responses in real time. Therefore, to address these limitations as well as to establish stronger security in medical image sharing, the research proposes a Biometric Fused Federated Generative Adversarial Network (BF2GAN) method. The federated learning (FL) concept is included in this research, which provides collaborative training while reducing the risks of gradient inversions. Moreover, the decentralized training strengthens data integrity and reduces the risk of data exposures. The biometric information is included for user verification and key generation, which assists in creating highly secure images that are difficult to decode without the correct keys. The method minimizes the reliance on key management algorithms and establishes a simplified encryption process that strengthens the overall system security. In the LUNA16 database, the BF2GAN achieves a significant performance in terms of 3.08&#xa0;s encryption time, 0.975 structural similarity index measure, 0.82 Feature similarity index, 3.9&#xa0;s decryption time, 58.38 decibels of peak signal-to-noise ratio, and minimum memory usage of 280.87 kilobytes, compared to the state-of-the-art methods.</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

BF2GAN: A Fused Federated Learning–Based Biometric Enabled Deep Learning Approach for Secure Medical Image Sharing in Cloud Environment

  • 期刊论文

  • 发表日期：

    2026年4月27日

查看全文
  • Zeeshan Ahmed Mohammed,
  • D. Rajeshwari,
  • Shanavaz Mohammed,
  • M. Sreelaxmi

摘要

Medical images contain sensitive and private health information of patients, which is crucial to be safeguarded from unauthorized access. Various encryption schemes are used for improving security; however, they are subject to adversaries associated with higher complexity, ineffective compression ratios, and slower responses in real time. Therefore, to address these limitations as well as to establish stronger security in medical image sharing, the research proposes a Biometric Fused Federated Generative Adversarial Network (BF2GAN) method. The federated learning (FL) concept is included in this research, which provides collaborative training while reducing the risks of gradient inversions. Moreover, the decentralized training strengthens data integrity and reduces the risk of data exposures. The biometric information is included for user verification and key generation, which assists in creating highly secure images that are difficult to decode without the correct keys. The method minimizes the reliance on key management algorithms and establishes a simplified encryption process that strengthens the overall system security. In the LUNA16 database, the BF2GAN achieves a significant performance in terms of 3.08 s encryption time, 0.975 structural similarity index measure, 0.82 Feature similarity index, 3.9 s decryption time, 58.38 decibels of peak signal-to-noise ratio, and minimum memory usage of 280.87 kilobytes, compared to the state-of-the-art methods.