Scholarlink

With the use of computers, physicians may more precisely detect anomalies in the liver and reduce the need for surgical interventions. Medical imaging methods such as computed tomography (CT) and magnetic resonance imaging (MRI) are primarily responsible for the early identification and detection of liver abnormalities. Due to the high noise and low resolution of these pictures, it might be difficult to segment and identify liver lesions in them. On this issue, several novel machine learning and image analysis approaches have been employed progressively but their effectiveness remains limited. There is still a need for an automated and precise model that can monitor, identify, and diagnose hepatic lesions in the 3D volumes of CT and MRI. In this research, we propose a privacy-preserving architecture based on federated learning to help clinicians use their time more effectively. Several pre-trained models are employed; our suggested FedVGG19 model obtains a test accuracy of 97.03%.

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

Transforming Healthcare Analytics: Federated Learning for Privacy-Preserving Liver Lesion Detection and Classification

  • 章节

  • 发表日期：

    2026年

查看全文
  • Md. Shahidul Islam,
  • Jayonto Dutta Plabon,
  • Nazneen Nahar,
  • Nishat Tasnim,
  • Shah Murtaza Rashid Al Masud,
  • Jungpil Shin,
  • Moshiur Rahman Tonmoy

摘要

With the use of computers, physicians may more precisely detect anomalies in the liver and reduce the need for surgical interventions. Medical imaging methods such as computed tomography (CT) and magnetic resonance imaging (MRI) are primarily responsible for the early identification and detection of liver abnormalities. Due to the high noise and low resolution of these pictures, it might be difficult to segment and identify liver lesions in them. On this issue, several novel machine learning and image analysis approaches have been employed progressively but their effectiveness remains limited. There is still a need for an automated and precise model that can monitor, identify, and diagnose hepatic lesions in the 3D volumes of CT and MRI. In this research, we propose a privacy-preserving architecture based on federated learning to help clinicians use their time more effectively. Several pre-trained models are employed; our suggested FedVGG19 model obtains a test accuracy of 97.03%.