Scholarlink

<p>Early detection of breast cancer significantly enhances treatment outcomes, making advanced imaging technologies essential. While traditional methods such as X-ray mammography, MRI, and ultrasound are commonly used, microwave imaging (MI) has emerged as a safe, low-cost, and non-ionizing alternative. However, microwave tomographic image formation is inherently nonlinear, ill-posed, and low in spatial resolution. This study focuses on improving image reconstruction accuracy using the Gauss–Newton Krylov (GNK) method, a computationally efficient iterative approach for solving nonlinear inverse problems. The GNK algorithm, implemented in MATLAB with finite element method (FEM)-based forward solvers, was evaluated using MRI derived numerical phantom dataset from the University of Wisconsin-Madison. Comparative analysis with Gauss–Newton (GN) and Quasi-Newton (L-BFGS) methods reveals that GNK offers faster convergence, improved spatial resolution, and better artifact suppression. These results suggest that the GNK method holds significant promise for clinical microwave imaging applications, providing a reliable alternative to conventional iterative algorithms.</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

Microwave Imaging: Using Gauss Newton Krylov Method for Breast Cancer Detection Purpose

  • 期刊论文

  • 发表日期：

    2026年4月9日

查看全文
  • Moutusi Mondal,
  • Debipriya Dutta,
  • Somnath Mahato

摘要

Early detection of breast cancer significantly enhances treatment outcomes, making advanced imaging technologies essential. While traditional methods such as X-ray mammography, MRI, and ultrasound are commonly used, microwave imaging (MI) has emerged as a safe, low-cost, and non-ionizing alternative. However, microwave tomographic image formation is inherently nonlinear, ill-posed, and low in spatial resolution. This study focuses on improving image reconstruction accuracy using the Gauss–Newton Krylov (GNK) method, a computationally efficient iterative approach for solving nonlinear inverse problems. The GNK algorithm, implemented in MATLAB with finite element method (FEM)-based forward solvers, was evaluated using MRI derived numerical phantom dataset from the University of Wisconsin-Madison. Comparative analysis with Gauss–Newton (GN) and Quasi-Newton (L-BFGS) methods reveals that GNK offers faster convergence, improved spatial resolution, and better artifact suppression. These results suggest that the GNK method holds significant promise for clinical microwave imaging applications, providing a reliable alternative to conventional iterative algorithms.