Subspace-Based Two-Step Iterative Shrinkage/Thresholding Algorithm for Microwave Tomography Breast Imaging

Area of Science:

• Medical Imaging
• Electromagnetics
• Computational Science

Background:

• Microwave tomography is a non-invasive breast imaging technique.
• Accurate reconstruction in noisy environments is a significant challenge for current methods.

Purpose of the Study:

• To develop an adaptive subspace-based two-step iterative shrinkage/thresholding (S-TwIST) algorithm.
• To improve the accuracy of microwave tomography breast imaging, particularly in noisy conditions.

Main Methods:

• Utilized singular value decomposition (SVD) for deterministic contrast source extraction.
• Implemented an adaptive strategy for optimal singular value selection.
• Incorporated deterministic induced currents for improved total field approximation, differing from conventional DBIM methods.

Main Results:

• Validated the S-TwIST algorithm using synthetic data and 45 digital breast phantoms.
• Demonstrated robust reconstruction across noise levels from 0-20 dB SNR.
• Achieved an average relative error of 0.4847% in breast tissue reconstruction without prior noise level information.

Conclusions:

• The S-TwIST algorithm significantly enhances reconstruction accuracy in microwave tomography for breast imaging.
• The method successfully recovers complex tissue structures and density distributions.
• Shows strong potential for future clinical applications in breast cancer detection.