Breast Imaging with Microwaves: Interpreting Images and Signals through Multimodal Registration and Signal Classification
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Early detection of breast cancer is critical to improving outcomes for patients. While x-ray mammography is the current gold standard, limitations include ionizing radiation and reduced sensitivity in dense breast tissue. Microwave breast imaging (MBI) offers a promising, complementary technology to mammography that measures the dielectric properties (permittivity) of the tissue to identify abnormalities. However, clinical validation of MBI remains challenging because the geometry and placement of the breast in the imaging system differs from mammography and few tools exist to relate these images to established clinical modalities. These limitations of MBI can be addressed by using laser scanning to understand the breast placement, supporting image interpretation and tissue classification. This research aims to improve the accuracy and clinical relevance of MBI by combining laser scanning of breast placement, multimodal image registration with mammography, and frequency domain signal analysis, demonstrating sub-centimeter consistency of breast placement and quantifiable correspondence between microwave permittivity and fibroglandular tissue. First, a laser scanning workflow was developed to measure 3D breast shape and volume and assess the consistency across multiple placements of the breast in the microwave scanner. This enables assessment of positional changes between scans and corresponding impact on the image. Second, contours describing the surface of the breast were derived from the laser scans and used to register microwave images with mammograms, enabling comparison between dielectric properties and fibroglandular tissue distribution. This has not been possible with other MBI systems. Finally, frequency domain microwave signals were analyzed using principal component analysis and clustering to identify signal features that correspond to the underlying tissue types. This analysis aims to provide interpretability to the signals used to create images and therefore improves confidence in identifying regions of interest such as fibroglandular tissue or tumours. Results demonstrated spatial correspondence between high permittivity regions and fibroglandular tissue. Expected signal characteristics reflecting tissue composition were related to the permittivity values in the microwave image. These findings establish a foundation for more accurate, consistent, and clinically interpretable microwave breast imaging, supporting its translation toward breast health and diagnostics applications.
Fields of Study & Subjects
Microwave Imaging and Scattering Analysis
Infrared Thermography in Medicine
Electrical and Bioimpedance Tomography
Publication details
- Publication date
- 26 March 2026
- Language
- EN
- Document type
- Thesis