A-prior-knowledge-informed Microwave Medical Imaging for Breast Cancer Diagnosis
Lide Hu,Yahui Ding,Zheng Gong,Yifan Chen
DOI: https://doi.org/10.1109/piers62282.2024.10618734
2024-01-01
Abstract:Microwave medical imaging (MMI), functioning across a wide frequency range from hundreds of megahertz to tens of gigahertz, shows potential for offering proactive healthcare solutions to patients with acute (e.g., suitable for early diagnosis) or chronic (e.g., ideal for continuous monitoring) conditions. However, developing accurate and robust MMI algorithms is challenging due to inherently complex inverse scattering problems and low dielectric contrast between healthy and diseased tissues. Therefore, employing A-Priori-Knowledge (APK) is crucial for enhancing MMI’s effectiveness in practical applications and clinical settings. This paper proposes a microwave quantitative imaging algorithm DBIM-APK based on spatial APK. It involves fusing data from APK based on the current patient’s MMI and from APK based on the existing MRI database, resulting in a hybrid APK. This hybrid APK is then incorporated into DBIM as the final APK. Compared to traditional DBIM, this algorithm doesn’t solve the contrast function through step-by-step iteration, but instead, it is directly provided by the hybrid spatial APK. According to the numerical results, the boundary artifacts caused by the inverse scattering nonlinear problem are basically eliminated after the introduction of spatial APK and the SSIM (Structural Similarity Index Measure) of the reconstructed image compared to the original image has improved from 0.62 to 0.95. This algorithm optimizes computational resources, reduces imaging time, improves imaging accuracy, enhances spatial resolution, and more clearly restores the structure of the breast.