Practical Microwave Breast Imaging Device

About This Grant

Project Summary The objective of this project is to develop an advanced microwave (MW) imaging system for early detection of breast cancer (BC). MW, a form of non-ionizing radiation, has the potential to serve as an effective imaging/screening tool for various diseases, including BC. In MW breast imaging (MBI), the dielectric properties of cancerous breast tissue differ significantly (0.1~10 times larger) from those of the healthy breast tissues at MW frequencies. This forms the foundation of MBI for high-contrast breast image. Next, the nonionizing-radiation dose associated with a regular MBI examination is equivalent to or less than that of a wireless phone call, making it safer than ionizing approaches. As a result, MBI has the potential to replace X-ray, the current primary method for BC screening, and serve as a routine screening tool. This project aims to address two key challenges present in existing MW breast imaging, bridging the gap between laboratory prototypes and a market-ready product. Existing MBI prototypes in laboratories employ multiple antennas positioned around the breast to form a synthetic aperture array, collecting response signals from tumors. However, the limited workspace allows for only a few antennas (usually 10~20), resulting in a restricted number of signals available for image reconstruction. To overcome this limitation and effectively address the ill-posed inverse problem, a larger image with more pixels necessitates an increased amount of measurement data. Furthermore, the operation frequency of MBI is typically limited to a few Gigahertz, restricting the spatial resolution to the centimeter level, even when utilizing the best image reconstruction algorithm available. This level of resolution is insufficient for identifying small tumors near breast glands. These two shortcomings have been the primary reasons that have delayed the realization of a practical MBI product for clinic use. To address these challenges, this project aims to develop an enhanced MBI device featuring a dynamic antenna system and an AI image-formation model. The system will comprise only two antennas, one functioning as a transmitter and the other as a receiver. Both antennas have independent rotation capabilities around the breast, allowing signal collection from any desired azimuthal positions. This setup creates a virtual synthetic aperture array equivalent to multiple real antennas but provides a greater number of signals for image reconstruction. Additionally, recent studies have demonstrated that machine learning (ML) technologies can enhance the spatial resolution of near-field MW imaging. By leveraging our previous experience in ML-based MW imaging, we will incorporate ML technology into MBI in this project and compare it with conventional algorithms. The proposed ML algorithm will adopt an unsupervised approach, eliminating the need of patient information or high-quality phantom images as ground truth. By addressing these two critical issues in MBI, a practical MBI product capable of performing BC screening in clinics can be produced, thereby effectively serving women patients very soon.