Results
Using deep neural networks, we achieved a breast cancer prediction accuracy of 0.95 ± 0.003. MWR diagnostics provide results much faster than traditional mammography and ultrasound, offering rapid and highly accurate diagnostic predictions.
Passive microwave radiometry (MWR) has a long history. Recent advances in engineering and AI allowed its use for diagnostics of different pathologies.
MWR is a non-invasive technique that passively measures natural microwave emissions from human tissues. Cancerous tissues exhibit increased metabolic rates and produce elevated levels of microwave radiation, allowing effective differentiation between malignant and benign tissues.
In our studies, we employed the MWR2020 dual-band point-of-care device developed by MMWR Ltd. The device captures infrared emissions (indicating skin temperature) and microwave emissions (reflecting internal tissue temperature) between 3.5 and 4.2 GHz.
Temperature measurements were taken at 22 anatomical locations. The collected data was analyzed using deep neural networks to distinguish between healthy, benign, and malignant states.
Using deep neural networks, we achieved a breast cancer prediction accuracy of 0.95 ± 0.003. MWR diagnostics provide results much faster than traditional mammography and ultrasound, offering rapid and highly accurate diagnostic predictions.
We plan to enhance our AI-MWR system by integrating additional clinical biomarkers and refining our deep learning algorithms. Our goals include: