With over 1.7 million new cases annually, breast cancer is the leading cause of cancer death in women around the globe. Importantly, medical intervention has a greater chance of success if tumors can be detected during the early stages of the disease. This has led to a concerted effort on the part of the scientific community to develop advanced imaging technology which can detect smaller volumes of cancerous tissue.
The topic of this work is an algorithm designed to supplement the existing modality Digital Breast Tomosynthesis (DBT) with Microwave Radar (MWR) sensing. Using the DBT image, a physician would be asked to first identify any suspicious regions which may contain a tumor. Then a MWR computational model, such as Finite Difference Frequency Domain (FDFD), would be used with the DBT image to simulate two scenarios: a healthy case, and a diseased case with tissue dielectric constant increased to that of cancer over the suspicious region. The two results would then be compared to the actual measured MWR data to find the closest match using Maximum Likelihood Estimation.
In a preliminary numerical experiment, datasets representing MWR measurements with added Gaussian noise were generated using FDFD and openly available DBT data. In each of five cases, a team of undergraduate students, having no knowledge of the ground truth, correctly detected the presence (or absence) of a lesion using this algorithm. This result suggests that the algorithm has potential to augment existing technology. Ultimately, this information could be crucial for medical intervention.