Comparison of Machine Learning Algorithms for Tumor Detection in Breast Microwave Imaging

Conventional breast imaging methods like Magnetic Resonance Imaging, ultrasound and X-Ray are relied upon by most clinics and doctors worldwide. Breast microwave imaging (BMI) is an alternative imaging technology which uses nonionizing radiation which safer for the body and has a lower cost. A pre-clinical BMI system using breast phantoms is used to create the open source University of Manitoba- BMI dataset (UM-BMID). In this paper, we explore the usability of the dataset, implement different machine learning classification algorithms for tumor detection on UM-BMID and compare our findings with the previously published results. The accuracy achieved was a maximum of 94% which shows great promise for use of machine learning techniques in breast microwave imaging.