SAFE: A microwave imaging device for breast cancer early screening

Abstract

Background: SAFE (Scan and Find Early) is a microwave breast cancer imaging (MBI) device with the protentional for non-invasive and non-ionizing breast cancer early screening. The MBI technology utilizes the difference in dielectric properties of cancerous and healthy tissue to make further clinical predictions. In this study, we assess the ability of the SAFE to accurately detect and classify the existing lesion inside patient’s breast. Materials and Methods: Only patients scheduled for the biopsy were included in the study. The study was approved by the Ethics committee of Marmara University School of Medicine and in accordance with both institutional and national ethical standards in research and with the World Medical Association Declaration of Helsinki. Machine learning (ML) approach, namely Stochastic Gradient Descent (SGD), was used to detect the lesions inside patient’s breasts based on the difference in backscattered signals of healthy and lesion affected breasts, while ML Adaptive Boosting (AdaBoost) approach was used to determine the pathology of detected lesions. Due to the limited dataset, stratified 5-fold cross-validation was used to assess the proposed models and test their performance. Localization of detected lesions was assessed based on the inverse scattering algorithm, namely linear sampling method (LSM), used to reconstruct the image of the patient’s breasts. Results: Dataset included 113 patients, 70 with benign and 43 with malignant findings in one of the patient’s breast. The proposed detection model achieved the sensitivity, specificity and accuracy of 80.5%, 81.4% and 81%, respectively. Furthermore, proposed classification model had the accuracy of 82.5%, sensitivity at 84.6% and specificity at 81%. Device correctly localized 83% of the detected lesions. Conclusion: The study results show that our MBI system is capable to detect, localize and classify majority of breast lesions present. SAFE may have the potential to impact breast cancer early screening positively, due to its non-invasive and harmless nature. Further clinical studies are planned to validate acquired results.