Signal Processing in the Problem of Breast Tumors Detection by Impedansometry Method Based on Havriliak-Negami Model

This article is devoted to the processing of impedance data for breast examination. Breast tumors are a serious medical problem. Their timely detection requires periodic examinations over a long period of time. For monitoring, it is advisable to use methods that are not associated with radiation exposure and other harmful factors. These include impedance measurements, since the dielectric properties of tumors differ from the properties of healthy tissues. The authors propose the use of the Havriliak-Negami model for processing primary measurement data. As a result of solving the inverse problem, informative parameters of the model are calculated that describe the dielectric properties of the object of study. To implement laboratory experiments, an original multielectrode design of a dielectric sensor has been proposed, which can be used both for detection and localization of tumors. The data received from the sensor includes about a thousand parameters. It is proposed to process such a data set using machine learning methods. A dataset is created to train and test the intelligent classifier. It is formed based on data from laboratory experiments using a phantom. Assessing the sensitivity and localization capabilities of the proposed method is the immediate task of further research.