Guided classifier in the diagnosis of breast cancer according to microwave radiothermometry

This work was carried out as part of the problem of developing a consultative expert system for the diagnosis of breast cancer according to microwave radiothermometry. Microwave radiometry is a biophysical methodof non-invasive examination, which consists in measuring internal and surface temperatures by the intensity of their thermal radiation, respectively, in the microwave and infrared ranges. Over the past two decades, this method has become widespread in various fields of medicine.
The quite complicated problem has arisen, difficulty in perception of information arising from medical personnel without special long-term training. This greatly reduces the possibility of widespread use of the method in screening.
Thus, the actual task is is the creation of an expert system that has the ability to substantiation of the proposed diagnostic solution. Therefore, the task of constructing a sufficiently complete space of highly informative signs of diseases is of particular importance. The problem of managing the quality of classification algorithms also acquires special importance, since there is a need to use an expert system in a wide variety of patient groups (age, etc.)
During the analysis of the results of examinations over the past two decades, specialists from various clinics identified a number of qualitative signs of the behavior of temperature fields in patients with breast cancer.
The first studies of a quantitative analysis of functions describing the presence of anomalies in the temperature fields of patients appeared. The complex of such functions determines the presence of anomalies in the behavior of temperature fields inside the mammary glands with a sufficiently high efficiency.
The description of the characteristic subdomains of sets of changes in functions is one of the most significant aspects of such an approach to the construction of a feature space.
However, the use of multidimensional thermometric features gives a greater effect. Their construction is based on the construction of a feature space using vector functions.
This article discusses the construction of multidimensional thermometric diagnostic features and evaluates the effectiveness of their use. The article also discusses the problem of developing a controlled classifier based on two-dimensional thermometric features.
The effectiveness of the proposed approach has been proved in determining the severity of temperature anomalies, and, as a result, the effectiveness of the method in the diagnosis of breast cancer in the examined patients.