Filatova H. Methods and means of decision support in biomedical systems based on the morphological analysis of biomedical signals and images

Object of the study - the process of formation of diagnostic decisions in conditions of significant a priori uncertainty on the basis of instrumental methods of examination of the functional state of human body subsystems. The purpose of the study is to increase the effectiveness of the instrumental examination of patients and reduce the risk of making incorrect decisions in decision support systems by developing new methods for morphological analysis of biomedical signals and images with locally concentrated features that take into account the specifics of processed signals and images in the form of useful signal models. Methods of the study: system analysis, the theory of morphological analysis, the theory of digital signal processing and images, the theory of decision making, cluster analysis, the method of registration of X-ray biomedical images, digital filtering methods and elemental image transformations. Theoretical and practical results: the models and methods of morphological analysis of biomedical signals and images with locally concentrated features developed are the scientific and methodological basis for the development of appropriate algorithms and software. The generalized method for the morphological analysis of biomedical signals with locally concentrated features using a multichannel matched morphological filter was developed. The system of alternative diagnostic features in the morphological analysis of biomedical signals with locally concentrated features taking into account the model of a useful one-dimensional signal and the properties of the detecting function of matched morphological filter was developed. . The mathematical model of a digital X-ray image that takes into account the visualization features of biological objects and their structural elements was developed. The method for improving the quality of visualization of biological objects on X-ray images of a wide class was developed on the basis of the morphological analysis of biomedical images with locally concentrated features. The objective integral criterion for assessing the quality of visualization of biological objects and their structural elements on medical low-contrast grayscale X-ray images was developed. The structure of the generalized biomedical decision support system as well as the software for modules of morphological analysis of biomedical signals and images with locally concentrated features were developed. The adequacy of the developed methods of morphological analysis in the processing of real biomedical signals and images was checked. The scientific novelty of the results is that the complex approach to the solution of the scientific and applied problem of increasing the efficiency of functioning of biomedical decision support systems on the basis of the morphological analysis of biomedical signals and images with locally concentrated features using models of useful signal was first proposed, justified and applied. Degree of implementation: the results of dissertation work are introduced in the company "Radmir" of subsidiary of the joint stock company of the Research Institute of Radio Engineering Measurements (Kharkiv); LLC "TREDEX Company" (Kharkiv); State institution "Institute of Medical Radiology named after S.P. Grigoryev of National Academy of Sciences of Ukraine" (Kharkiv); Municipal institution of healthcare "Center for Emergency Medical Care and Disaster Medicine" (Kharkiv); Department of "Computer Engineering and Programming" NTU "KhPI". Scope of use: in the field of medical instrumentation.