Search academic texts

Information × Registration Number 0223U003173, 0120U102134 , R & D reports Title Mathematical and software methods for processing multimodal monitoring data of biomedical objects to diagnose patients' health status popup.stage_title Head Dychka Ivan A., Доктор технічних наук Registration Date 03-04-2023 Organization National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute" popup.description2 The work is devoted to the development of methods, algorithms and software tools that will allow medical institutions to improve the quality of diagnosis and monitoring of patients' conditions due to the comprehensive presentation of multimodal medical data and their processing using the latest intelligent technologies. Proposed solutions to the problem of presentation and processing of temporal multimodal data of medical diagnostics. A mathematical apparatus for aggregation and synchronization of medical diagnostic data based on an algebraic system of aggregates has been developed. A three-level generalized semantic model of the patient was developed. The most popular CNN architectures for binary and multi-class classification of oncological diseases are studied. The problems of processing medical images of human organs, mammary gland, lungs and brain, cancer detection and development of medical diagnostic systems as the ultimate goal are considered. New approaches have been developed to increase the effectiveness of cancer detection compared to known approaches. The main general idea of the proposed approach is the development of a new class of CNN structures - hybrid CNNs using different classes of neural networks combined into a common CNN architecture. Numerous experimental studies have proven their better performance on various classification criteria compared to well-known CNN architectures. Product Description popup.authors popup.nrat_date 2023-04-03 Close