Registration Number
0218U005000,
0115U000393
, R & D reports
Title
Development of neural-like methods of parallel-hierarchical processing of complex textured biomedical images
popup.stage_title
Head
Timchenko Leonid Ivanovich,
Registration Date
29-01-2018
Organization
State University of Infrastructure and Technology
popup.description2
The object of the study is the process of digital analysis of complex textural biomedical images in computer systems based on neural-like methods of parallel-hierarchical processing. The process of diagnostics using tomographic and thermographic imaging of the brain, tomography of the macular region of the retina of the eye, based on the use of modern methods for analyzing biomedical information through the application of the developed system. Subject of research - geometrical sizes and information signs of biomedical objects. Errors of digital processing of complex textural biomedical images of objects of various nature. The research methods are based on the developed theory of parallel-hierarchical image processing, the theory of digital processing and image analysis, on separate sections of applied mathematics and mathematical statistics. In the course of the research, the theory of algorithms was used to construct an algorithm for processing biomedical images of the macular region of the retina of the eye and the algorithm for the analysis of structural changes in the fundus of the fundus, based on fuzzy sets; fuzzy sets for evaluation of information signs of biomedical objects; methods of image filtering to eliminate noise of different nature and improve their quality; methods of clinical and medical research to verify the work and reliability of the developed system. The expected results are the development of neural-like principles and approaches for the recognition of biomedical images based on the theory of parallel-hierarchical transformation. Development of information technologies for forecasting the state of indicators of individual diseases. Development of methods for biomedical imaging analysis. Further development of mathematical models for the analysis of biomedical images on the basis of spatially-coherent preparation in the part of the allocation of characteristic features of images, its fragmentation (reduction of dimension), taking into account the invariance to the rotation of images, which allows to reduce hardware costs while increasing the impedance of recognition of biomedical images. The method of automatic selection of the optimal threshold of the gradient filter is improved, which increases the reliability of prediction of the rate of change of the studied indicators of biomedical images in the conditions of dynamic perturbing factors. Переводчик Google для бизнеса –Инструменты переводчикаПереводчик сайтов
Product Description
popup.authors
Боклаг Л.П.
Герцій О.А.
Кокряцька Н.І.
Маліщук В.В.
Павлов С.В.
Степанюк Д.С,
Тимченко Л.І.
Шишковський С.О.
popup.nrat_date
2020-04-02
