UDK: 004.032.26

PHYSICAL INTERPRETATION OF NEURON TRAINING IN FUZZY CLUSTERING NETWORKS WHEN USING MICROTOMOGRAPHY IN MEDICAL AND BIOLOGICAL RESEARCH

S. V. Gorbachev1, L. S. Bykhalov2,3, N. N. Gorbacheva2

1Национальный исследовательский Томский государственный университет, 2Волгоградский государственный медицинский университет, 3Волгоградский научный медицинский центр

Abstract

The physical bases and interpretation of the procedure for training neurons in the fuzzy cellular neural networks of Kohonen have been investigated. Based on them, the method of automatic calibration of digital X-ray microtomographs has been described. This method involves neural-fuzzy segmentation of a distorted image into a finite number of clusters. The method is of great practical importance for the use of microtomography in biomedical research.

Keywords

X-ray microtomographs, detectors, images, calibration, neuro-fuzzy networks.

Contacts

Горбачев Сергей Викторович – к. т. н., с. н. с. международной лаборатории «Системы технического зрения», Национальный исследовательский Томский государственный университет, e-mail: spp03@sibmail.com