intelligent approach to segmentation and classification of microbiological photographic images

Abstract

The task of urgent detection of pathogen microorganisms in the human body is a relevant problem in the field of medicine. An approach based on seeding the biomaterial into the media nutritious and monitoring the bacterial colony growth is on the popular side by today’s standards. At the same time it possesses a set of downsides, caused generally by the human factor, introducing possible mistakes to medical verdicts. This work is dedicated to the development of intelligent data-driven technologies for processing microbiological analysis data in the form of photographic images of Petri dishes. Such technologies must allow reducing the dependence on the human factor as well as increasing the key factor quality of the analysis. The results of the conducted experiments allow making a conclusion that developed heuristic and neural network methods for detection and classification of the microorganism colonies surpass existing approaches and allow automating the key stages of the microbiological analysis as well as introducing proposed methods in practice.