DEVELOPMENT OF METHODS FOR THE PROCESSING OF MEDICAL IMAGES USING GENETIC ALGORITHMS

In this paper we deal with analysis and evaluation of objects of interest which are present in ultrasound images as well as assessment of the progress or regressions that occurred in these objects. These objects are highly significant from a medicinal perspective and include atherosclerostic plaque in carotid arteries, the intima-media thickness in the distal part of the common carotid artery, cerebral cortex size and brain stem Findings in cases of Parkinson disease. Here, we describe procedures employing combination of common methods and evolutionary algorithms for recognizing points of interest in the images that may serve in determining various parameters and properties of analyzed objects. We use the evolutionary algorithms to optimize the energy function of deformable models used to approximate the locations and shapes of object boundaries in images. We suppose that evolutionary algorithms can be used to find the desired global solution. Evolutionary algorithms are based on principles of evolution found in nature and respect the Darwin's theory of natural selection according to the defined cost function and gene recombination and mutation. As the computation of gradient vector flow field and also the evolution of active contour are computationally very expensive, we investigate the suitability of the GPU for a parallel implementation. In conclusion, we compare our approach with common numerical methods on real medical images segmentation.