The method for structure learning of DWI-images, based on the analysis of the few fractal dimensions of the set of image pixels was developed. It is shown that the study about the structure of DWI-images using few fractal dimensions can be used to describe the ischemic focus and to identify areas of irreversible changes in the brain. The statistically significant differences and correlations between fractal dimensions of healthy tissue and ischemic focus were detected. It confirms the sensitivity of the method and demonstrates the dependence of the infarction area structure from the original morphology state of the brain tissue, that is determined not only the features of the formation of the ischemic focus, but the clinical severity of stroke. It was found that the structural change of the ischemic focus deceased patients significantly different from that of the surviving patients, which may be due to more accurate identification of areas of irreversible changes in comparison with the calculation of the diffusion coefficient. This increases the opportunity of usage of this method as a predictor of stroke outcome.
Prof. Dr. Bilal BİLGİN