Singling out ischemic lesion zones and transplanted mesenchymal stem cells in the rat brain MRI

The paper is devoted to the study and development of methods for automatic finding and localization of ischemic lesion zones and transplanted mesenchymal stem cells in the brain of laboratory animals. The developed methods are aimed at supporting the work of doctors-researchers. The methods of preliminary processing are proposed aimed at improvement of MRI data, as well as singling out the area of the brain, which is necessary for automation of finding and visualization of zones of interest. The following methods were used to single out the ischemic lesion zones: method of nearest neighbor, method of random forest, artificial neural networks, analysis of textural features using a classifier on the basis of Euclidean-Mahalanobis distance. The stem cells are searched and visualized on the slice images with the help of many-stage processing of pairs of images before and after transplantation. Instruments for cognitive combined 2D and 3D visualization are presented. © 2018 IEEE