In this paper authors present their new proposition of system for cognitive analysis of dynamic computer tomography perfusion maps (dpCT). The novel contribution of this article is introducing an augmented reality visualization module that supports real time volume rendering (VR) of derived data. Authors also presents the results of their researches on optimization of VR algorithm memory usage by dynamic computation of volume gradient instead of pre-generation of gradient Authors compare five different discrete gradient computation schemas taking into account image quality and processing speed on two VR algorithms: volume ray casting and texture based visualization with view aligned slices.