LAUSR

Glioblastoma Multiforme is the most common and most aggressive type of brain tumors grade four astrocytoma. Although accurate prediction of Glioblastoma borders and shape is absolutely essential for neurosurgeons, there are not many in silico platforms that can make such predictions. In the current study, an automatic patient-specific simulation of Glioblastoma growth is described. A finite element approach is used to analyze the magnetic resonance images from patients in the early stages of their tumors. For segmentation of the tumor, support vector machine method, which is an automatic segmentation algorithm, is used. Using in situ and in vivo data, the main parameters of tumor prediction and growth are estimated with high precision in proliferation-invasion partial differential equation, using genetic algorithm optimization method. The results show that for a C57BL mouse, the differences between the surface and perimeter of in vivo test and simulation prediction data, as objective function, are 3.7% and 17.4%, respectively.