Study Of Brain Tumor Growth Model Based On MR Images

Malignant brain tumors grow rapidly and invade the surrounding brain tissue,which is a serious threat to human health.It is difficult to completely remove tumor cells through surgery,resulting in high recurrence rate and poor prognosis after surgery.With the development of information technology,three-dimensional reconstruction of human tissues based on MR images can provide a more intuitive and clear understanding of the patient’s condition;tumor growth modeling by computer can predict tumor development and quantify tumor growth rate and invasion degree.Therefore,with the help of medical imaging and computer modeling can assist doctors in diagnosing the condition and formulating surgical plans,thus improving the treatment effect and increasing the cure rate of patients.However,due to the complex tissue structure and biological properties of the brain,it is still difficult for existing methods to accurately simulate brain tumor growth.In this thesis,we investigated how to accurately reconstruct brain structure and how to accurately simulate brain tumor growth process in brain tumor growth model,and the main research contents and contributions are as follows:1.A method of brain tissue reconstruction based on MR images is proposed for the problems of brain tissue deformation caused by brain tumor growth and large differences in brain tissue structure among different patients.The patient-specific brain tissue anatomical structure was firstly obtained by the segmentation method based on atlas alignment,followed by the approximate alignment of the deformed brain tissue using the image alignment method to obtain the non-deformed brain tissue anatomical structure,and finally the 3D structure of the patient’s brain was reconstructed based on the obtained brain tissue structure to achieve personalized modeling.The results show that the method can achieve accurate brain tissue structure reconstruction while preserving the specificity of the patient’s brain tissue and providing a realistic and accurate growth domain for the subsequent brain tumor growth simulation.2.A new brain tumor growth model is proposed to address the problem that existing models using global parameters cannot characterize the spatial changes of tumor cell proliferation and infiltration and are difficult to accurately simulate tumor growth.Based on the reaction-diffusion equation to simulate brain tumor growth process,the tissue heterogeneity was characterized by local diffusion coefficient;and two methods based on cell density gradient and inverse distance weighted interpolation were used to construct spatial parameters to characterize the spatial distribution of proliferation rate,respectively.The simulation experimental results show that the proposed brain tumor growth model with spatial parameters can simulate tumor growth more accurately than the traditional brain tumor growth model;among the brain tumor growth models with spatial parameters,the interpolation-based method is more advantageous,reduces the parameter complexity,and can simulate brain tumor growth more accurately.3.Based on the finite difference method to solve the reaction-diffusion equation,the multi-resolution adaptive mesh is introduced to improve the solution accuracy and speed.The finite difference method discretizes the computational domain into grid nodes,and solves them by replacing the partial derivatives of the grid nodes with the difference quotient approximation of the functions.The computational domain in the multi-resolution adaptive grid consists of blocks processed in parallel,and the resolution of each block adaptively changes according to the computational situation,thus accurately expressing the boundary conditions and effectively overcoming the errors caused by the finite difference method with the medium distance difference grid on the boundary.The simulation results show that the method can effectively reduce the computational errors and improve the computational efficiency.