DTI Image Registration Based On Active Demons Algorithm With Full Variable Mean Gradient Residual Term

In recent years,the number of people suffering from Alzheimer’s disease（AD）has been on the rise.At present,there are about 50 million AD patients in the world,and China accounts for about 25%of them.Clinical symptoms of AD patients are memory impairment,aphasia,etc.,which brings heavy psychological impact and economic burden to the families of the patients,which brings heavy psychological impact and economic burden to the families of the patients.Diffusion tensor imaging（DTI）,which is based on the diffusion degree of water molecules,can effectively reveal the structural changes of the brain white matter in AD patients by tracing the fiber bundles.It is of great value for the early prevention and diagnosis of AD patients.DTI image registration is the most important prerequisite for fiber bundle tracking based on DTI image.Active Demons algorithm was widely used in image registration because of its complete mathematical theory.The DTI multi-channel Image Registration based on variable parameter Active demons algorithm is realized in this paper,but its convergence speed is slow and the registration accuracy is insufficient.In order to improve the convergence speed and registration accuracy of the Algorithm,the Active demons algorithm with variable inertia coefficient is proposed in this paper.The inertia coefficient b is added to the iterative formula₉)of the driving force,and then the value of b is adjusted dynamically,the convergence speed and registration accuracy of the algorithm are improved.On this basis,it is found that when both the equilibrium coefficient and the inertia coefficient are changed,value of coefficient of a single homogenization can’t juggle registration deformation caused by the size of the registration accuracy is lower.Aiming to reduce the impact of this problem on the registration results,an Active demons algorithm based on full variable is proposed,balance coefficient and inertia coefficient would take more than one variable value known as the state variable method,to further improve the registration accuracy.However,when the gray gradient information of the image tends to 0,the image gradient information will be missing and then the intensity of the driving force will be weakened.Subscribe to solve the problem of missing gradient information.This article will be more than the average gradient Active demons algorithm is the driving force of the introduction of variable inertia coefficient formula in the gradient information of image registration.In this paper,the average gradient residual term was introduced into the driving force formula of Active Demons algorithm with variable inertia coefficient to supplement the gradient information of image registration,and an Active Demons algorithm based on the full variable average gradient residual term was proposed to further improve the registration accuracy.In this paper,the Active demons algorithm,MSE extremum method.Changing parameter Active demons method,increasing learning factor PSO method based on Active demons.Active demons variable inertia coefficient algorithm and based on all variables than the average gradient algorithm of Active demons six algorithm to do contrast simulation experiment,the results directly revealing the the put forward from the variable than average gradient algorithm of Active demons to achieve the best MSE values.The topology of the image registration is maintained while the accuracy is improved.