Correlation Model And Experimental Study Of Prostate Movement Caused By Lung Respiration

Radiotherapy is an effective method for the treatment of prostate cancer at present.However,some physiological factors(such as breathing,coughing,etc.)in the process of surgery can lead to changes in the location of tumors,making the radiation emitted by radiation sources unabl e to accurately kill tumor cells.In this paper,aiming at the problem of uncertainty in radiotherapy for prostate cancer,the human lung breathing movement and prostate movement were studied,and a correlation model was established to lay a certain theore tical foundation for the dynamic perception of the prostate.Firstly,according to the performance of the analyzed sensors,this chapter establishes a set of pulmonary respiratory motion measurement system based on acceleration sensors.In order to reduce the influence of interference signals generated during detection,the sampling data was smoothed by the five-point moving average method.Then the smoothed data was analyzed in the frequency domain by Fourier transform,and the quadratic integration was us ed to obtain the displacement data of lung breathing movement.Secondly,the obtained displacement data of lung breathing motion were predicted.The two models of ARIMA(Autoregressive Integrated Moving Average model)and the fuzzy information granulation prediction model based on SVM(Support Vector Machine)were compared to predict the movement of the body surface,and their prediction errors are compared.The simulation results show that the root mean square error of the fuzzy information granulation mod el based on SVM is significantly smaller than the two models of ARIMA.Then,a set of prostate motion simulation measurement experiment system was set up.The simulated acceleration motion information of the prostate was collected by the acceleration sensor,and then converted into displacement motion information;introduced from three levels of deformation model,gray-based measurement and optimization algorithm Non-rigid registration of images;BP(Back Propagation)neural network algorithm is introduced,and a correlation model is established based on the BP neural network algori thm;at the same time,the acceleration warning threshold coefficient is defined.When the acceleration is abruptly changed due to coughing or sneezing,it will output an alarm signal to control the puncture needle mechanism to stop the needle feeding.Finally,the correlation model based on BP neural network is verified,and the motion data of lung breathing predicted by the fuzzy information granulation model based on SVM is used as the input test data of BP neural network to achieve the combination of the two models,and analyze the impact of single and double hidden layers on the accuracy of the two models.The results show that the model has the smallest root mean square error combined with the fuzzy information granulation model based on SVM and the do uble hidden layer BP neural network model.This model can be used as a clinical model of prostate radiotherapy to achieve dynamic perception of the prostate during prostate radiotherapy surgery.