Study On The Respiratory Tumor Movement Tracking Method For Radiosurgery Robot

Radiation therapy is currently one of the main methods to treat lung cancer.Its main purpose is to use high dose ray to radiate the target lesion with multi-angle,while avoiding damage to normal tissues surrounding tumor,and reduce the incidence of complications while kill cancer cells.But the tumor location changes caused by human respiration bring great difficulty to precision radiotherapy,and the impact of respiratory movement persist in the entire treatment process.In order to compensate for the disturbance caused by respiratory movement on tumor location and improve the precision of radiation treatment,at present the most effective method is to treat lung tumor using synchronous respiratory tracking technology.This technology can establish the internal and external respiratory movement correlation model,and then obtain the movement position of the target by the movement information of the external markers,so as to adjust the treatment beam to always follow the tumor target,and finally realize real-time tracking and precise treatment of tumor.This paper is aimed at radiotherapy robot for tumor movement tracking method research and experiment.The tumor tracking method based on respiratory motion model can predict tumor position at any moment by using the movement data of the markers,so as to achieve real-time tracking and treatment of tumor.Concerning the traditional modeling method does not consider the sensor noise and model uncertainty,this study put forward a modeling method based on UT(Unscented Transformation)to build correlation model,and applied this method to various prediction algorithms,the validity of UT-based modeling is verified by comparing the prediction errors.At the same time,this paper designed a respiratory motion simulation platform for radiotherapy robot for the treatment of lung cancer,and based on which studies and experiments concerning tumor motion tracking methods were carried out.The designed 3D respiratory motion simulation platform can be used to simulate the association between internal markers and external tumor,and using the data provided by the platform can be effective to verify this tumor tracking methods.The main research contents in the paper are listed as follows:(1)Analyzing the movement relationship between tumor and markers in the body during free breathing,studying exhaling and inhaling phase partition method based on the PCA principle component analysis method,putting forward a modeling method to establish correlation model based on UT transformation in view of the traditional modeling method does not consider the sensor noise and model uncertainty,and establishing the association model of probability optimization,based on which tumor tracking prediction algorithm was investigated.(2)Tumor motion prediction methods are taken into research,including extended kalman filter algorithm,unscented kalman filter algorithm and particle filter algorithm.Traditional correlation model and the correlation model established based on UT transformation were applied to these three algorithms respectively.The effectiveness of the model established based on UT transformation is verified by comparing prediction error.(3)The research was carried out to study the principle of the real-time tracking radiation treatment of tumor for radiotherapy robot and analyze the characteristic of the human respiratory movement.Independently designed a three-dimensional respiratory motion simulator that can simulate the movement correlation between in vivo tumor and markers on chest,abdomen and body surface during radiotherapy.This simulator can represent real human breathing movement,and provide data source and verification platform for tumor respiratory tracking model of radiosurgery robot.(4)Based on the open database and respiratory motion simulation platform study respiratory tracking algorithm,using public data in the database and collect data from platform to verify the validity of the new modeling method and precision of the tumor tracking method.The results of public database verification and platform experiments show that the tracking algorithm based on UT transformation modeling has higher prediction accuracy than the tracking algorithm based on traditional model.