Research On The Key Technologies Of Tracking Moving Tumor For The Stereotactic Body Radiotherapy Of Small Liver Cancer

Purpose:Induced by respiration,small liver cancer can move automatically,which deteriorates the tumor location precision and influences the application of stereotactic body radiotherapy(SBRT).This thesis proposes a technical scheme to estimate and predict the moving tumor position during radiation treatment.With a correlation model,the moving tumor position is inferred from the skin marker motion.Methods and Results:The technical scheme include three parts:(a)the calibration of the skin marker motion data,(b)the calibration and measurement of the internal target motion and(c)the design and development of the correlation model.Among them,part(a)is patient setup in radiotherapy.It keeps the patient body in the same position for each treatment,so that the signal baseline drift of the skin marker motion can be avoided as much as possible.Part(b)is to measure the small liver cancer drift and the respiration-induced tumor motion,by using a three-dimensional ultrasound technique which is based on a free-form registration.Part(c)is to relate the internal target motion and skin marker motion.According to the implementation of the above technical scheme,the study work can be classified into two aspects:1.The research,development and clinical application of the free-form registration in SBRT of small liver tumor.This aspect of work included two parts:(1)the calibration of the skin marker motion data,(2)the calibration and measurement of the internal target motion.(1)Research and implementation of a scheme,based on the free-form registration,to reduce the tumor location errors that are caused by two sources.One is the relative displacement between the patient and the immobilization system(RDBPI),and the other one is the deformation of the thermoplastic mask.The scheme to correct RDBPI was verified on 6 patients with liver cancer.The tumor location result of our method was compared with the one of the cone-beam computed tomography(CBCT).20 clinical experiments were conducted.The scheme to measure the deformation of the thermoplastic mask was verified using a simulation experiment.During verification,the actual deformation is obtained by an optically positioning device.The difference of the measured and actual deformation was defined as the error.The clinical verification showed that the scheme of correcting RDBPI attained an average liver tumor location error of 2.76mm,3.72mm and 1.80mm which were along the three orthogonal directions;and after correction,the tumor location error was reduced.Simulation verification suggested that the mean estimation error of thermoplastic mask deformation was 1.7639mm.(2)Research on the free-form registration-based ultrasound tracking method of small liver cancer,which is used to measure the liver drift(caused by the volume change of the surrounding organ)and its spontaneous motion induced by respiration.The method was verified on a phantom.The phantom was a small liver cancer which was reconstructed from CT scans.The actual drift and spontaneous motion of the small liver cancer were randomly set by a computer.The difference of the actual value and the estimated one was defined as the error.Simulation verification displayed that the proposed method measured the liver drift with an average estimation error of 1.11 mm and tracked the tumor motion with an average error of 2.94mm.2.The research and development of estimating and predicting the motion of the small liver tumor.To infer the current position(estimation)and the future position(prediction)of tumor based on the current skin marker motion,the major work was:Design and implement a model to correlate the skin marker motion and the internal tumor motion,based on a nearest neighbor estimation(NNE).In the practical application of the model,NNE doesn't update model parameters or process image in real time.This model was verified on an open database consisting of the skin and the internal target motion.Experiment results verified that(1)NNE could estimate the small liver cancer motion with a root-mean-square error(RMSE)of less than 3 mm and a computation delay of 1ms and(2)NNE could perform a prediction with a RMSE of less than 3mm and a system delay of less than 0.42ms.The estimation and prediction results both met the clinical requirement.Conclusion:This thesis researches and develops the critical techniques that are related to the challenging tasks of tumor location in liver SBRT,Verification proved that the method proposed by the thesis could guarantee a relatively high-precision small liver tumor location guidance and real-time tracking.