Technologies Of Respiratory Motion Prediction Based On Nonparametric Regression During Radiation Oncology

The goal of radiation therapy is to precise!y deliver a lethal dose to tumour while minimizing the dose to surrounding healthy tissues and critical structures.Recent technical developments such as intensity-modulated radiation therapy(IMRT)have advanced the capability of delivering hightly conformal radiation dose distributions to small,localized target volumes.One of the challenges in using those techniques is real-time tracking and predicting target motion,which is necessary to accommodate system latencies.For image guided radiotherapy(IMRT)systems,it is also desirable to minimize sampling rates to reduce imaging dose.Therefore the motion estimation of the tumor obj ects has to be implemented during respiration because the tumors in the pans of lung or abdomen are more impacted by respiration.This study focuses on predicting respiratory motion,which can significantly affect lung tumours.Predicting respiratory motion in real-time is challenging,due to the complexity of breathing patterns and the many sources of variability.We propose a prediction method based on nonparametric local regression.There are three major ingredients of this approach:(1) augmentation of the state space to capture system dynamics;(2)local regression in the augmented space to train the predictor from previous observation data using semi-periodicity of respiratory motion;(3)local weighting adjustment to incorporate fading temporal correlations.To evaluate prediction accuracy,we computed he root mean square error between predicted tumour motion and its observed lacation for 1 0 patients.For comparison,we investigated commonly used predi ctive methods,namely linear prediction,neural networks and Kalman filtering to the same data.The proposed method reduced the prediction error for all imaging rates and latency lengths,particularly for long prediction lengths.