Real-time prediction of respiratory motion using a cascade structure of an extended Kalman filter and support vector regression

S-M Hong, W Bukhari
  • Physics in Medicine and Biology, June 2014, Institute of Physics Publishing
  • DOI: 10.1088/0031-9155/59/13/3555

Hybrid of Kalman filter and support vector regression for time series prediction

What is it about?

In this article, a new algorithm for prediction of respiratory motion is proposed. The proposed algorithm is a hybrid of a model-bassed Kalman filter and learning-based support vector regression. The algorithm exploit the benefit of both the approaches and yield better performance than individual algorithms.

Why is it important?

The proposed algorithm is applied for the prediction of respiratory motion. The results of our algorithm are compared with state-or-the-art artificial neural networks (ANN) and support vector regression (SVR). Our proposed algorithm significantly outperform both the ANN as well SVR algorithm for the prediction of respiratory motion.

Read Publication

http://dx.doi.org/10.1088/0031-9155/59/13/3555

The following have contributed to this page: Waqas Bukhari