Abstract:Radiation therapy for tumors has become a mainstream treatment option at present, and the current method used to predict tumor information is mainly to establish association models by body surface motion information and tumor motion information. Most of the human breathing patterns are divided into thoracic and abdominal respiration, so determining the human breathing pattern before establishing the correlation model is of great help for the accuracy of radiotherapy. This paper proposed a support vector machine (SVM)-based method for classifying human respiratory patterns. It can effectively distinguish the respiratory patterns of test subjects. Two depth cameras were used to collect the point cloud information of human thorax and abdomen, and Octomap was used to reconstruct the surface of thorax and abdomen. A chunking idea was proposed for the respiratory motion characterization of the body surface to make it as the training set of the SVM to bring into training and finally bring out the test set to verify the feasibility of the method. Experimental results on the collected experimental data from three test subjects showed that the proposed method can accurately and effectively discriminate human breathing patterns.

SVM Based Human Respiratory Pattern Classification Method for Stereo Radiotherapy Robot