关键词: 肺, 肿瘤, CT图像, 分割, 统计跳变回归分析

Abstract: The segmentation of tumor is an important step for automatic locating of tumor in the applications of radiotherapy for lung cancer.With the existence of small bronchus，vessels and textures in lung area，it is a challenge to segment tumor from the background.Conventional methods of filtering cannot restrain the details within the lung area without blurring the boundary of lung and tumor.Using a method based on statistical jump regression analysis which is mentioned in this paper as a preprocessing on CT image，small bronchus，vessels and textures can be restrained.Additionally，the method has an ideal feature of edge preservation，which is advantageous for further segmentation of lung and tumor.Experiment results show that this method is effective.

Key words: lung, tumor, CT image, segmentation, statistical jump regression analysis