结合伪彩色与上下文感知的肺癌PET图像分割算法

由于肺癌PET成像质量较低且待分割区域边界没有明显的灰度差,使得基于颜色特征的图像分割算法不能做到有效分割。本文提出了结合伪彩色与上下文感知的肺癌PET图像分割算法。首先,将原始的肺癌PET图像根据彩色查找表生成对应的伪彩图；然后,使用改进的上下文感知模型获得伪彩图对应的显著图,并采用大津法对显著图进行二值化处理,初始化显著图的分割区域；最后,使用改进的GrabCut算法迭代分割图像。算法应用于肺癌的PET图像分割。实验结果表明,本文算法有效提升肺癌PET图像的分割效率、提升分割精度,取消GrabCut算法、Snake算法的用户操作,实现图像分割自动化,具有较高的可靠性、执行效率、以及实际应用价值。

Research on Segmentation Algorithm of Lung Cancer PET Images Based on Pseudo Color and Context Aware

Because of the low quality of lung cancer PET image and no obvious gray scale difference in the region, the graph segmentation algorithm based on color feature cannot be effectively segmented. In this paper, we propose segmentation algorithm for lung cancer PET image based on pseudo color and context awareness. First, the original lung cancer PET images according to the color-lookup table to generate the corresponding pseudo color pictures; then, using improved context aware model to obtain the pseudo color pictures corresponding to the saliency map, and the Otsu algorithm for binarization processing of the saliency map, initialization of the saliency map segmentation; and finally, use improve GrabCut algorithm to segment images by iterative method. The algorithm is applied to the segmentation of lung cancer PET image. Experimental results show that the proposed algorithm can effectively improve the segmentation efficiency of lung cancer PET image, improve the segmentation accuracy, ignore the user operation of GrabCut algorithm and Snake algorithm, realize the automation of image segmentation, which has high reliability, efficiency, and practical application value.