图像分割在肺部计数器虚拟刻度中的应用

基于边缘检测及阈值分割算法，采用Matlab编程对一物理体模的CT图片进行了自动分割，区分出肺、肌肉、骨骼3种组织，并对分割出的数据进行了三维重建，实现了体素模型的快速建立。将自动分割出的器官的体积与实际器官的比较，相对偏差小于2%。将体素模型用于肺部计数器的虚拟刻度，结果显示，虚拟刻度与实验结果的相对偏差小于10%。

Application of Image Segmentation for Virtual Calibration of Lung Counting System

A voxel phantom was established easily using Matlab software, based on the edge detection and gray threshold methods. Lung, muscle and bone were segmented from the original computerized tomography images of a physical phantom. Comparing the volume of the segmented organ and the physical organ, the relative deviation less than 2% was obtained. The segmented phantom was used for the in vivo calibration of lung counting systems. The result shows that the virtual calibration coefficient agrees with the experimental values very well, with a relative deviation less than 10%.