介绍用共生矩阵和局部灰度差分矩阵对MR图像进行纹理分析,提取熵、粗糙度等纹理特征,对比骨质疏松图像和正常图像,有显著的差别。又对MR图像进行二值化处理,分离骨小梁和骨髓。在此基础上对骨小梁的形态和网络结构进行分析,提取面积、骨架长度和欧拉数等特征参数,对比骨质疏松图像和正常图像,也获得显著的差别。 Introduction Co-occurrence matrix and local grayscale difference matrix are used to analyze the texture of MR images, extract texture features such as entropy and roughness, and compare the difference between osteoporosis images and normal images. The MR images were also binarized to separate trabecular bone and bone marrow. On this basis, the morphology and network structure of trabecular bone were analyzed, the characteristic parameters such as area, length of skeleton and Euler number were extracted, and the differences between osteoporosis images and normal images were also obtained.