【摘要】 目的评价骨质疏松高分辨MR图像处理及其应用价值。方法本组12例,其中女7例,男5例,年龄25-80岁,平均54.8岁。经双能X线骨密度仪测量(dual-energyX-rayabsorptiometry,DXA),分为骨松组和对照组(骨密度正常),每组6例。所有病例均行高分辨MR扫描。SIEMENS1.5T超导MR,3D-DESS序列,层厚0.73mm,像素大小312×312×730μm。扫描部位左跟骨,矢状面,扫描时间为11min。数据用共生矩阵(co-occurrenceMatrix)和局部灰度差分矩阵(neighborhoodgray-tonedifferencematrix,NGTDM)对MR图像进行了纹理分析,提取了熵、粗糙度等纹理特征,并对MR图像进行二值化,分离了骨小梁和骨髓。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,欧拉数等数据,对骨松组和对照组进行统计学处理。结果两组数据在15个结构参数中,有11个参数有显著性差异,占73%。结论HR-MR二值化图像技术能区分骨松组和正常对照组。为HR-MR研究骨质疏松小梁微结构起到积极作用。部分参数的意义还有待进一步研究。更多还原

【Abstract】 Objective To assess osteoporosis with HR MRI. Methods There were 12 patients including seven females and five males in The group, The mean age was 54.8 years (25-80 years). Bone densitometry was measured by dual-energy X-ray absorptiometry both at the lumbar spine and thehe group nondominant femoral neck. All the patients were divided into the osteoporosis group and the control group (BMD normal) according to their bone status. Every group consisted of six patients. All underwent MRI with 1.5T super conductive MRI, left calcaneus sagittal scan, 3D-DESS sequence, slice thick 0.73 mm. pixel size: 312×312×730μm. Time acquirment was 11 min. Texture features (entropy, coarse, etc...) derived from co-occurrence matrix and neighborhood gray-tone difference matrix were obtained. These features were significantly different between patients and control subjects. Then the gray scale MR image was transformed to binary image, The shape and topology features (area, skeleton length, euler number, etc...) were obtained from the binary image. Results The 11of 15 (73%) parameters from the patients and the control showed significant difference. Conclusions The parameters of the binary image from HR MRI can distinguish osteoporosis from control group. It is valuable for trabecular micro-architecture in osteoporosis. Of course, the significance of the part numbers will be discussed in the future.更多还原