Dunstan C R, Lauren P D, Somers N M, Evans R A
Metabolic Unit, Concord Hospital, New South Wales, Australia.
J Bone Miner Res. 1993 Feb;8(2):139-45. doi: 10.1002/jbmr.5650080204.
An automated image analysis method is described for measurement of osteoclasts and resorbing surface in calcified bone. Osteoclasts, osteoid, and mineralized bone were measured in a single section, reacted for acid phosphatase activity, and then stained with orange G and light green stain. Three images were acquired of each field with a monochromatic camera at illuminating wavelengths of 635, 540, and 480 nm (selected using a stage monochromator). These wavelengths were chosen according to the absorption spectra of the different image components to maximize absorption differences between osteoclasts (red), mineralized bone (blue/green), and osteoid (orange). These components were then discriminated according to operator-defined ranges of color density (mineralized bone) or color fraction (osteoid and osteoclasts). A gray level coded segmented image was produced, from which was determined the area, perimeter, and number of each component and the length of contact zones with the marrow and between these components. The method was evaluated by twice measuring 10 bone sections from patients with end-stage liver failure awaiting liver transplantation. The method was quite reproducible, with coefficients of variation varying between 4% for bone volume (% tissue volume) and 22% for osteoid surface (% bone surface). The sections were also measured using a previously established semiautomated method. Coefficients of variation between methods were higher varying between 4% for bone volume (% tissue volume) and 56% for osteoid volume (% bone volume). The automated method gave a substantial time saving compared to the semiautomated method. An interactive technique was used in adjacent sections to evaluate tetracycline labeling and osteoblast surfaces.
本文描述了一种用于测量钙化骨中破骨细胞和吸收表面的自动图像分析方法。在单个切片中测量破骨细胞、类骨质和矿化骨，使其进行酸性磷酸酶活性反应，然后用橙G和亮绿染色。使用单色相机在635、540和480nm的照明波长下（使用载物台单色仪选择）获取每个视野的三张图像。根据不同图像成分的吸收光谱选择这些波长，以最大化破骨细胞（红色）、矿化骨（蓝/绿色）和类骨质（橙色）之间的吸收差异。然后根据操作员定义的颜色密度范围（矿化骨）或颜色分数范围（类骨质和破骨细胞）区分这些成分。生成了一个灰度编码的分割图像，从中确定每个成分的面积、周长和数量以及与骨髓和这些成分之间接触区域的长度。通过对10例等待肝移植的终末期肝衰竭患者的骨切片进行两次测量来评估该方法。该方法具有相当高的可重复性，骨体积（%组织体积）的变异系数在4%之间，类骨质表面（%骨表面）的变异系数在22%之间。还使用先前建立的半自动方法测量切片。方法之间的变异系数更高，骨体积（%组织体积）的变异系数在4%之间，类骨质体积（%骨体积）的变异系数在56%之间。与半自动方法相比，自动方法节省了大量时间。在相邻切片中使用交互式技术来评估四环素标记和成骨细胞表面。
