Takahashi Masaya, Wehrli Felix W
Harvard Medical School, Department of Radiology, Beth Israel Deaconess Medical Center.
Clin Calcium. 2004 Dec;14(12):47-54.
Recent progress in noninvasive imaging techniques--notably magnetic resonance imaging (MRI) and computed tomography (CT)--in terms of improvements in achievable signal-to-noise ratio (SNR) and spatial resolution, has led to specialized modalities (often referred to micro-CT and micro-MRI) that now enable visualization and quantitative analysis of trabecular bone architecture in vivo in animals and humans. We have during the past decade been dedicated to the development of acquisition and processing methods permitting quantitative characterization of the complex architecture of trabecular and cortical bone from micro-MRI data obtained in the limited spatial resolution regime inherent to measurements performed in vivo. Fundamental to achieving such a goal is an image acquisition strategy that ensures faithful retrieval of the structural features, which includes appropriate imaging pulse sequences and control of and correction for involuntary subject motion. Preprocessing of the raw images results in noise-free grayscale images representing maps of bone volume fraction, which are subvoxel-processed (a method for reducing partial volume blurring with application) to yield images in higher spatial resolution. From the binarized and skeletonized images the three-dimensional topology of the network can be retrieved allowing each voxel to be characterized as belonging to a surface, curve or junction between the fundamental topological types. Surfaces and curves are the lower-dimensionality counterparts of plates and rods for better differentiation of bone quality.
无创成像技术——尤其是磁共振成像(MRI)和计算机断层扫描(CT)——在可实现的信噪比(SNR)和空间分辨率方面取得的最新进展,催生了专门的成像方式(通常称为微型CT和微型MRI),现在能够在动物和人体体内可视化并定量分析小梁骨结构。在过去十年中,我们致力于开发采集和处理方法,以便根据在体内测量固有的有限空间分辨率条件下获得的微型MRI数据,对小梁骨和皮质骨的复杂结构进行定量表征。实现这一目标的基础是一种图像采集策略,该策略可确保忠实地获取结构特征,这包括适当的成像脉冲序列以及对受试者非自愿运动的控制和校正。对原始图像进行预处理可得到表示骨体积分数图的无噪声灰度图像,然后对其进行亚体素处理(一种减少部分容积模糊的应用方法),以生成更高空间分辨率的图像。从二值化和骨架化图像中可以获取网络的三维拓扑结构,从而可以将每个体素表征为属于基本拓扑类型之间的表面、曲线或交点。表面和曲线是板和杆的低维对应物,以便更好地区分骨质。