Orthopaedic Biomechanics Laboratory, Sunnybrook Health Sciences Centre, 2075 Bayview Avenue, Toronto, ON, Canada.
Med Eng Phys. 2011 Mar;33(2):188-94. doi: 10.1016/j.medengphy.2010.09.022. Epub 2010 Oct 29.
Preclinical models of spinal metastases allow for the application of micro-image based structural assessments, however, large data sets resulting from high resolution scanning motivate a need for robust automated analysis tools. Accurate assessment of changes in vertebral architecture, however, may depend both on the resolution of images acquired and the models used to represent the structural data.
To apply a recently developed automated μCT based analysis tool to quantify the effect of diffuse metastatic disease on rat vertebral architecture at multiple resolutions. It was hypothesized that automated methods could accurately quantify differences in vertebral microstructure and that diffuse metastatic disease could be shown to have significant negative architectural effects on trabecular bone independent of stereologic model and resolution.
μCT images acquired at 14 μm(3) of healthy and metastatically involved whole lumbar rat vertebrae were analyzed at high, medium and low (8.725, 17.45, and 34.9 μm(3)) resolutions using an automated algorithm to yield micro-structural measures of the trabecular centrum and cortical shell. The images analyzed at different resolutions were obtained via up/downsampling of the acquired image data. Trabecular thickness was evaluated with the Parfitt and Hildebrand models, and anisotropy was evaluated through calculation of mean intercept length.
Significant differences in microstructural parameters measured in comparing healthy and metastatically involved vertebrae were affected by resolution, however, relative anisotropy was maintained. The Parfitt and Hilderbrand models yielded similar structural differences between healthy and metastatic vertebrae, however, the Hildebrand model was limited due to segmentation accuracy required for its automated application.
Differences in microstructural parameters generated through automated analysis at high resolution suggest that diffuse MT1 osteolytic destruction in whole rat vertebrae results primarily in loss of trabeculae in the metastatic vertebrae, as opposed to trabecular thinning. The sensitivity of the bony architectural parameters to resolution motivates the need for high resolution scanning or post-processing of images.
脊柱转移的临床前模型允许应用基于微图像的结构评估，然而，由于高分辨率扫描产生的大数据集，需要强大的自动化分析工具。然而，准确评估椎体结构的变化可能既取决于所获取图像的分辨率，也取决于用于表示结构数据的模型。
应用最近开发的基于自动 μCT 的分析工具，在多个分辨率下定量测量弥漫性转移性疾病对大鼠椎体结构的影响。假设自动化方法可以准确地量化椎体微结构的差异，并且弥漫性转移性疾病可以被证明对小梁骨具有显著的负结构影响，而与体视学模型和分辨率无关。
使用自动算法对来自健康和转移性受累的整个大鼠腰椎的 μCT 图像进行分析，以获取 14μm(3)分辨率的高、中、低（8.725、17.45 和 34.9μm(3)）分辨率的图像，以产生小梁骨中心和皮质壳的微观结构测量值。不同分辨率下分析的图像是通过对获取的图像数据进行上/下采样获得的。使用 Parfitt 和 Hildebrand 模型评估小梁厚度，通过计算平均截距长度评估各向异性。
在比较健康和转移性受累椎体时，通过分辨率评估的微结构参数存在显著差异，然而，相对各向异性得以保持。Parfitt 和 Hildebrand 模型在健康和转移性椎体之间产生了相似的结构差异，然而，由于其自动化应用所需的分割准确性，Hildebrand 模型受到限制。
通过高分辨率自动化分析生成的微结构参数差异表明，在整个大鼠椎体中，弥漫性 MT1 溶骨性破坏主要导致转移性椎体中小梁的丢失，而不是小梁变薄。骨性结构参数对分辨率的敏感性促使我们需要进行高分辨率扫描或对图像进行后处理。
