結合動態X光與立體像素骨模型量測活體三維頸椎骨運動

Abstract

人體脊椎是個複雜的結構，它保護脊髓以及傳遞頭顱與軀幹的重量至骨盆。脊椎的損傷可能傷及神經，甚至造成下半身癱瘓。因此，精確的三維動態量測人體脊椎運動學，可以對脊椎的研究、手術與治療提供相當程度的幫助。但是目前的量測方法比如皮膚標記、醫學影像或者試體實驗等，皆有其無法消除的誤差與限制。因此，本研究的目的在於結合三維電腦模型以及二維動態X光影像重建脊椎骨空間位置，以提供一準確的非侵入式的人體脊椎三維動態活體量測方法。 本研究使用雙平面動態Ｘ光量得頸關節運動影像，配合電腦斷層掃描，得到骨骼之立體模型，並且產生數位重建立體像素影像(Digitally reconstructed radiography)，並使用相似度比較動態X光影像與數位重建立體像素影像，藉由最佳化方法控制模型位置，求得頸椎骨空間位置及方向。 本研究並且使用四段新鮮相連的豬頸椎骨(C1-C7)，分別彎成不同姿態埋入水晶球並與石蠟封成剛體，進行電腦斷層掃瞄以得知水晶球與骨頭相對關係，之後以雙平面動態X光，搭配DLT(Direct linear transformation)以得到豬頸椎骨空間位置。實驗拍攝靜態影像與動態影像，另外由電腦斷層掃瞄(Computed tomography, CT)提供黃金準則(gold standard)比較得知精度完成驗證實驗。單平面與雙平面的比對精度也可以同時被本研究確立，並且測試不同最佳化手段以及解決頸椎骨在動態X光影像疊影(overlapping)問題並且改善精度。 本方法具有三維量測、活體量測、動態量測、非侵入式、準確量測並且可以避免皮膚移動誤差以及後續運算時間較短等優點。可以適用於量測人體三維頸關節運動學。 本研究量測四位正常人的頸椎骨活動，量測出在頭進行屈曲/伸展、旋轉、側彎等日常生活基本運動時的頸椎骨的運動，另外使用兩台數位攝影機量測出頭相對於軀幹的角度。並且對各節貢獻度相對於頭運動角度、其聯帶運動(coupled motion)、以及有限螺旋軸(finite helical axis)的移動軌跡製圖。 未來，希望能應用我們所發展的方法，量測不同關節及其他骨骼肌肉系統疾患，以利未來骨科、復健科、物理治療、職能治療、運動醫學、電腦輔助手術與人工關節設計等領域的進步。

The human spine is a complex structure. It protects spinal cord and transfer weight of head and trunk to pelvis. The spinal injury could affect nerve system, even cause paraplegia. Therefore, accurate 3-dimensional dynamic measurement of human spine kinematics could be helpful for spinal research, operation, and clinical management. However, present measurement techniques such as skin marker, medical imaging or in vitro studies have different kind of limitation. Therefore, the goal of this study is to combine 3-dimensional computer bone model and 2-dimensional fluoroscopic image to reconstruct the vertebra position, In order to provide an accurate, in vivo, non-invasive, 3-dimensional, dynamic measurement technique for human cervical spine. A fluoroscopy system was used to capture the cervical spine movement. Computed tomography was also used to reconstruct the bone model and digitally reconstructed radiography (DRR). Similarity measurement was used to compare the DRR and fluoroscopic image to find the model position and orientation using optimization. The validation experiment used 4 fresh connected porcine cervical spine cadavers (C1-7), bent into 4 different statuses, and fixed them with crystal balls using paraffin. The relation between crystal balls and vertebrae were defined by the computed tomography. Direct linear transformation and bi-plane fluoroscopy was used to locate the position of crystal balls and vertebrae. Static and dynamic images were captured by the bi-plane fluoroscopy. And the gold standard was provided by computed tomography. The registration accuracy of single plane and bi-plane fluoroscopy were determined. 2 optimization technique was compared and provide an solution for overlapping image. 4 normal subjects were participated in our study. The dynamic motion of flexion/extension, rotation, side bending were measured. 2 digitally cameras was also used for the movement of head. The contribution of each vertebra, coupled motion, and the finite helical axis were plotted. In the future, the method we develop will be adopted in other joint components and be extended in studying various kinds of diseases. Thus, the application of our method is helpful for the progress in the fields of orthopedic, rehabilitation, physical therapy, occupational therapy, sports medicine, computer-aid surgery, and implements designs, etc…