Burckhardt Kathrin, Székely Gábor, Nötzli Hubert, Hodler Jürg, Gerber Christian
Computer Vision Laboratory, Swiss Federal Institute of Technology (ETH), Gloriastr. 35, 8092 Zürich, Switzerland.
IEEE Trans Med Imaging. 2005 May;24(5):676-88. doi: 10.1109/TMI.2005.846849.
Assessing the displacement of bony implants is an important topic in arthroplasty, particularly in total hip replacement (THR). The observation of the migration is supposed to provide an insight into the fixation of the implant. Diagnostic standard radiographs of the pelvis are an advantageous data source for this purpose. The previous methods based on these images, however, lack of a thorough consideration of their projective nature. They do, hence, not reach the desired precision, which should lie in the submillimeter range to allow a detection of migration in the first one or two years after implantation. The aim of the work presented here was, therefore, a method for measuring the distance of the artificial hip socket to the bone with an error of less than 0.5 mm. The approach has been on the one hand to define the bone-cup distance measured in the radiograph so that the variability of the intrinsic and extrinsic parameters at exposure has a minimal impact. On the other, specialized matching techniques are applied in order to optimize the localization of the necessary bony landmarks and the cup in the X-ray image. The coordinates of the bony landmarks are determined by means of a template matching algorithm. The position of the implant is estimated by intensity-based registration using the cup's CAD-model. The method was validated theoretically, experimentally, and clinically. In the clinical radiographs, the standard deviation of the migration measurements resulted to be 0.28 mm when using only natural bony landmarks. The implantation of a bony marker was found to increase the precision to a standard deviation of 0.20 mm. The interobserver variability in the two cases was estimated to be 0.11 mm and 0.04 mm.
评估骨植入物的位移是关节置换术中的一个重要课题，尤其是在全髋关节置换术（THR）中。观察植入物的移位情况有助于深入了解植入物的固定情况。骨盆的诊断性标准X线片是实现这一目的的有利数据源。然而，以往基于这些图像的方法没有充分考虑其投影特性。因此，它们无法达到所需的精度，而所需精度应在亚毫米范围内，以便在植入后的头一两年内检测到移位。因此，本文所介绍工作的目标是开发一种测量人工髋臼与骨骼之间距离的方法，误差小于0.5毫米。该方法一方面是定义在X线片中测量的骨杯距离，以使曝光时内在和外在参数的变化影响最小。另一方面，应用专门的匹配技术来优化必要的骨标志和髋臼在X线图像中的定位。骨标志的坐标通过模板匹配算法确定。植入物的位置通过使用髋臼的CAD模型基于强度的配准来估计。该方法在理论、实验和临床方面都得到了验证。在临床X线片中，仅使用自然骨标志时，移位测量的标准差为0.28毫米。发现植入骨标志物可将精度提高到标准差为0.20毫米。两种情况下观察者之间的变异性估计分别为0.11毫米和0.04毫米。
