In-vivo kinematic analysis of the ACL-reconstructed knee using fuzzy ROI based 2-D/3-D image matching

研究内容2-D/3-D图像匹配是膝关节韧带损伤运动学分析的尝试之一。然而，到目前为止，还没有达到很高的准确度，临床应用仍然受到限制。在本研究中，我们提出了一种基于模糊感兴趣区域(ROI)的配准技术来提高测量质量。本研究的目的是评估我们提出的方法的准确性，并评估其在前交叉韧带重建膝关节评估中的可行性。方法使用16-MDCT扫描仪获得三维多层螺旋CT图像，而使用配备宽视平板探测器的数字摄影设备获得二维图像。在基于模糊感兴趣区(ROI)的2-D/3-D图像匹配中，确定了两种类型的感兴趣区(具有特征骨块的“注意区域”和具有模糊骨骼轮廓的“禁忌区域”)。在随后的分析中，对ROI应用不同的值并计算匹配分数。在误差分析中，生成了计算机仿真图像并进行了分析。比较了传统的基于梯度的配准和基于模糊感兴趣区域的配准的计算误差。结果采用基于模糊感兴趣区域的配准方法，提高了配准精度。应用本文提出的方法计算得到的旋转误差小于0.5度。在对前交叉韧带重建膝关节和对侧完整膝关节的分析中，分别对过约束和欠约束的运动学模式进行了定性识别。讨论将模糊ROI逻辑引入到2-D/3-D图像匹配中，可以提高分析的精度和效率。对正常膝关节和重建前交叉韧带的膝关节进行运动学比较，将为评估手术质量提供有价值的信息。

INTRODUCTION2-D/3-D image matching has been investigated as one of the attempts in kinematic analysis of the knees with ligamentous injuries. To date, however, high accuracy has not been achieved, and the clinical application is still limited. In this study, we proposed a fuzzy ROI (region of interest) based registration technique to improve the measurement quality. The purpose of this study was to evaluate the accuracy of our proposed method, and to assess its feasibility in the evaluation of the ACL-reconstructed knee.METHODSThe 3-D MDCT image was obtained using a 16-MDCT scanner, while the 2-D image was obtained with a digital radiography equipment fitted with a wide-vision flat panel detector. In the fuzzy ROI (region of interest) based2-D/3-D image mating, two types of ROIs (“attention region" with characteristic bony land mark and "taboo region" with murky bony contour) were determined. In the subsequent analysis, different values were applied to the ROIs and matching score was calculated. In the error analysis, computer-simulated image was generated and analyzed. The calculated errors were compared between the conventional gradient based registration and the fuzzy ROI based registration. Subsequently, our proposed method was applied to the ACL-reconstructed knees.RESULTSApplication of the fuzzy ROI based registration resulted in improved accuracy. The resultant error in rotation was calculated to be less than 0.5 degree with the application of our proposed method. In the analysis of the ACL-reconstructed knee and the contralateral intact knee, both overconstrained and underconstrained kinematic patterns were quatitatively identified.DISCUSSIONThis study showed that introduction of fuzzy ROI logic into the 2-D/3-D image matching could improve accuracy and efficiency of the analysis. Kinematic comparison between the normal and ACL-reconstructed knees will provide valuable information in evaluation of the quality of our surgical procedure.