透视立体影像分析技术在假体无菌性松动检测中的应用研究

Application of fluoroscopic stereophotogrammetric analysis in the detection of aseptic loosening of prostheses

目的·验证基于2D‒3D配准的透视立体影像分析(fluoroscopic stereophotogrammetric analysis,FSA)技术在关节假体无菌性松动早期迁移检测的精度及临床应用可行性。方法·通过图像合成实验验证以光源为中心和以投影物体为中心的二维(two dimension,2D)‒三维(three dimension,3D)配准算法在不同工况下FSA技术的精度,通过真实模型实验验证临床应用可行性。图像合成实验通过在虚拟环境下建立与真实环境参数相同的透视投影,记录3D模型(骨或假体)在六自由度变换时的2D透视图像,使用不同的2D‒3D配准算法还原3D模型的六自由度变换,计算配准算法的误差。真实模型验证通过骨‒假体高精度迁移模拟器模拟关节置换术后骨与假体之间的迁移,利用计算机断层成像及光学扫描重建骨或假体的3D模型,使用透视设备拍摄假体迁移前后的2D透视图像,再通过基于2D‒3D配准的FSA技术还原假体的迁移,计算FSA技术的误差。结果·以光源为中心的2D‒3D配准算法在不同工况下的精度均高于以投影物体为中心的算法。配准初始条件较好时,以光源为中心的算法与以投影物体为中心的算法相比:旋转误差降低,差异存在统计学意义(P=0.021);位移误差降低,差异存在统计学意义(P=0.000)。并且,以光源为中心的算法满足临床应用要求所需的相似度更低、配准次数更少。结论·基于2D‒3D配准的FSA技术在人工关节假体早期迁移检测中精度满足临床应用要求。该技术可通过检测关节置换术后假体的早期迁移预警假体的晚期无菌性松动,有望通过进一步研究应用于临床。

Objective·To verify the accuracy and clinical feasibility of fluoroscopic stereophotogrammetric analysis(FSA)technology based on two dimension(2D)-three dimension(3D)registration for early migration detection of aseptic loosening of joint prostheses.Methods·2D-3D registration algorithms centering on the light source and projected object respectively in FSA technology were verified under various working conditions through image synthesis experiments,and the feasibility of clinical application was verified through real model experiments.The image synthesis experiment established a perspective projection environment with the same parameters as the real environment in a virtual environment,the 2D perspective images of the 3D model(bone or prosthesis)during the six degrees of freedom transformation were recorded,and the six degrees of freedom transformation of the 3D model was restored by using different 2D-3D registration algorithms.The error of each registration algorithm was calculated.For real model validation,the migration between bone and prosthesis after joint replacement surgery was simulated with a high precision bone prosthesis migration simulator.The 3D model of the bone or prosthesis was reconstructed by using computed tomograph(CT)images and optical scanning,and the 2D perspective images before and after prosthesis migration were captured by using a fluoroscopy device.The migration of the prosthesis was restored by using FSA technology based on 2D-3D registration,and the error of FSA technology was calculated.Results·The accuracy of the 2D-3D registration algorithm centering on the light source was higher than that of the algorithm centering on the projected object under different working conditions.When the initial registration conditions were favorable,the algorithm centering on the light source reduced the rotation error compared to the algorithm centering on the projected object,with a statistical difference(P=0.021),and the displacement error decreases,with a significant statistical difference(P=0.000).Moreover,algorithms centering on the light sources required lower similarity and fewer registration times to meet clinical application requirements.Conclusion·The accuracy of FSA technology based on 2D-3D registration in early migration detection of artificial joint prostheses meets clinical application requirements.This technology can warn of late aseptic loosening of prostheses by detecting early migration of prostheses after joint replacement surgery,and is expected to be applied to clinical practice through further research.