基于视觉的智能脊柱畸形测量研究方法及临床初探

目的本研究旨在研制一款无需摄片即可了解脊柱形态的可视化智能方法。方法采用“单摄像机、多视角”的系统构建方案以便后期应用于便携式设备。在原理验证实验中,利用相机测量一系列画在纸上然后贴在板上的标记物的三维坐标,验证该方法的可行性与精度。真实试验纳入了自2018年6月至2018年12月长海医院骨科就诊的11例患者,分析其临床特点、脊柱全长正侧位片以及基于该方法与Surgimap软件测量的冠状面Cobb角度数,对该方法的有效性进行评估。结果原理验证实验中,认为利用五个图像计算Cobb角为最优策略,所得XOY和YOZ平面Cobb角的误差值基本控制在1°和2°之内,精度为1°。真实试验结果表明,本方法测量冠状面Cobb角具有可信度,但具体应用时仍需进一步优化。结论该方法成本较低,便于普及与使用,无放射性危害,为今后家庭或社区式小儿脊柱形态监测,以及动态信息反馈提供客观依据,有广阔的应用前景。

A new vision-based intelligent method for spinal deformity measurement and its clinical verification

ObjectiveWe aim to find a visual and intelligent approach to measure spine morphology. MethodsThis study first proposed an image-based method, and evaluated the feasibility through a proof-of-principle experiment and a real test. The method adopts a "single camera, multi-view" system scheme for later application to portable devices. In the proof-of-principle experiment, a camera was used to measure the three-dimensional coordinates of a series of markers drawn on paper which were attached to the board, so as to verify the feasibility and accuracy of the method. The real trial collected data from 11 patients of the department of Orthopaedics, Changhai Hospital from June 2018 to December 2018. Their clinical features and full-length positive lateral radiographs were analyzed and coronary Cobb angles were measured by the method and surgimap software. The effectiveness of the method was evaluated. ResultsIn the proof-of-principle experiment, it was considered that use of five images was the optimal strategy for measuring Cobb angles, and the error values of the Cobb angles of the XOY and YOZ planes were controlled within 1° and 2° respectively, with an accuracy of 1°. The real test results showed that the method had a credibility to measure the Cobb angle of the coronal plane, but it needed further optimization for specific applications. ConclusionsCompared with X-ray, the method had similar precision and lower cost, it was easy to popularize and use as well as harmless to the human body. It provided an objective basis for future household or community-based spine monitoring. It also had dynamic information feedback, showing good application prospects.