构建“等效模型”快速验证个性化骨模型的精确度

Establishing equivalent model to verify the precision of personalized bone model rapidly

背景:目前个性化骨模型精确度的验证通常是通过配对t检验或组内相关系数等方法来进行,但这类方法往往需要制作大批量模型,不能满足个性化模型即时使用的需求。目的:探讨构建“等效模型”快速验证个性化骨模型精确度的可行性。方法:随机获取3例成年人骨骼CT影像并进行重建,运用3D打印制作个性化骨模型,再将个性化骨模型进行CT扫描并重建,使用Mimics对比重建模型和骨骼CT影像。运用Geomagic Studio对个性化骨模型CT影像的重建模型与骨骼CT影像的重建模型进行拟合偏差分析。对照骨骼CT影像的重建模型上所标注的测量位置和尺寸,对3D打印的个性化骨模型进行实物测量,并计算误差。结果与结论:①通过对比骨骼CT影像的重建模型与骨骼CT图像,二者在解剖结构和形态上相符,轮廓几乎重合;②通过拟合偏差分析,标准偏差依次为0.176,0.226,0.143 mm,所有结果均<0.25 mm;③通过实物测量并计算,相对误差均数依次为0.44%,0.21%,0.13%,所有结果均在5%的误差范围之内;④通过骨骼CT影像重建的模型符合制作个性化骨模型的基本条件;所构建的等效模型是符合临床需求和设计要求的,使用等效模型方法快速验证个性化骨模型精确度是可行的;⑤此方法可为个性化骨模型精确度的验证提供针对性强、验证迅速的途径,与常规配对t检验或组内相关系数等方法相比可缩短验证时间,能够达到提供临床即时使用的目的。

BACKGROUND:Currently,the verification of the precision of personalized bone models is usually performed by methods such as paired t-tests or intraclass correlation coefficient,but such methods often require the production of large batches of models,which do not satisfy the need for immediate use of personalized models.OBJECTIVE:To study the feasibility of establishing the equivalent model to verify the precision of the personalized bone model rapidly.METHODS:Bone CT images of three adults were randomly obtained for reconstruction.3D printing was used to create personalized bone models,and then the personalized bone models were scanned using CT and reconstructed.Mimics was used to compare the reconstructed models of bone CT images with the bone CT images.Geomagic Studio was used to analyze the fitting deviation between the reconstruction model of personalized bone model CT image and the reconstruction model of skeletal CT image.The 3D-printed personalized bone model was measured against the measurement positions and dimensions marked on the reconstruction model of skeletal CT image,and the error was calculated.RESULTS AND CONCLUSION:(1)By comparing the reconstructed bone CT image model with the bone CT scan image,the two were compatible in terms of anatomical structure and morphology,and the contours almost overlapped.(2)By fitting bias analysis,the standard bias was 0.176,0.226,and 0.143 mm in order,and all the results were<0.25 mm.(3)By measuring and calculating the model,the mean relative errors were 0.44%,0.21%,and 0.13%,and all the results were within 5%error.(4)The constructed equivalent model was in line with the basic conditions for making personalized bone models.The established equivalent model met the clinical needs and design requirements,and it was feasible to use the method of the equivalent model to verify the precision of the personalized bone model quickly.(5)This method could provide a targeted and rapid way to verify the precision of personalized bone models.It could achieve the goal of providing immediate clinical use without the need to produce large batches of models compared to conventional methods such as paired t-tests or intraclass correlation coefficient.