分享:
分享到微信朋友圈
X
临床研究
3.0 T MRI T2 mapping纹理特征在膝关节骨性关节炎软骨损伤分级中的价值
刘晓艺 蒲如剑 梁洁 鞠文萍 王现亮

Cite this article as: Liu XY, Pu RJ, Liang J, et al. The value of T2 mapping texture features of 3.0 T MRI in grading cartilage injury of knee osteoarthritis[J]. Chin J Magn Reson Imaging, 2021, 12(7): 34-38.本文引用格式:刘晓艺, 蒲如剑, 梁洁, 等. 3.0 T MRI T2 mapping纹理特征在膝关节骨性关节炎软骨损伤分级中的价值[J]. 磁共振成像, 2021, 12(7): 34-38. DOI:10.12015/issn.1674-8034.2021.07.007.


[摘要] 目的 探究基于3.0 T MRI T2 mapping的纹理特征在膝关节骨性关节炎(keen osteoarthritis,KOA)患者不同程度软骨损伤分级中的诊断性能。材料与方法 回顾性分析实验组骨性关节炎患者72个膝关节及对照组健康志愿者22个膝关节。通过矢状位T2 mapping生成T2伪彩图,在T2伪彩图中画取ROI并标记国际软骨修复学会(International Cartilage Repair Society,ICRS)分级,选取ICRS MRI分级与关节镜分级一致的201个关节面图像,采用OK软件提取、分析纹理参数。按7∶3的比例随机选取143个关节面图像作为训练集,剩余58个关节面图像作为验证集。对训练集的参数用Spearman及sbf (select by filter)进行特征过滤,用随机森林函数进行特征选择,用ctree建立模型,给出特征在鉴别正常软骨及不同软骨损伤分级中的权重。用曲线下面积(area under the curve,AUC),敏感度、特异度,准确度来评价模型预测正常软骨及不同软骨损伤分级的性能。结果 MinLocation、MaxSize及Maximun3DDiameter权重均一致较大,其中MinLocation在各损伤分级中权重均最大,超过0.75。集训集中正常软骨的AUC值为0.91,Ⅰ级损伤的AUC值为0.82,Ⅱ级损伤的AUC值为0.84,Ⅲ级损伤的AUC值为0.88;验证集中正常软骨的AUC值为0.87,Ⅰ级损伤的AUC值为0.74,Ⅱ级损伤的AUC值为0.84,Ⅲ级损伤的AUC值为0.96。AUC最高的是验证集中Ⅲ级损伤软骨,为0.96;其次是训练集中正常软骨,为0.91。无论在训练集还是验证集中都表现出了良好的预测价值。敏感度最高的是训练集中Ⅰ级损伤软骨,为0.83;特异度最高的是训练集中Ⅲ级损伤软骨,为0.98。结论 通过T2 mapping提取的纹理参数在不同软骨损伤程度中有较好的鉴别能力。
[Abstract] Objective To explore the diagnostic performance of texture analysis based on 3.0 T MRI T2 mapping in different levels of cartilage injury in knee osteoarthritis patients. Materials andMethods Retrospective analysis of experimental group keen osteoarthritis patients (KOA) 72 knee joints and control group healthy volunteers (H) 22 knee joints. Through the sagittal T2 mapping T2 pseudocolor, to draw ROI and mark ICRS grading in a T2 artifact, a total of 201 articular surface images were selected consistent with MRI ICRS grading and arthroscopic grading, OK software was used to extract and analyze texture parameters. According to the ratio of 7∶3, 143 articular surface images were randomly selected as the training set, and the remaining 58 articular surface images were used as the verification set. The parameters of the training set were filtered by Spearman and sbf (select by filter), the characteristics were selected by random forest function, and the model was established by ctree to give the weight of the characteristics in the identification of normal cartilage and different cartilage injury grades. AUC, sensitivity, specificity and accuracy were used to evaluate the performance of the model in predicting normal cartilage and different cartilage injury grades.Results MinLocation, MaxSize and Maximun3DDiameter weights are consistent, among them MinLocation was the largest weight of each damage grade, over 0.75. AUC value of normal cartilage in training set was 0.91, grade Ⅰ damage AUC 0.82, grade Ⅱ damage AUC 0.84, grade Ⅲ damage AUC 0.88; verified that the AUC value of concentrated normal cartilage was 0.87, grade Ⅰ damage AUC 0.74, grade Ⅱ damage AUC 0.84, grade Ⅲ damage AUC 0.96. AUC highest was the validation of the grade Ⅲ damage to cartilage, 0.96. The second was the training of normal cartilage, 0.91. Both in the training set and the verification set show good predictive value. The most sensitive is the injury of cartilage in training set Ⅰ, 0.83 percent. The highest specificity was the injury of cartilage in training set Ⅲ, 0.98.Conclusions Texture parameters extracted by T2 mapping have better ability to distinguish different cartilage damage.
[关键词] 骨性关节炎;纹理分析;磁共振成像;T2 mapping
[Keywords] osteoarthritis;texture analysis;magnetic resonance imaging;T2 mapping

刘晓艺 1   蒲如剑 2   梁洁 1   鞠文萍 1   王现亮 1*  

1 潍坊市人民医院放射科,潍坊 261041

2 潍坊医学院医学影像学院,潍坊 261053

王现亮,E-mail:wangxianliang2011@126.com

全体作者均声明无利益冲突。


收稿日期:2021-02-07
接受日期:2021-03-25
DOI: 10.12015/issn.1674-8034.2021.07.007
本文引用格式:刘晓艺, 蒲如剑, 梁洁, 等. 3.0 T MRI T2 mapping纹理特征在膝关节骨性关节炎软骨损伤分级中的价值[J]. 磁共振成像, 2021, 12(7): 34-38. DOI:10.12015/issn.1674-8034.2021.07.007.

       膝关节骨性关节炎(keen osteoarthritis,KOA)是引起下肢功能障碍的重要原因之一[1]。随着社会的发展,原发性KOA的发病率不断提升[2]。膝关节软骨为透明软骨,缺乏血管、神经等组织的营养支持,只能通过关节滑膜分泌的滑液获取营养,因此软骨损伤后的修复能力十分有限,如果早期诊治不及时,很容易发展至骨性关节炎阶段,最终引起关节功能障碍甚至导致肢体残疾,严重影响患者的生活质量[3, 4]。由此可见,膝关节软骨损伤严重程度的早期分级诊断是决定患者治疗及预后的重要环节,也是现阶段骨关节外科相关领域研究的热点及重点问题。磁共振成像(magnetic resonance imaging,MRI)是目前评价关节软骨损伤最敏感的无创检查,T2 mapping序列通过组织的横向磁化衰减来反映组织的结构特性,对软骨内的水分子、胶原纤维及组织各向异性的改变尤为敏感,可以在关节软骨形态学发生改变前检测到异常信号变化[5],是当前用于诊断膝关节软骨损伤分级及检测软骨术后修复情况的最佳影像学检查。医学影像纹理分析(texture analysis,TA)是通过一定图像后处理技术,分析在医学影像图像中像素或体素灰度的分布及关系,提取肉眼无法观察到的定量或定性的纹理特征,可以早期无创地明确病灶的性质以及疾病的疗效评价、预后判断等[6]

1 材料与方法

1.1 研究对象

       本实验研究对象分为KOA组及健康志愿者组。

       KOA组:选取2018年9月到2020年2月因膝关节不适等症状来我院就诊,经关节外科医生诊断为KOA并入院治疗(关节镜手术)的受检者(排除体质量指数大于28 kg/m2,有膝关节手术史、存在膝关节感染和肿瘤病史等) 54例,女性30例,男性24例,共72个膝关节。关节镜下软骨损伤的标准采用Outerbridge[7]分级(图1, 2, 3, 4)。

       健康志愿者组:选取年龄介于18~55周岁,未有膝关节不适、非从事体育相关职业、未进行膝关节手术的健康愿者11例,女性6例,男性5例,共22个膝关节。本研究经过潍坊市人民医院医学伦理委员会批准(批准文号:2021伦审批第007号),免除受试者知情同意。

图1  男,28岁,右膝关节股骨内侧髁软骨Ⅰ级损伤。T2 mapping伪彩图中大部分区域呈现黄红色色阶,局部区域表现为斑片状绿色色阶(左图,白箭);关节镜下可见软骨局部软化,无裂纹状溃疡改变(右图,黑箭)
图2  女,30岁,左膝关节股骨外侧髁软骨Ⅱ级损伤。T2 mapping伪彩图中表现为红黄色色阶部分缺失,缺失深度<50% (左图,白箭);关节镜下可见关节软骨中出现轻至中度纤维化、浅溃疡,为“鲨鱼腮”状(右图,黑箭)
图3  女,57岁,左膝关节股骨内侧髁软骨Ⅲ级损伤。T2 mapping伪彩图中表现为红黄色色阶缺失深度>50%,但未累及全层软骨(左图,白箭);关节镜下可见软骨重度纤维化,部分剥脱(右图,黑箭)
图4  女,52岁,左膝关节股骨内侧髁软骨Ⅳ级损伤。T2 mapping伪彩图中可见软骨红黄色色阶全层缺失,代之为蓝绿色,软骨下骨裸露(左图,白箭)。关节镜下示软骨全层缺失、软骨下骨暴露(右图,黑箭)
Fig. 1  Male, 28 years old, grade Ⅰ injury of medial femoral condyle cartilage in right knee joint. Most of the T2 mapping pseudocolor images showed yellowish red color scale, and the local area showed patchy green color scale (left, white arrow). No crack ulcer changes (right, black arrow).
Fig. 2  Female, 30, grade Ⅱ injury of lateral femoral condyle of left knee joint. T2 mapping partial absence of red-yellow color scale in pseudo-color image, deletion depth<50% (left, white arrow). Under the arthroscopy, mild to moderate fibrosis and shallow ulcers could be seen in the articular cartilage, which were "shark gills". (right, black arrow).
Fig. 3  Female, 57 years old, grade Ⅲ injury of medial femoral condyle cartilage of left knee joint. T2 mapping pseudocolor image showed red and yellow color scale deletion depth >50%, but not whole layer cartilage (left, white arrow); severe fibrosis of cartilage was observed under arthroscopy. Partial stripping (right, black arrow).
Fig. 4  Female, 52 years old, left knee medial femoral condyle cartilage grade Ⅳ injury. T2 mapping pseudo-color image can be seen cartilage red and yellow color scale loss, replaced by blue-green, subchondral bone exposed (left,white arrow). Arthroscopy shows total cartilage loss, subchondral bone exposure (right, black arrow).

1.2 扫描方法与技术参数

       受检者取仰卧位,膝关节自然伸直,身体长轴与检查床长轴平行,足先进。选择合适的膝关节线圈,线圈中心线大致位于受检者髌骨下缘并完全包裹整个受检膝关节。应用GE 3.0 T Discovery 750 w Silent MR仪行膝关节8通道相控常规扫描序列及8回波SE T2 mapping序列(表1)。

表1  常规序列及T2 mapping序列的扫描参数
Tab. 1  Scanning parameters of conventional and T2 mapping sequences

1.3 图像分析

       应用GE-ADW 4.6工作站,Functool 2 T2 MAP功能选项将矢状位T2 mapping原始图像生成T2伪彩图。由2名影像科骨肌组副主任医师运用Functool 2 T2 MAP功能选项自带的勾画组件,在T2伪彩图中的股骨关节面下、胫骨平台及髌骨表面3处关节软骨画取损伤范围ROI并标记国际软骨修复学会(International Cartilage Repair Societ,ICRS)[8] MRI分级(图1, 2, 3, 4),因Ⅳ级骨性关节炎中软骨全层缺失,软骨下骨裸露,无法测得软骨的纹理特征参数,本次研究对Ⅳ级损伤不做分析。选取MRI ICRS损伤分级与关节镜分级一致的201个关节面图像。

1.4 统计学分析

       采用组内相关系数(intraclass correlation coeffcient,ICC)评价2名医师测量不同损伤分级的一致性,ICC值>0.80为一致性较好。采用Omnikinetice软件在T2伪彩图标记过的ROI上进行纹理参数提取、分析,并将提取的纹理参数导入Excel表格。按照7∶3的比例,随机选取143个关节面图像作为集训集,剩余58个关节面图像作为验证集(表2)。提取的纹理特征包括:First-Order and Distribution Statistics (一阶分布统计特征)、Shape and Morphology Metrics (形态学特征)、Histogram (直方图特征)、GLCM (灰度共生矩阵特征)、GLRLM (灰度游程矩阵特征),共77个特征参数。用影像组学数据处理的软件包R软件(Version:3.5.1),对集训集中每一级损伤的ROI所提取的特征参数之间进行Spearman相关性分析,剔除相关性大于0.9的特征。然后,为进一步选择最优特征,使用caret包中sbf (select by filter)对剩下的特征参数实施过滤,特征参数选择函数为随机森林函数。模型建立使用caret包train函数中的ctree函数,给出特征在鉴别正常软骨及不同软骨损伤分级中的权重。使用pROC包来绘制受试者特征曲线(receiver operating characteristic curve,ROC)曲线。采用ROC下面积(area under the curve,AUC),敏感度(sensitivity),特异度(specificity),准确度(accuracy)来评价模型预测正常软骨及不同软骨损伤分级的性能。

表2  正常关节及不同损伤分级关节面个数(个)
Tab. 2  Number of articular surfaces of normal joints and different injury grades (n)

2 结果

2.1 2名医师的测量结果

       2名医师测量不同损伤分级的ICC值均>0.80。

2.2 特征筛选及模型建立

       在77个纹理特征参数中,保留7个较有预测价值的特征,分别是MinLocation、MaxSize、ClusterShade、HighGreyLevelRunEmphasis、Maximum3DDiameter、GreyLevelNonuniformity以及Sphericity,给出特征在鉴别正常软骨及不同软骨损伤分级中的权重(图5)。可以看出,在鉴别正常软骨及不同程度损伤软骨中,MinLocation权重最大,预测价值最高。

图5  经随机森林函数过滤后的软骨损伤特征在四组中的重要性排序。X0=正常软骨;X1=Ⅰ级损伤软骨;X2=Ⅱ级损伤软骨;X3=Ⅲ级损伤软骨。Y轴代表相应特征,X轴代表特征的相对权重
Fig. 5  Importance of cartilage injury characteristics filtered by random forest function in four groups. X0=Normal cartilage. X1=GradeⅠdamaged cartilage. X2=GradeⅡdamaged cartilage. X3=GradeⅢ damaged cartilage. The Y-axis represents the corresponding characteristics. The X-axis represents the relative weight of the features.

2.3 模型评估

       使用AUC评估影像组学模型在鉴别正常软骨及不同损伤软骨的性能(表3图6),集训集中正常软骨的AUC值为0.91,Ⅰ级损伤的AUC值为0.82,Ⅱ级损伤的AUC值为0.84,Ⅲ级损伤的AUC值为0.88;验证集中正常软骨的AUC值为0.87,Ⅰ级损伤的AUC值为0.74,Ⅱ级损伤的AUC值为0.84,Ⅲ级损伤的AUC值为0.96。AUC最高的是验证集中Ⅲ级损伤软骨,为0.96;其次是训练集中正常软骨,为0.91。无论在训练集还是验证集中都表现出了良好的预测价值。敏感度最高的是训练集中Ⅰ级损伤软骨,为0.83;特异度最高的是训练集中Ⅲ级损伤软骨,为0.98。

图6  训练集及验证集受试者工作特性曲线。0=正常膝关节软骨,1=Ⅰ级损伤软骨,2=Ⅱ级损伤软骨,3=Ⅲ级损伤软骨
Fig. 6  Working characteristic curves of subjects in training set and validation set. 0=Normal knee cartilage,1=Grade Ⅰ damaged cartilage,2=Grade Ⅱ damaged cartilage,3=Grade Ⅲ damaged cartilage.
表3  纹理分析特征影像组学模型在鉴别正常软骨及不同损伤软骨的诊断效能
Tab. 3  Diagnostic efficacy of texture analysis imaging models in differentiating normal cartilage from different damaged cartilage

3 讨论

3.1 膝关节软骨组织结构特点及分子功能

       膝关节软骨组织主要由软骨细胞和细胞外基质(extracellular matrix,ECM)[9]组成,软骨细胞是软骨组织中仅有的细胞成分,均匀嵌入ECM内并能产生ECM,仅约占关节软骨湿重的3/4[10, 11]。而ECM是关节软骨的主要成分,约占关节软骨湿重的3/4,主要包括水分子、胶原纤维、蛋白多糖聚合体等,其主要作用是把软骨细胞结合在一起,借以支撑、维持软骨组织的生理结构和功能体现。ECM中胶原纤维构成膝关节软骨的网状支架,起到承重的作用,当它发生糖化反应后有很强的吸水性,可以锁住关节软骨内的水分[12, 13]。胶原纤维构成的网状支架内镶嵌着蛋白多糖,又称为黏多糖,也是亲水分子[14]

3.2 T2 mapping纹理特征在鉴别KOA软骨损伤分级中的价值

       纹理是宏观结构下微观组织所体现的缓慢性、周期性变化的一种属性[15]。纹理特征主要运用像素、相邻空间区域的灰度分布,明显区分于颜色等图像构成特征来表现膝关节内部结构的同质性,并通过对图像中像素灰度值的局部分布特征及变化规律进行分析。建立在T2 mapping序列基础上的膝关节软骨不同损伤情况的纹理分析结果,可以阐明关节软骨损伤改变后的情况。本实验经Spearman相关性分析和随机森林函数对特征参数进行过滤筛选后,通过ctree建立模型,来自First-Order and Distribution Statistics (一阶分布统计特征)中的MinLocation在鉴别正常软骨和Ⅰ~Ⅲ级损伤软骨中权重最大,预测价值最高。表明分子量越小,像素定位越准确。正常关节软骨水分子分布均匀,发生损伤时,关节软骨肿胀,水的通透性增高,胶原纤维内的蛋白多糖侧链基团更多地暴露,引起水的吸收增多[16, 17, 18]。因此在正常软骨及损伤软骨中,水分子变化明显,代表水分子的特征参数敏感准确,在软骨损伤方面有最高的鉴别能力,可以更早地定量体现损伤的不同[19, 20]。此外,来自Shape and Morphology Metrics (形态学特征)中的MaxSize、Maximun3DDiameter在鉴别正常软骨及不同程度损伤软骨中权重也较大,也具有较好的预测价值[21, 22]。MaxSize代表病灶最大尺寸。这两个特征都与损伤范围有关,损伤范围越大,包含的基本纹理单元越多,预测性及准确度越高。基于T2 mapping序列所获纹理参数特征对膝关节软骨的生理状态和病理改变有较好的说明性。

       AUC值通常在0.5-Order and Distribution Statistics (一阶分布统计特征)中的MinLocation在鉴别正常软骨和Ⅰ-1.0之间,当AUC=0.5时,说明诊断方法完全不起作用,无诊断价值。当AUC>0.5的情况下,AUC越接近于1,表明诊断效能越好;AUC在0.5-Order and Distribution Statistics (一阶分布统计特征)中的MinLocation在鉴别正常软骨和Ⅰ-0.7时有较低准确度,AUC在0.7-Order and Distribution Statistics (一阶分布统计特征)中的MinLocation在鉴别正常软骨和Ⅰ-0.9时有一定准确度,AUC在0.9以上时有较高准确度[10]。本次实验中集训集中正常软骨的AUC值为0.91,Ⅰ级损伤的AUC值为0.82,Ⅱ级损伤的AUC值为0.84,Ⅲ级损伤的AUC值为0.88,验证集中正常软骨的AUC值为0.87,Ⅰ级损伤的AUC值为0.74,Ⅱ级损伤的AUC值为0.84,Ⅲ级损伤的AUC值为0.96;无论在训练集还是验证集中都表现出了良好的预测价值。Ⅲ级损伤的AUC值最高,为0.96;其次为正常软骨,为0.91,有较高的准确度,这个结果可以很好地阐明在正常关节软骨与Ⅲ级损伤关节软骨方面建立的模型准确度高,临床应用可信度大。

       此外,在正常软骨和Ⅰ~Ⅲ级骨损伤分级中,Sensitivity在训练集中分别为:0.75、0.83、0.44、0.33,在验证集中分别为0.77、0.71、0.47、0.30。Specificity在训练集中分别为0.88、0.68、0.90、0.98,验证集中为0.91、0.66、0.84、1.00。可看出Ⅱ级损伤和Ⅲ级损伤中敏感度较低,可能的原因是:正常膝关节软骨中ECM成分分布规律一致,胶原纤维排列规整,蛋白多糖含量适中,水分子在细胞外基质中均匀分布;而KOA患者随着关节软骨损伤分级的增加,细胞外基质ECM成分混杂,胶原纤维排列紊乱,蛋白多糖浓度减低,水的吸收增多,纹理信息不清晰导致对KOA患者鉴别的敏感性降低。

       本研究存在的不足:(1)样本例数相对较少,可能会引起结果偏倚;(2)在纹理分析中ROI的选取时,ROI的选取范围越大,所含图像信息也越多,纹理特征参数的提取相对也越准确,但实际软骨损伤范围并不一致,不能统一范围大小,因而相对主观,故需要提升ROI选取的标准化、多样化、智能化。

       总而言之,通过T2 mapping提取的纹理参数可以在不同软骨损伤程度中做出鉴别。

1
Roemer FW, Kassim JM, Guermazi A, et al. Anatomical distribution of synovitis in knee osteoarthritis and its association with joint effusion assessed on non-enhanced and contrast-enhanced MRI[J]. Osteoarthritis Cartilage, 2010, 18(10): 1269-1274. DOI: 10.1016/j.joca.2010.07.008.
2
Madelin G, Xia D, Brown R, et al. Longitudinal study of sodium MRI of articular cartilage in patients with knee osteoarthritis: initial experience with 16-month follow-up[J]. Eur Radiol, 2018, 28(1): 133-142. DOI: 10.1007/s00330-017-4956-z.
3
Tsai PH, Wong CC, Chan WP, et al. The value of MR T2* measurements in normal and osteoarthritic knee cartilage: effects of age, sex, and location[J]. Eur Radiol, 2019, 29(8): 4514-4522. DOI: 10.1007/s00330-018-5826-z.
4
Alizai H, Walter W, Khodarahmi I, et al. Cartilage imaging in osteoarthritis[J]. Semin Musculoskelet Radiol, 2019, 23(5): 569-578. DOI: 10.1055/s-0039-1695720.
5
Eagle S, Potter HG. Morphologic and quantitative magnetic resonance imaging of knee articular cartilage for the assessment of post-traumatic osteoarthritis[J]. J Orthop Res, 2017, 35(3): 412-423. DOI: 10.1002/jor.23345.
6
钟毅, 刘欣, 肖云丹, 等. 医学影像纹理分析在骨肌系统疾病中的研究进展[J]. 磁共振成像, 2020, 11(5): 394-397. DOI: 10.12015/issn.1674-8034.
Zhong Y, Liu X, Xiao YD, et al. advances in medical imaging texture analysis in diseases of the skeletal muscle system[J]. Magnetic resonance imaging, 2020, 11(5): 394-397. DOI: 10.1002/jor.23345.
7
Slattery C, Kweon CY. Classifications in brief: outerbridge classification of chondral lesions[J]. Clin Orthop Relat Res, 2018, 476(10): 2101-2104. DOI: 10.1007/s11999.0000000000000255.
8
Paatela T, Vasara A, Nurmi H, et al. Assessment of cartilage repair quality with the international cartilage repair society score and the oswestry arthroscopy score[J]. J Orthop Res, 2020, 38(3): 555-562. DOI: 10.1002/jor.24490.
9
Schenk H, Simon D, Waldenmeier L, et al. Regions at risk in the knee joint of young professional soccer players: longitudinal evaluation of early cartilage degeneration by quantitative T2 mapping in 3T MRI [J]. Cartilage, 2020. [ DOI: ]. DOI: 10.1177/1947603520924773.
10
Mittal S, Pradhan G, Singh S, et al. T1 and T2 mapping of articular cartilage and menisci in early osteoarthritis of the knee using 3-Tesla magnetic resonance imaging[J]. Pol J Radiol, 2019, 84: e549-e564. DOI: 10.5114/pjr.2019.91375.
11
de Windt TS, Welsch GH, Brittberg M, et al. Is magnetic resonance imaging reliable in predicting clinical outcome after articular cartilage repair of the knee? A systematic review and meta-analysis[J]. Am J Sports Med, 2013, 41(7): 1695-1702. DOI: 10.1177/0363546512473258.
12
Wei B, Mao F, Guo Y, et al. Using 7.0 T MRI T2 mapping to detect early changes of the cartilage matrix caused by immobilization in a rabbit model of immobilization-induced osteoarthritis[J]. Magn Reson Imaging, 2015, 33(8): 1000-1006. DOI: 10.1016/j.mri.2015.06.007.
13
Ishijima M, Watari T, Naito K, et al. Relationships between biomarkers of cartilage, bone, synovial metabolism and knee pain provide insights into the origins of pain in early knee osteoarthritis[J]. Arthritis Res Ther, 2011, 13(1): R22. DOI: 10.1186/ar3246.
14
Liebl H, Joseph G, Nevitt MC, et al. Early T2 changes predict onset of radiographic knee osteoarthritis: data from the osteoarthritis initiative[J]. Ann Rheum Dis, 2015, 74(7): 1353-1359. DOI: 10.1136/annrheumdis-2013-204157.
15
刘菲涤, 杜芳, 程海泉. 基于MRI图像纹理分析的应用及研究进展[J]. 中国医学计算机成像杂志, 2018, 24(5): 426-429. DOI: 10.3969/j.issn.1006-5741.2018.05.014.
Liu FD, Du F, Cheng HQ. The Application and research progress of texture analysis based on MRI image[J]. Chin J Med Comput Imaging24(5): 426-429. DOI: 10.3969/j.issn.1006-5741.2018.05.014.
16
Lazik-Palm A, Kraff O, Geis C, et al. Morphological imaging and T2 and T2*mapping of hip cartilage at 7 tesla MRI under the influence of intravenous gadolinium[J]. Eur Radiol, 2016, 26(11): 1-9. DOI: 10.1007/s00330-016-4247-0.
17
Cubukçu D, Ardiç F, Karabulut N, et al. Hylan G-F 20 efficacy on articular cartilage quality in patients with knee osteoarthritis: clinical and MRI assessment[J]. Clin Rheumatol, 2005, 24(4): 336-341. DOI: 10.1007/s10067-004-1043-z.
18
Hochberg MC, Guermazi A, Guehring H, et al. Effect of intra-articular sprifermin vs placebo on femorotibial joint cartilage thickness in patients with osteoarthritis: the FORWARD randomized clinical trial[J]. JAMA, 2019, 322(14): 1360-1370. DOI: 10.1001/jama.2019.14735.
19
Eagle S, Potter HG, Koff MF. Morphologic and quantitative magnetic resonance imaging of knee articular cartilage for the assessment of post-traumatic osteoarthritis[J]. J Orthop Res, 2017, 35(3): 412-423. DOI: 10.1002/jor.23345.
20
Argentieri EC, Burge AJ, Potter HG. Magnetic resonance imaging of articular cartilage within the knee[J]. J Knee Surg, 2018, 31(2): 155-165. DOI: 10.1055/s-0037-1620233.
21
Maerz T, Newton MD, Matthew HW, et al. Surface roughness and thickness analysis of contrast-enhanced articular cartilage using mesh parameterization[J]. Osteoarthritis Cartilage, 2016, 24(2): 290-298. DOI: 10.1016/j.joca.2015.09.006.
22
Pedoia V, Haefeli J, Morioka K, et al. MRI and biomechanics multidimensional data analysis reveals R2 -R1ρ as an early predictor of cartilage lesion progression in knee osteoarthritis[J]. J Magn Reson Imaging, 2018, 47(1): 78-90. DOI: 10.1002/jmri.25750.

上一篇 度中心性的静息态功能磁共振成像探讨原发性痛经患者月经期痛经的中枢机制
下一篇 基于健康人皮质脊髓束模板分析脑卒中皮质脊髓束损伤
  
诚聘英才 | 广告合作 | 免责声明 | 版权声明
联系电话:010-67113815
京ICP备19028836号-2