常规磁共振影像组学诊断早期膝关节退变软骨损伤的价值

Value of conventional magnetic resonance imaging radiomics in the diagnosis of early degenerative knee joint with cartilage injury

目的探究基于常规磁共振影像组学特征诊断早期膝关节退变软骨损伤的价值。方法收集2021年3月至2022年6月皖南医学院弋矶山医院50例经膝关节镜检查诊断为早期膝关节退变软骨损伤的患者为观察组,30例性别、年龄匹配的健康志愿者为对照组。所有受试者均行膝关节常规MRI扫描。将所有患者按照7∶3的比例完全随机地分为训练集(56例)和验证集(24例)。使用ITK-SNAP软件,分别在矢状位T1WI及矢状位脂肪抑制PDWI图像的所有层面沿关节软骨边缘勾画感兴趣区(ROI),生成三维ROI,导入AK软件中提取纹理特征。采用最大相关最小冗余(mRMR)和最小绝对收缩和选择算子(LASSO)对提取的纹理特征降维,建立矢状位T1WI、矢状位脂肪抑制PDWI及联合序列的影像组学标签并计算获得分值(rad-score),采用100次留组交叉验证(LGOCV)验证模型的可靠性。采用多因素Logistic回归分析将临床特征中差异有统计学意义的变量建立常规模型,基于常规模型及诊断效能最高的影像组学模型建立综合诊断模型。绘制受试者工作特征(ROC)曲线评价模型的诊断效能。结果在训练集中,常规模型的BMI为诊断早期膝关节退变软骨损伤的独立因子(OR=1.338,95%CI:1.055~1.696,P=0.016),常规模型在训练集和验证集中的曲线下面积(AUC)分别为0.78、0.76。矢状位T1WI、矢状位脂肪抑制PDWI及联合序列图像中提取的纹理特征经降维后各获得10、9、10个最优特征,联合序列影像组学模型的诊断效能高于各单序列影像组学模型,联合序列影像组学模型在训练集和验证集中的AUC分别为0.86和0.91。训练集中,综合诊断模型的联合序列影像组学标签(OR=2.275,95%CI:1.350~3.834,P=0.002)为独立因子,综合诊断模型在训练集和验证集中的AUC分别为0.90和0.91。综合诊断模型对于早期膝关节退变软骨损伤的诊断效能优于常规模型(P=0.008)。结论常规磁共振影像组学模型诊断早期膝关节退变软骨损伤具有较高的价值。

Objective To explore the value of conventional magnetic resonance imaging(MRI)radiomics in the diagnosis of early degenerative knee joint with cartilage injury.Methods This study selected 50 patients with early degenerative knee joint and cartilage injury diagnosed by knee arthroscopy in Yijishan Hospital of Wannan Medical College from March 2021 to June 2022 as the observed group and 30 healthy volunteers incorresponding gender and age as the control group.All subjects underwent routine MRI scan of knee joint.According to the proportion of 7∶3,all patients were randomly divided into a training set(n=56)and a verification set(n=24).ITK-SNAP software was adopted to delineate the regions of interest(ROI)along the edge of articular cartilage at all levels of sagittal T1WI(T1 weighed images)and sagittal fat-suppressed PDWI(proton density-weighted images).3D ROI was generated and imported into AK software to extract texture features.mRMR(Maximum relevance minimum redundancy)and LASSO(least absolute shrinkage and selection operator)were used to reduce the dimension of the texture features.This research moved on to establish the radiomic tags of sagittal T1WI,sagittal fat-suppressed PDWI and the combined sequenceand to calculate the rad-score.The 100-timesleave-group-out cross-validation(LGOCV)was employed to verify the reliability of the model.Multivariate Logistic regression analysis was conducted to establish a routine model for the variables on clinical features with statistically significant differences.A comprehensive diagnostic model was established based on the conventional model and the radiomic model of the highest diagnostic efficiency.Receiver operating characteristic(ROC)curve was drawn to evaluate the diagnostic efficiency of the model.Results In the training set,BMI of the conventional model was an independent factor for the diagnosis of early degenerative knee joint with cartilage injury(OR=1.338,95%CI:1.055~1.696,P=0.016).The AUC of the conventional model in the training set and the verification set were 0.78 and 0.76,respectively.Texture features extracted from sagittal T1WI,sagittal fat-suppressed PDWI and their combined-sequence images obtained 10,9 and 10 optimal features,respectively,after dimension reduction.The diagnostic efficiency of the combined-sequence imaging model was higher than that of the single-sequence imaging model.The AUC of the combined-sequence radiomic model in the training set and validation set were 0.86 and 0.91,respectively.In the training set,the combined-sequence radiomic tag of the comprehensive diagnosis model(OR=2.275,95%CI:1.350~3.834,P=0.002)was the independent factor;the AUC of the comprehensive diagnosis model in the training set and the validation set were 0.90 and 0.91,respectively.The diagnostic efficiency of the comprehensive diagnosis model was better than that of the conventional model for early degenerative knee joint with cartilage injury(P=0.008).Conclusion The conventional MRIradiomic model is of high value in the diagnosis of early degenerative knee joint with cartilage injury.