摘要
目的探讨基于冠状动脉CT血管造影(CCTA)图像的纹理分析在鉴别冠状动脉功能性狭窄和解剖学狭窄中的可行性和诊断效能。方法回顾性分析89例经有创性冠状动脉造影(ICA)证实冠状动脉狭窄程度为30%~90%患者的CCTA图像,基于CCTA图像,应用冠状动脉血流储备分数(FFR)计算软件计算出冠状动脉CT血流储备分数(CT-FFR),CT-FFR≤0.80的冠状动脉狭窄患者为功能性狭窄组(45例),CT-FFR>0.80为解剖学狭窄组(44例),利用MaZda分析软件提取患者左心室心肌的纹理特征,建立预测模型,分析模型中的纹理特征及预测模型鉴别冠状动脉功能性狭窄和解剖学狭窄的诊断效能。结果基于CCTA图像纹理分析的预测模型的组成特征为2_S(5,5)SumEntrp、2_Variance、2_Area、5_WavEnHH_s-3和5_WavEnLL_s-1,两组患者预测模型中纹理特征参数组间比较差异均具有统计学意义(P<0.05),功能性狭窄组的2_S(5,5)SumEntrp、2_Variance、2_Area和5_WavEnLL_s-1均高于解剖学狭窄组,5_WavEnHH_s-3低于解剖学狭窄组,其中2_S(5,5)SumEntrp的曲线下面积(AUC)最大(AUC=0.780,特异度86.4%,敏感度64.4%),诊断效能最高;预测模型鉴别冠状动脉功能性狭窄和解剖学狭窄的诊断效能(AUC=0.866,特异度72.7%,敏感度84.4%)较单个特征参数更好。结论基于CCTA图像纹理分析对鉴别冠状动脉功能性狭窄和解剖学狭窄具有一定的可行性,并具有良好的诊断效能。
Objective To investigate the feasibility and diagnostic efficacy of texture analysis combined with coronary CT angiography images in identifying coronary functional and anatomical stenosis.Methods A retrospective analysis was performed in total of 89 patients who confirmed with coronary stenosis of 30%~90%by invasive coronary angiography.Coronary CT-FFR was calculated using coronary fractional flow reserve calculation software.The functional stenosis group included 45 patients with CT-FFR≤0.8,while the remaining 44 cases were divided into the anatomical stenosis group with CT-FFR>0.8.MaZda analysis software was used to extract the texture characteristics of the left ventricular myocardium,and a predictive model was established.The texture characteristics and prediction model to identify the diagnostic efficiency of coronary functional stenosis and anatomical stenosis were analyzed.Results The texture characteristics of CCTA images prediction model were:2_S(5,5)SumEntrp、2_Variance、2_Area、5_WavEnHH_s-3 and 5_WavEnLL_s-1,which all presented statistical difference between the two groups(P<0.05).2_S(5,5)SumEntrp,2_Variance,2_Area and 5_WavEnLL_s-1 in the coronary functional stenosis group were all higher than the anatomical stenosis group while 5_WavEnHH_s-3 was less than the anatomical stenosis group.2_S(5,5)SumEntrp had the highest diagnostic efficacy(AUC=0.780,specificity=86.4%,sensitivity=64.4%).And the prediction model had better diagnostic efficiency in identifying the two groups than the individual characteristic parameters(AUC=0.866,specificity=72.7%,sensitivity=84.4%).Conclusion Based on texture analysis combined with CCTA images is feasible to identify coronary functional and anatomical stenosis and has good diagnostic capability.
作者
李操
王世界
熊义林
陈静
LI Cao;WANG Shijie;XIONG Yilin(Department of Radiology,The Affiliated Hospital of Southwest Medical University,Luzhou,Sichuan Province 646000,P.R.China)
出处
《临床放射学杂志》
北大核心
2022年第8期1465-1470,共6页
Journal of Clinical Radiology
