【摘要】 目的 探讨磁敏感加权成像（SWI）在脑深部核团区自回归模型纹理分析在帕金森病（PD）中的应用价值。资料与方法回顾性分析2014年8月—2017年9月在徐汇区中心医院神经内科临床确诊的36例PD患者和72例健康志愿者（对照组），采用MRI对PD患者和健康志愿者进行常规MRI序列和SWI序列成像。采用MaZda软件分别手动勾画SWI图和相位图中双侧黑质致密带（SNc）、黑质网状带（SNr）、红核（RN）、壳核（PUT）、苍白球（GP）、尾状核头（CN）、丘脑（THA）、齿状核（DN）共8个感兴趣区，提取自回归模型纹理参数值（Teta 1、Teta 2、Teta 3、Teta 4及Sigma），并进行统计学分析。比较两组间SWI图和相位图中各深部核团区纹理参数的差异，采用受试者工作特征曲线评价有统计学差异核团区纹理参数值诊断PD的效能，采用Pearson线性相关分析各核团区纹理参数值与Hoehn-Yahr量表分级间的相关性。结果 PD组和对照组间相位图在RN、GP和CN区Teta 3和Teta 4(t=-1.129、1.772、-0.609、0.599、-2.018、2.568,P均<0.05)、SNc区Teta 4和Sigma参数值（t=0.723、0.948,P均<0.05）差异有统计学意义，SWI图在GP区Sigma(t=-1.889,P=0.004)、SNc区Teta 4参数值（t=-0.332,P=0.041）差异有统计学意义。受试者工作特征曲线显示，相位图SNc区Sigma值的AUC最大为0.683，其他依次为Phase＿GP＿Teta 3(0.672)、Phase＿GP＿Teta 4(0.655)、Phase＿RN＿Teta 4(0.638)、SWI＿GP＿Sigma(0.615)、Phase＿SNc＿Teta4(0.576)、Phase＿CN＿Teta 4(0.574)、Phase＿RN＿Teta 3(0.562)、Phase＿CN＿Teta 3(0.560)和SWI＿SNc＿Teta 4(0.551)。各感兴趣区纹理参数值与Hoehn-Yahr量表分级值均无相关性（P均>0.05）。结论 PD患者深部灰质核团区SWI图和相位图均存在自回归模型纹理参数值变化，且相位图应用效能更高，自回归模型纹理参数分析在PD诊断中可提供一定的信息。更多还原

【Abstract】 Purpose To explore the application value of texture analysis of autoregressive model on susceptibility weighted imaging(SWI of deep brain nucleuses in Parkinson’s disease(PD). Materials and Methods Thirty-six patients with PD and 72 healthy volunteers(control group) from August 2014 to September 2017 were retrospectively analyzed. All subjects were performed conventional MRI and SWI sequence imaging. Eight regions of interest were manually mapped by MaZda software in SWI image and phase image, including bilateral substantia nigra compact zone(SNc), substantia nigra reticular zone(SNr), red nucleus(RN), putamen(PUT), globus pallidus(GP), caudate nucleus(CN), thalamus(THA) and dentate nucleus(DN). Texture parameters(Teta 1, Teta 2, Teta 3, Teta 4 and Sigma) were extracted from the autoregression model and analyzed statistically. The differences of texture parameters of each deep nucleus in phase image and SWI image were compared between the two groups. ROC curve was used to evaluate the diagnostic efficacy of texture parameters of the nucleus regions with statistical difference in PD group. The correlation between texture parameter values of each nucleus region and Hoehn-Yahr scale grading was analyzed by Pearson linear correlation. Results There were significant differences in Teta 3 and Teta 4 in RN, GP and CN region(t=-1.129, 1.772,-0.609, 0.599,-2.018, 2.568, all P<0.05) and Teta 4 and Sigma in SNc region(t=0.723, 0.948, P<0.05) in the phase image between PD group and control group. There were significant differences in Sigma of GP region(t=-1.889, P=0.004) and in Teta 4 of SNc region(t=-0.332, P=0.041) in the SWI image between PD group and control group. ROC curve showed that the maximum AUC of Sigma in SNc region was 0.683, the others were Phase＿GP＿Teta 3(0.672), Phase＿GP＿Teta 4(0.655), Phase＿RN＿Teta 4(0.638), SWI＿GP＿Sigma(0.615), Phase＿SNc＿Teta 4(0.576), Phase＿CN＿Teta 4(0.574), Phase＿RN＿Teta 3(0.562), Phase＿CN＿Teta 3(0.560) and SWI＿SNc＿Teta 4(0.551). There was no significant correlation between texture parameters and Hoehn-Yahr grading in all regions of interest(all P>0.05).Conclusion Both SWI image and phase image of deep gray matter nucleuses in PD show changes in autoregressive model texture parameter,and the phase image have higher application efficiency. The analysis of autoregressive model texture parameters can provide certain diagnostic information in the diagnosis of PD disease.更多还原