基于fMRI动态低频振幅应用于帕金森病辅助诊断研究

The Application of Dynamic Low Frequency Amplitude for Auxiliary Detection of Parkinson’s Disease:A Resting State fMRI Study

目的:利用帕金森病(PD)患者和健康对照(HCs)全脑动态低频振幅(dALFF),实现对PD与HCs的分类,进一步探索dALFF与PD病程的关系。方法:24例PD患者和24名HCs接受3.0T血氧水平依赖(BOLD)-功能磁共振成像(fMRI)检查。使用DPARSF软件对原始数据进行处理,并计算出dALFF。将dALFF载入PRoNTo软件包对PD及HCs进行分类计算,采用分类精度、ROC及曲线下面积(AUC)对分类结果进行评价。利用偏相关分析观察感兴趣脑区dALFF与病程的相关性。结果:全脑dALFF对PD与HCs的分类精度达93.75%(P=0.0002,置换检验5000次),分类权重最主要的脑区为双侧楔前叶、双侧小脑半球。其中,左侧楔前叶dALFF变异系数与病程呈正相关(r=0.645,P=0.001)。结论:利用dALFF能够较为理想地区分PD和HCs,为临床辅助诊断PD提供影像学参考。左侧楔前叶自发脑活动的动态变异程度可反映PD患者疾病的进程。

Purpose:To distinguish patients with Parkinson’s disease(PD)form healthy controls(HCs)using whole brain dynamic amplitude of low-frequency fluctuation(dALFF),and to further explore the correlation between dALFF and the disease duration of PD.Methods:Twenty-four patients with PD and 24 HCs underwent 3.0 T blood oxygenation level dependent(BOLD)-functional magnetic resonance imaging(fMRI)scanning.DPARSF software was employed to preprocess the original data and calculate dALFF.dALFF was loaded into the PRoNTo software package to generate classification of PD and HCs.Classification results were evaluated through accuracy,ROC curve and area under the ROC curve(AUC).Partial correlation analysis was used to observe the correlation between the dALFF and disease duration of PD.Results:The classification accuracy of dALFF for PD and HCs was 93.75%(P=0.0002,permutation tests,5000 times).The brain areas with the most important classification weights were located in the bilateral precuneus and bilateral cerebellar hemispheres.The dALFF coefficient of variation in the left precuneus was positively correlated with the disease duration(r=0.645,P=0.001).Conclusions:Patients with PD and HCs can be distinguished with high accuracy through dALFF,which could help for clinical diagnosis of PD.The degree of dynamic variation of spontaneous brain activity in the left precuneus can reflect the disease progression.