Regional amplitude abnormities in the major depressive disorder: A resting-state fMRI study and support vector machine analysis
Regional amplitude abnormities in the major depressive disorder: A resting-state fMRI study and support vector machine analysis
复制标题
重度抑郁症的区域振幅异常：静息态功能磁共振成像研究和支持向量机分析
DOI：
10.1016/j.jad.2022.03.079
复制
发表时间：
2022-04-09
影响因子：
6.6
通讯作者：
Lv, Zhiping
中科院分区：
文献类型：
--
作者：
Chen, Qing;Bi, Yanmeng;Lv, Zhiping
目的：抑郁症（Major Depression Disorder，MDD）是一种常见的心境障碍.然而，目前选择敏感的生物标志物和建立可靠的诊断方法仍然具有挑战性。本研究应用支持向量机（Support Vector Machine，SVM）方法研究MDD患者的自发脑电活动异常，并探讨3种异常脑电活动幅度指标的临床诊断价值。计算低频波动幅度（ALFF）、低频波动幅度分数（fALFF）和波动幅度百分比（PerAF）指标，以评估局部自发脑活动。然后我们进行相关分析，以检查脑异常和临床特征之间的关联。结果：双样本t检验显示MDD患者右侧尾状核和胼胝体的ALFF值较HC升高，fALFF值较HC升高，顶下小叶和右侧尾状核的PerAF值较HC升高。顶下小叶PerAF值与慢因子分呈负相关。SVM结果表明，选择右侧尾状核和胼胝体中的平均ALFF和fALFF作为特征的组合实现了最高的曲线下面积（AUC）值（0.89），准确度（79.79%），敏感性（65.12%）和特异性结论：平均ALFF和fALFF的增加可作为区分MDD和HC的潜在神经影像学指标。
Purpose: Major depressive disorder (MDD) is a common mood disorder. However, it still remains challenging to select sensitive biomarkers and establish reliable diagnosis methods currently. This study aimed to investigate the abnormalities of the spontaneous brain activity in the MDD and explore the clinical diagnostic value of three amplitude metrics in altered regions by applying the support vector machine (SVM) method.Methods: A total of fifty-two HCs and forty-eight MDD patients were recruited in the study. The amplitude of low frequency fluctuation (ALFF), fractional amplitude of low-frequency fluctuation (fALFF) and percent amplitude of fluctuation (PerAF) metrics were calculated to assess local spontaneous brain activity. Then we performed correlation analysis to examine the association between cerebral abnormalities and clinical characteristics. Finally, SVM analysis was applied to conduct the classification model for evaluating the diagnostic value.Results: Two-sample t -test exhibited that MDD patients had increased ALFF value in the right caudate and corpus callosum, increased fALFF value in the same regions and increased PerAF value in the inferior parietal lobule and right caudate compared to HCs. Moreover, PerAF value in the inferior parietal lobule was negatively correlated with the slow factor scores. The SVM results showed that a combination of mean ALFF and fALFF in the right caudate and corpus callosum selected as features achieved a highest area under curve (AUC) value (0.89), accuracy (79.79%), sensitivity (65.12%) and specificity (92.16%).Conclusion: Collectively, we found increased mean ALFF and fALFF may serve as a potential neuroimaging marker to discriminate MDD and HCs.