Objective： Although lithium has been a commonly prescribed neurotrophic/neuroprotective mood-stabilizing agents, its effect on spontaneous brain activity in patients with bipolar depression remains unclear. The aim o...Objective： Although lithium has been a commonly prescribed neurotrophic/neuroprotective mood-stabilizing agents, its effect on spontaneous brain activity in patients with bipolar depression remains unclear. The aim of this study is to reveal the basic mechanism underlying the pathological influences of lithium on resting-state brain function of bipolar depression patients. Methods：97 subjects including 9 bipolar depression patients with lithium treatment, 19 bipolar depression patients without lithium treatment and 69 healthy controls, were recruited to participate in this study. Amplitude of low-frequency fluctuation （ ALFF ） and fractional amplitude of low-frequency fluctuation （ fALFF） were used to capture the changes of spontane-ous brain activity among different groups. In addition, further analysis in terms of Hamilton Depression Rating Scale, the number of depressive episodes, and illness duration in pooled bipolar depression patients were conducted, which combined FLEF and fALEF to identify the basic neural features of bipolar depression patients. Results： It was observed from the imaging results that both the bipolar depression patients receiving lithium treatment and healthy control subjects showed signifi-cantly decreased ALFF/fALFF values in the right anterior cingulate cortex and right middle frontal gyrus compared to that from the bipolar depression patients without lithium treatmetn. The ALFF values of the right＆amp;nbsp;middle temporal gyrus was also found to be negative related to the number of depressive episode and the total episodes. Conclusions：Our findings suggested that the bipolar depression subjects were identified to have ab-normal ALFF/ fALFF in the corticolimbic systems, in-cluding regions like right anterior cingulate cortex, bilateral middle frontal gyrus, right orbital frontal gyrus, and right middle temporal gyrus. In addition, it was also revealed that the decreased ALFF/fALFF in the right anterior cingulate cortex and right middle frontal gyrus might be a biomarker that is related to the lithium effects.展开更多
本研究在睁眼(eyes-open,EO)和闭眼(eyes-closed,EC)两种静息态下提取了45位健康被试的脑功能参数比率低频振幅(fractional amplitude of low frequency fluctuation,fALFF)和局部一致性(regional homogeneity,ReHo)数据,比较并分析了...本研究在睁眼(eyes-open,EO)和闭眼(eyes-closed,EC)两种静息态下提取了45位健康被试的脑功能参数比率低频振幅(fractional amplitude of low frequency fluctuation,fALFF)和局部一致性(regional homogeneity,ReHo)数据,比较并分析了基于线性核的支持向量机(SVM)、基于RBF核的支持向量机、朴素贝叶斯、决策树、随机森林和自适应增强(Adaboost)6种机器学习方法在数据上的分类效果.实验表明,对单一特征数据分类时,朴素贝叶斯算法对fALFF数据的分类效果最好,线性核的SVM算法对ReHo数据的分类效果最好;对fALFF和ReHo数据相融合的多层次特征数据分类时,朴素贝叶斯算法的分类效果最好.此外,本研究对单一特征数据与多层次特征数据在6种机器学习方法上进行分类比较,结果表明利用多层次特征数据时,基于RBF核的SVM,朴素贝叶斯和随机森林算法的分类效果有所提升.本研究基于不同机器学习方法和不同层次特征数据的分类比较,为EO和EC静息态脑功能活动和其他脑病理的研究提供了相关的参考依据.展开更多
基金Acknowlegements ： The authors gratefully acknowledge the Beijing Normal University Imaging Center for Brain Research and Prof. Yufeng Zang for their contributions in MRI data acquisition. This study was supported by grants from the Beijing Municipal Science ＆ Technology Commission （Grant no. D121100005012002） , Beijing Municipal Administration of Hospitals Clinical Medicine Development of Special Funding Support （ Grant no. ZYLX201403） , the National Natural Science Foundation of China （ Grant no. 81471389 ） , the High level health technical personnel in Beijing （ Grant no. 2014 - 3-095）, the MYRG2014 - 00093 - FHS and MYRG 2015-00036- FHS grants from the University of Macao in Macao, and FDCT 026/2014/A1 and FDCT 025/ 2015/A1 grants from Macao government.
文摘Objective： Although lithium has been a commonly prescribed neurotrophic/neuroprotective mood-stabilizing agents, its effect on spontaneous brain activity in patients with bipolar depression remains unclear. The aim of this study is to reveal the basic mechanism underlying the pathological influences of lithium on resting-state brain function of bipolar depression patients. Methods：97 subjects including 9 bipolar depression patients with lithium treatment, 19 bipolar depression patients without lithium treatment and 69 healthy controls, were recruited to participate in this study. Amplitude of low-frequency fluctuation （ ALFF ） and fractional amplitude of low-frequency fluctuation （ fALFF） were used to capture the changes of spontane-ous brain activity among different groups. In addition, further analysis in terms of Hamilton Depression Rating Scale, the number of depressive episodes, and illness duration in pooled bipolar depression patients were conducted, which combined FLEF and fALEF to identify the basic neural features of bipolar depression patients. Results： It was observed from the imaging results that both the bipolar depression patients receiving lithium treatment and healthy control subjects showed signifi-cantly decreased ALFF/fALFF values in the right anterior cingulate cortex and right middle frontal gyrus compared to that from the bipolar depression patients without lithium treatmetn. The ALFF values of the right＆amp;nbsp;middle temporal gyrus was also found to be negative related to the number of depressive episode and the total episodes. Conclusions：Our findings suggested that the bipolar depression subjects were identified to have ab-normal ALFF/ fALFF in the corticolimbic systems, in-cluding regions like right anterior cingulate cortex, bilateral middle frontal gyrus, right orbital frontal gyrus, and right middle temporal gyrus. In addition, it was also revealed that the decreased ALFF/fALFF in the right anterior cingulate cortex and right middle frontal gyrus might be a biomarker that is related to the lithium effects.
文摘本研究在睁眼(eyes-open,EO)和闭眼(eyes-closed,EC)两种静息态下提取了45位健康被试的脑功能参数比率低频振幅(fractional amplitude of low frequency fluctuation,fALFF)和局部一致性(regional homogeneity,ReHo)数据,比较并分析了基于线性核的支持向量机(SVM)、基于RBF核的支持向量机、朴素贝叶斯、决策树、随机森林和自适应增强(Adaboost)6种机器学习方法在数据上的分类效果.实验表明,对单一特征数据分类时,朴素贝叶斯算法对fALFF数据的分类效果最好,线性核的SVM算法对ReHo数据的分类效果最好;对fALFF和ReHo数据相融合的多层次特征数据分类时,朴素贝叶斯算法的分类效果最好.此外,本研究对单一特征数据与多层次特征数据在6种机器学习方法上进行分类比较,结果表明利用多层次特征数据时,基于RBF核的SVM,朴素贝叶斯和随机森林算法的分类效果有所提升.本研究基于不同机器学习方法和不同层次特征数据的分类比较,为EO和EC静息态脑功能活动和其他脑病理的研究提供了相关的参考依据.
