基于全脑功能连接的网络游戏障碍分类研究

Classification of Internet Gaming Disorder Using Whole Brain Resting-state Functional Connectivity

目的:利用静息态磁共振数据构建全脑功能连接网络,通过多元模式分析建立诊断模型,实现网络游戏障碍(internet gaming disorder, IGD)和正常对照组之间的分类识别。方法:采集71例IGD患者及88例正常对照的大脑静息态磁共振数据,采用功能连接分析技术构建全脑功能连接网络。将大脑功能连接作为分类特征,采用支持向量机和多种特征选择方法,探索IGD患者区别于正常人的异常网络模式。综合多种特征选择方法筛选的共有特征最终确定IGD客观识别的影像学标志。结果:基于全脑功能连接建立的分类模型准确率最高可达80.6%(敏感性为78.5%,特异性为81.2%)。用于区分IGD患者和正常对照的神经影像学标记主要位于左侧背外侧前额叶、右侧前扣带回、左侧眶额回、右侧海马旁回和双侧颞叶等负责认知控制、动机和学习记忆的脑区。结论:基于静息态全脑功能连接的诊断模型对IGD有较好的区分能力,未来可以为临床智能诊断提供补充手段。

Objective: To explore diagnostic models at an individual level to differentiate individuals with internet gaming disorder(IGD) from healthy controls with resting state functional magnetic resonance imaging(fMRI) and multivariate pattern analysis methods. Methods: Resting state fMRI data was enrolled from 71 IGD individuals and 88 healthy controls.Whole brain functional connectivity was computed and viewed as classification features. Support vector machine(SVM)combined with the frequently-used feature selection methods such as least absolute shrinkage and selection operator(Lasso), recursive feature elimination(RFE) and random forest(RF) respectively was used to classify these data. Results: The classification results demonstrated that the feature subset selected by Lasso could identify IGD effectively with a high accuracy of 80.6%(sensitivity of 78.5%, specificity of 81.2%). The robust features selected by three methods were among brain regions implicating with cognitive control, motivation, learning and memory such as left dorsolateral prefrontal lobe, right anterior cingulate gyrus, left orbitofrontal gyrus, right parahippocampal gyrus, and bilateral temporal lobes. Conclusion: The diagnostic model based on resting whole brain functional connection has a good ability to distinguish IGD, which can provide a supplementary means for clinical intelligent diagnosis.