Song Zhiwei, Li Wenjie, Bi Hui, Wang Suhong, Zou Ling
School of Information Science and Engineering, Changzhou University, Changzhou, Jiangsu 213164, P.R China;Key Laboratory of Biomedical Information Technology, Changzhou University, Changzhou, Jiangsu 213164, P.R China.
Brain Science Research Center, The Third Affiliated Hospital of Soochow University, Changzhou, Jiangsu 213003, P.R China.
Sheng Wu Yi Xue Gong Cheng Xue Za Zhi. 2020 Oct 25;37(5):749-755. doi: 10.7507/1001-5515.201912058.
Aiming at the difference between the brain networks of children with attention deficit hyperactivity disorder (ADHD) and normal children in the task-executing state, this paper conducted a comparative study using the network features of the visual function area. Functional magnetic resonance imaging (fMRI) data of 23 children with ADHD ［age: (8.27 ± 2.77) years］ and 23 normal children ［age: (8.70 ± 2.58) years］ were obtained by the visual capture paradigm when the subjects were performing the guessing task. First, fMRI data were used to build a visual area brain function network. Then, the visual area brain function network characteristic indicators including degree distribution, average shortest path, network density, aggregation coefficient, intermediary, etc. were obtained and compared with the traditional whole brain network. Finally, support vector machines (SVM) and other classifiers in the machine learning algorithm were used to classify the feature indicators to distinguish ADHD children from normal children. In this study, visual brain function network features were used for classification, with a classification accuracy of up to 96%. Compared with the traditional method of constructing a whole brain network, the accuracy was improved by about 10%. The test results show that the use of visual area brain function network analysis can better distinguish ADHD children from normal children. This method has certain help to distinguish the brain network between ADHD children and normal children, and is helpful for the auxiliary diagnosis of ADHD children.
针对注意力缺陷多动障碍（ADHD）儿童与正常儿童在任务执行状态下脑网络的差异，本文利用视觉功能区的网络特征进行了对比研究。通过视觉捕捉范式，获取了23名ADHD儿童［年龄：（8.27±2.77）岁］和23名正常儿童［年龄：（8.70±2.58）岁］在执行猜测任务时的功能磁共振成像（fMRI）数据。首先，利用fMRI数据构建视觉区脑功能网络。然后，获取包括度分布、平均最短路径、网络密度、聚集系数、中介中心性等在内的视觉区脑功能网络特征指标，并与传统的全脑网络进行比较。最后，使用机器学习算法中的支持向量机（SVM）等分类器对特征指标进行分类，以区分ADHD儿童和正常儿童。在本研究中，利用视觉脑功能网络特征进行分类，分类准确率高达96%。与构建全脑网络的传统方法相比，准确率提高了约10%。测试结果表明，采用视觉区脑功能网络分析能够更好地区分ADHD儿童和正常儿童。该方法对区分ADHD儿童与正常儿童的脑网络具有一定帮助，有助于ADHD儿童的辅助诊断。
