论文部分内容阅读
本文尝试通过脑电信号检测方法辅助多动症儿童进行临床个体化诊断。首先基于一种经典的干扰控制试验任务Simon-spatial Stroop范例采集14名多动症儿童和16名正常儿童的脑电数据,并完成滤波、分段、去伪迹等预处理;然后采用主成分分析(PCA)进行电极优化选择,分别选取每种刺激模式下出现率90%以上的优化电极作为共有电极,并提取共有电极潜伏期(200~450ms)波幅的均值特征;最后采用基于欧氏距离的k-最近邻(KNN)和基于径向基核函数的支持向量机(SVM)分类器来分类。实验发现同种试验任务中多动症儿童比正常儿童表现出更低的反应正确率和更长的反应时间;多动症儿童与正常儿童的前额叶优化电极均出现N2,顶枕叶均有P2出现,且多动症儿童的峰值更低;在该实验中KNN分类准确率高于SVM分类器,StI刺激模式下KNN分类器的最高分类准确率为89.29%。以上结果说明,干扰控制试验中多动症儿童与正常儿童的前额叶及顶枕叶的脑电信号存在差异,该结果可为多动症个体的脑电信号临床诊断提供一定科学依据。
This article attempts to assist the diagnosis of ADHD children by EEG detection method. Firstly, the EEG data of 14 ADHD children and 16 normal children were collected based on a classic example of Simon-spatial Stroop for interference control experiment tasks. Filtering, segmentation and artifact pretreatment were performed. Then, principal component analysis PCA) were used to optimize the electrode selection. The optimal electrode with 90% or more of each stimulation mode was selected as the common electrode and the mean value of the amplitude of the common electrode latency (200 ~ 450ms) was extracted. Finally, the Euclidean distance k- Nearest neighbor (KNN) and support vector machine (SVM) classifier based on radial basis function. Experiments found that children with ADHD in the same kind of test task showed lower reaction accuracy and longer reaction time than normal children. N2 in the prefrontal lobe optimized electrodes of ADHD children and normal children, P2 in the top occipital lobe, and The peak value of ADN children was lower. In this experiment, the accuracy of KNN classification was higher than that of SVM classifier. The highest classification accuracy of KNN classifier was 89.29% under StI stimulation model. The above results indicate that there is a difference in EEG signals between ADHD and normal occipital lobe and occipital lobe in ADHD patients. This result may provide a scientific basis for the clinical diagnosis of ADHD.