Department of Medical Bioengineering, Faculty of Advanced Medical Sciences, Tabriz University of Medical Sciences, Golgasht Ave, 51666, Tabriz, Iran.
Department of Psychiatry, School of Medicine, Tabriz University of Medical Sciences, Tabriz, Iran.
Phys Eng Sci Med. 2021 Sep;44(3):855-870. doi: 10.1007/s13246-021-01038-7. Epub 2021 Aug 9.
Schizophrenia is one of the serious mental disorders, which can suspend the patient from all aspects of life. In this paper we introduced a new method based on the adaptive neuro fuzzy inference system (ANFIS) to classify recorded electroencephalogram (EEG) signals from 14 schizophrenia patients and 14 age-matched control participants. Sixteen EEG channels from 19 main channels that had the most discriminatory information were selected. Possible artifacts of these channels were eliminated with the second-order Butterworth filter. Four features, Shannon entropy, spectral entropy, approximate entropy, and the absolute value of the highest slope of autoregressive coefficients (AVLSAC) were extracted from each selected EEG channel in 5 frequency sub-bands, Delta, Theta, Alpha, Beta, and Gamma. Forty-six features were introduced among the 640 possible ones, and the results included accuracies of near 100%, 98.89%, and 95.59% for classifiers of ANFIS, support vector machine (SVM), and artificial neural network (ANN), respectively. Also, our results show that channels of alpha of O1, theta and delta of Fz and F8, and gamma of Fp1 have the most discriminatory information between the two groups. The performance of our proposed model was also compared with the recently published approaches. This study led to presenting a new decision support system (DSS) that can receive a person's EEG signal and separates the schizophrenia patient and healthy subjects with high accuracy.
精神分裂症是一种严重的精神障碍，它可以使患者的生活的各个方面都受到影响。在本文中，我们介绍了一种基于自适应神经模糊推理系统（ANFIS）的新方法，用于对 14 名精神分裂症患者和 14 名年龄匹配的对照组记录的脑电图（EEG）信号进行分类。我们从具有最多判别信息的 19 个主要通道中选择了 16 个 EEG 通道。用二阶巴特沃斯滤波器消除这些通道中的可能伪影。从每个选定的 EEG 通道的 5 个频带（Delta、Theta、Alpha、Beta 和 Gamma）中提取了 4 个特征，即 Shannon 熵、谱熵、近似熵和自回归系数的绝对值最高斜率（AVLSAC）。在 640 个可能的特征中引入了 46 个特征，ANFIS、支持向量机（SVM）和人工神经网络（ANN）的分类器的准确率分别接近 100%、98.89%和 95.59%。此外，我们的结果表明，O1 的 alpha 通道、Fz 和 F8 的 theta 和 delta 通道以及 Fp1 的 gamma 通道在两组之间具有最多的判别信息。我们提出的模型的性能也与最近发表的方法进行了比较。这项研究提出了一种新的决策支持系统（DSS），它可以接收一个人的 EEG 信号，并以高精度区分精神分裂症患者和健康受试者。
