IEEE Trans Neural Syst Rehabil Eng. 2019 Apr;27(4):594-602. doi: 10.1109/TNSRE.2019.2900725. Epub 2019 Feb 22.
The P300 is regarded as a psychosis endophenotype of schizophrenia and a putative biomarker of risk for schizophrenia. However, the brain activity (i.e., P300 amplitude) during tasks cannot always provide satisfying discrimination of patients with schizophrenia (SZs) from healthy controls (HCs). Spontaneous activity at rest indices the potential of the brain, such that if the task information can be efficiently processed, it provides a compensatory understanding of the cognitive deficits in SZs. In this paper, based on the resting and P300 task electroencephalogram (EEG) data sets, we constructed functional EEG networks and then extracted the inherent spatial pattern of network (SPN) features for both brain states. Finally, the combined SPN features of the rest and task networks were used to recognize SZs. The findings of this paper revealed that the combined SPN features could achieve the highest accuracy of 90.48%, with the sensitivity of 89.47%, and specificity of 91.30%. These findings consistently implied that the rest and task P300 EEGs could actually provide comprehensive information to reliably classify SZs from HCs, and the SPN is a promising tool for the clinical diagnosis of SZs.
P300 被认为是精神分裂症的精神病学表型和精神分裂症风险的潜在生物标志物。然而，任务期间的大脑活动（即 P300 幅度）并不总能令人满意地区分精神分裂症患者（SZ）和健康对照（HC）。静息状态下的自发活动反映了大脑的潜力，如果能够有效地处理任务信息，它可以为 SZ 的认知缺陷提供一种补偿性的理解。在本文中，我们基于静息和 P300 任务脑电图（EEG）数据集，构建了功能脑电图网络，然后提取了两种脑状态的固有空间模式网络（SPN）特征。最后，将静息和任务网络的联合 SPN 特征用于识别 SZ。本文的研究结果表明，联合 SPN 特征可以达到最高的准确率 90.48%，其中灵敏度为 89.47%，特异性为 91.30%。这些发现一致表明，静息和任务 P300 EEG 实际上可以提供全面的信息，从而可靠地将 SZ 与 HC 区分开来，而 SPN 是 SZ 临床诊断的一种有前途的工具。
