Interictal high-density scalp EEG in focal epilepsy patients

This dataset includes de-identified interictal spike information in 20 focal epilepsy patients who became seizure-free after surgery. This dataset has been used and analyzed to study epilepsy sources and the results are reported in: Sun R, Sohrabpour A, Worrell GA, He B: “Deep Neural Networks Constrained by Neural Mass Models Improve Electrophysiological Source Imaging of Spatiotemporal Brain Dynamics.” <em>Proceedings of the National Academy of Sciences of the United States of America</em> 119.31 (2022): e2201128119. <br> Please cite the above paper if you use any data included in this dataset. Codes related to this study are also available from (https://github.com/bfinl/DeepSIF). Clinical information of the patients are described in the Supplementary Table S1 of the paper. This dataset was collected under support from the National Institutes of Health via grants NS096761 and EB021027 to Dr. Bin He and Dr. Greg Worrell. The human data were collected and de-identified as part of NIH funded research at Mayo Clinic, Rochester, overseen by Dr. Greg Worrell, and processed and organized at Dr. Bin He’s lab at Carnegie Mellon University. The data are shared for information only. The EEG recorded in this dataset follows a 10-10 system and contains 76 channels of recording. One channel is the reference channel and given that we use a common reference channel for source imaging this channel is removed from the topographical map data and the corresponding rows of the lead-field matrix for each patient, equals 75. The EEG data were filtered between 1-40 Hz and a common average reference was used to pre-process the data. <br> The data include three variables: * spike_peak_data: Contains the topographical EEG map at the peak of the spike selected in each individual patient. * patient_id: patient index corresponding to the patient ID in Supplementary Table S1 of the paper. * eloc: electrode information in EEGLAB (https://sccn.ucsd.edu/eeglab/index.php) format. <br>

本数据集包含20名术后无癫痫发作的局灶性癫痫（focal epilepsy）患者的去标识化（de-identified）发作间期棘波信息。
本数据集已被用于癫痫源相关研究，相关研究成果发表于：Sun R, Sohrabpour A, Worrell GA, He B：“受神经质量模型（neural mass models）约束的深度神经网络可改善脑时空动态的电生理源成像（electrophysiological source imaging）”，《美国国家科学院院刊》119卷31期（2022年）：e2201128119。
若使用本数据集内的任何数据，请引用上述论文。本研究相关代码可从https://github.com/bfinl/DeepSIF获取。患者的临床信息详见论文的补充表S1。
本数据集的采集受美国国立卫生研究院（National Institutes of Health, NIH）资助，项目编号为NS096761与EB021027，资助方为Bin He博士与Greg Worrell博士。本研究的人类数据由罗切斯特梅奥诊所（Mayo Clinic, Rochester）在Greg Worrell博士监管下，作为NIH资助研究的一部分完成采集与去标识化，并由卡内基梅隆大学Bin He博士的实验室完成数据处理与整理。本数据集仅用于学术信息共享。
本数据集记录的脑电图（Electroencephalogram, EEG）遵循10-10系统，共包含76个记录通道。其中1个为参考通道，由于源成像需使用公共参考通道，因此该参考通道将被从地形图数据以及每位患者对应的导联场矩阵（lead-field matrix）行中移除，最终有效通道数为75。EEG数据经过1-40 Hz带通滤波，并采用公共平均参考（common average reference）进行预处理。
本数据集包含三个变量：
* spike_peak_data：包含每位患者所选棘波峰值时刻的脑电地形图数据。
* patient_id：患者索引，对应论文补充表S1中的患者ID。
* eloc：采用EEGLAB（https://sccn.ucsd.edu/eeglab/index.php）格式记录的电极位置信息。