Journal of Neural Engineering · 2015 · 95 citations · 46 references
ComputingData ScienceElectrophysiologyPsychophysiologyTime Series AnalysisElectroencephalogram AnalysisElectroencephalogram Signal ProcessingElectroencephalographySpectral AnalysisTime Series Feature ExtractionTime-frequency AnalysisEeg Feature ExtractionEeg Signal ProcessingElectroencephalogram ClassificationElectroencephalographic Pattern ClassificationElectroencephalography Feature ExtractionStatistical Discriminant AnalysisTime-frequency Feature ExtractionTime-frequency Localization
OBJECTIVE: Although interest in using electroencephalogram (EEG) activity for subject identification has grown in recent years, the state of the art still lacks a comprehensive exploration of the discriminant information within it. This work aims to fill this gap, and in particular, it focuses on the time-frequency representation of the EEG. APPROACH: We executed qualitative and quantitative analyses of six publicly available data sets following a sequential experimentation approach. This approach was divided in three blocks analysing the configuration of the power spectrum density, the representation of the data and the properties of the discriminant information. A total of ten experiments were applied. MAIN RESULTS: Results show that EEG information below 40 Hz is unique enough to discriminate across subjects (a maximum of 100 subjects were evaluated here), regardless of the recorded cognitive task or the sensor location. Moreover, the discriminative power of rhythms follows a W-like shape between 1 and 40 Hz, with the central peak located at the posterior rhythm (around 10 Hz). This information is maximized with segments of around 2 s, and it proved to be moderately constant across montages and time. SIGNIFICANCE: Therefore, we characterize how EEG activity differs across individuals and detail the optimal conditions to detect subject-specific information. This work helps to clarify the results of previous studies and to solve some unanswered questions. Ultimately, it will serve as guide for the design of future biometric systems.
46
PhysioBank, PhysioToolkit, and PhysioNet
Ary L. Goldberger, Luı́s A. Nunes Amaral, Leon Glass et al. · Circulation · 2000 · 14.5K citations · Full text
DEAP: A Database for Emotion Analysis ;Using Physiological Signals
Sander Koelstra, Christian Mühl, Mohammad Soleymani et al. · IEEE Transactions on Affective Computing · 2011 · 4.8K citations · Full text
Nursing, Neuroscience, Psychology +17
Data Analysis: A Model-Comparison Approach
Geoff Vining, Charles M. Judd, Gary H. McClelland · Technometrics · 1992 · 941 citations