RESUMO
Background: West syndrome (WS) is a devastating epileptic encephalopathy with onset in infancy and early childhood. It is characterized by clustered epileptic spasms, developmental arrest, and interictal hypsarrhythmia on electroencephalogram (EEG). Hypsarrhythmia is considered the hallmark of WS, but its visual assessment is challenging due to its wide variability and lack of a quantifiable definition. This study aims to analyze the EEG patterns in WS and identify computational diagnostic biomarkers of the disease. Method: Linear and non-linear features derived from EEG recordings of 31 WS patients and 20 age-matched controls were compared. Subsequently, the correlation of the identified features with structural and genetic abnormalities was investigated. Results: WS patients showed significantly elevated alpha-band activity (0.2516 vs. 0.1914, p < 0.001) and decreased delta-band activity (0.5117 vs. 0.5479, p < 0.001), particularly in the occipital region, as well as globally strengthened theta-band activity (0.2145 vs. 0.1655, p < 0.001) in power spectrum analysis. Moreover, wavelet-bicoherence analysis revealed significantly attenuated cross-frequency coupling in WS patients. Additionally, bi-channel coherence analysis indicated minor connectivity alterations in WS patients. Among the four non-linear characteristics of the EEG data (i.e., approximate entropy, sample entropy, permutation entropy, and wavelet entropy), permutation entropy showed the most prominent global reduction in the EEG of WS patients compared to controls (1.4411 vs. 1.5544, p < 0.001). Multivariate regression results suggested that genetic etiologies could influence the EEG profiles of WS, whereas structural factors could not. Significance: A combined global strengthening of theta activity and global reduction of permutation entropy can serve as computational EEG biomarkers for WS. Implementing these biomarkers in clinical practice may expedite diagnosis and treatment in WS, thereby improving long-term outcomes.
