Automated estimation of frequency and spatial extent of periodic and rhythmic epileptiform activity from continuous electroencephalography data

Background and purpose: Rhythmic and periodic patterns (RPP) are harmful brain activity observed on continuous electroencephalography (cEEG) recordings of critically ill patients. The presence of RPPs at higher frequencies and on a larger scalp area (spatial extent) are associated with a higher probability of poor outcomes. This work describes automatic methods for detection of the frequency and spatial extent of specific RPPs: lateralized and generalized rhythmic delta activity (LRDA, GRDA) and lateralized and generalized periodic discharges (LPD, GPD). Methods: The frequency and spatial extent of RPPs is estimated using signal processing techniques combined with rule-based logic. The validation of the algorithms was performed on a total of 1087 cEEG segments. The annotations of three expert neurophysiologists for event frequency and spatial extent were considered the gold standard for the evaluation of the algorithm output. The inter-rater reliability (IRR) is evaluated for the assessment of performance. Results: The selected algorithms match or exceed the agreement of experts on the frequency and spatial extent of RPP segments. RDA1b-FFT (Fast Fourier Transform), the best algorithm for rhythmic delta activity, showed an expert-algorithm IRR ranging from a good to excellent intra-class correlation coefficient (ICC) of 66-96%, whereas the expert-expert IRR ranged from 60-92%. The best algorithm for periodic discharges, PD2a, showed an expert-algorithm IRR ranging from ICC of 13-80%, whereas the expert-expert IRR ranged from 13-86%. Conclusions: The proposed algorithms for estimating frequency and spatial extent of rhythmic and periodic patterns match expert performance and are a viable tool for large-scale cEEG analysis.