Figure 3
From: A robust deep learning detector for sleep spindles and K-complexes: towards population norms
SS and KC detection performance (F1-score) per parameter range. For SSs (from MODA dataset), the parameters considered were (a) duration, (b) PP amplitude, (c) spindle frequency and (d) age of the subject. For KCs (from MASS2-KC dataset), the parameters considered were (e) duration and (f) PP amplitude. Performance is measured by comparing detections and annotations that exist in a given range of the chosen parameter (e.g., between 0.6 s and 0.9 s of duration). Each data point represents the mean ± 2SD of the F1-score computed by micro-average.
