Fig. 3: Performance of SilenceMap on a real EEG dataset.
From: Neural silences can be localized rapidly using noninvasive scalp EEG
The first row shows the extracted ground truth regions of silence (red regions) overlaid on the resected cortical region of three patients based on their symmetric brain models extracted from the structural MRIs (see the MRI scans in Fig. 2); the second, third, and fourth rows show the performance in localization of the silent region using modified source localization algorithms (MNE, MUSIC, and sLORETA), through both visual illustration (red regions) and using performance metrics of center-of-mass (COM) distance (ΔCOM), Jaccard Index (JI), and size error (Δk). The fifth row shows the performance of SilenceMap without baseline, and the last two rows show the localization performance of SilenceMap with baseline, based on the Rest and Visual recordings respectively. p is the total number of sources in each brain model, and k is the size of ground truth region of silence.
