Preservation of Genuine Neural Activity Under Selective Correction

Determine whether SQI-guided selective artifact correction better preserves genuine neural activity than alternative automated EEG artifact-removal approaches, using artifact-annotated, controlled-contamination, or clean-reference recordings.

Background

The study evaluates automated artifact-removal methods through cross-dataset EEG age-prediction performance, but predictive outcomes alone cannot establish whether corrected information is neural or artifactual. The proposed SQI-guided GEDAI method reduces the proportion of channel–epoch pairs modified while retaining contaminated epochs, yet the modification ratio does not measure neural preservation or artifact attenuation.

The authors identify artifact-annotated recordings, controlled-contamination datasets, and clean-reference recordings as necessary for directly determining whether selective correction preserves genuine neural activity more effectively. This remains unresolved by the benchmark because it uses naturally acquired EEG without an independent clean signal reference.