Strengths & Weaknesses of RANSAC applied to Epidural & Intracortical Recordings

We evaluated the efficacy of the Random Sample Consensus (RANSAC) algorithm for detecting and correcting bad channels in invasive neural recordings. To our knowledge, this is one of the first studies applying RANSAC to high-resolution modalities specifically epidural microelectrocorticography (μECoG) and intracortical microelectrode arrays (MEA). Three scenarios were tested: epidural signals, intracortical signals, and a noise-only benchmark. A spatiotemporal analysis assessed RANSAC's impact on signal integrity. RANSAC effectively identified and interpolated faulty channels in epidural μECoG data using optimized parameters, yielding markedly cleaner signals with reduced noise and improved inter-channel correlation. In contrast, the less correlated intracortical MEA signals showed only minimal changes under the default parameters. In fact, using parameters optimized for epidural data on the intracortical recordings attenuated legitimate signal features. These findings highlight the need for modality-specific tuning. Overall, our results demonstrate that RANSAC can substantially enhance data quality in epidural recordings, while underscoring the algorithm's limitations in intracortical settings. This study provides initial guidance for integrating RANSAC into invasive recording preprocessing pipelines.Clinical relevance-RANSAC can be integrated into invasive brain recordings (live or offline) to enhance spatiotemporal analysis by detecting and correcting faulty channels. This is particularly valuable for epilepsy applications, such as closed-loop neurostimulation.