Epileptic seizures recorded with microelectrodes: A persistent multiscale gap between neuronal activity, micro-, and...

The cascade of events that occur in the human brain, from neurons to local circuits and global network dynamics during epileptic seizures, is barely understood. Ictogenesis in humans has been described in relation to electrophysiological concepts based on local field potentials (LFP) recorded by standard macroelectrodes (macro-LFP). Microelectrodes, however, record at the cellular scale. Despite over four decades of such recordings in patients with epilepsy, there remains a significant gap between these scales. This narrative review explores the contribution of microelectrode recordings of seizures in humans. By focusing closely on neuronal activity, researchers often overlook that microelectrodes also allow recording LFP at the micro-electrode level (micro-LFP). Above all, there is a gap between local circuits recorded at the micro-LFP level and large-scale network dynamics at the macro-LFP level, with little theoretical work to reconcile these two scales. Consequently, to date, analyses of seizures have been coarse, incomplete, and based on small numbers of patients. In particular, most multiscale seizure analyses have not included all three levels of scales (single units, micro-LFP, and macro-LFP) simultaneously, but doing so is key to providing a synthesis of ictal genesis. This review highlights the various challenges that face researchers using microelectrodes: (1) carrying out a systematic descriptive and quantitative analysis of the micro-LFP seizure signal, (2) improving the spatial correspondence between micro- and macroelectrodes in order to achieve better comparability between the two scales, (3) improving brain sampling thanks to specific devices, in particular deep electrodes with microwires, (4) reporting the reference electrode used in each study and how it may impact the results, (5) long duration of recordings over hours and days, and (6) shared simultaneous micro-LFP/macro-LFP databases.