Closed-loop control of cortical network dynamics

Cortical networks in vitro display highly synchronized bursts that drive the network into saturation, limiting plasticity. The aim is to implement a real-time closed-loop feedback stimulator based on Wagenaar and Potter’s protocol (2005) on 120-channel MEAs. This controller aimed to restore a more asynchronous state, using electrical or optical stimulation as substitutes for missing afferents; to achieve this, a fully automated software pipeline based on Multi Channel System (MCS) hardware was developed to acquire, process activity and stimulate in real-time. Each experiment began with noise calibration and baseline recording, using voltage calibration to select electrodes that reliably evoke network responses. A diagnostic sweep was introduced to estimate the highest firing rate sustainable without voltage saturation, addressing a gap in the original protocol which relied on fixed targets. This target rate drives an adaptive feedback controller that continuously adjusts stimulation voltage to maintain it, pausing adaptation during spontaneous bursts. The resulting change in network burstiness is quantified via a Burstiness Index, defined as the fraction of spikes clustered in highly active windows, to quantitatively assess protocol efficacy. The software created was able to as to bypass the GUI-based MCS Experimenter soft ware and develop a custom workflow capable of interfacing directly with the acquisition and stimulation hardware, laying the groundwork for future protocol adaptations; while to exactly replicate the results of the original protocol further tests and analysis are needed.