PT  - JOURNAL ARTICLE
AU  - Yochum Maxime
AU  - Modolo Julien
AU  - Benquet Pascal
AU  - Wendling Fabrice
TI  - Reconstruction of post-synaptic potentials by reverse modeling of local field potentials
AID  - 10.1101/346148
DP  - 2018 Jan 01
TA  - bioRxiv
PG  - 346148
4099  - http://biorxiv.org/content/early/2018/06/13/346148.short
4100  - http://biorxiv.org/content/early/2018/06/13/346148.full
AB  - Among electrophysiological signals, Local Field Potentials (LFPs) are extensively used to study brain activity, either in vivo or in vitro. LFPs are recorded with extracellular electrodes implanted in brain tissue. They reflect intermingled excitatory and inhibitory processes in neuronal assemblies. In cortical structures, LFPs mainly originate from the summation of post-synaptic potentials (PSPs), either excitatory (ePSPs) and inhibitory (iPSPs) generated at the level of pyramidal cells. The challenging issue, addressed in this paper, is to estimate, from a single extracellularly-recorded signal, both ePSP and iPSP components of the LFP. The proposed method is based on a model-based reverse engineering approach in which the measured LFP is fed into a physiologically-grounded neural mass model (mesoscopic level) in order to estimate the synaptic activity of a sub-population of pyramidal cells interacting with local GABAergic interneurons. The method was first validated using simulated LFPs for which excitatory and inhibitory components are known a priori and can thus serve as a ground truth. It was then evaluated on in vivo data (PTZ-induced seizures, rat; PTZ-induced excitability increase, mouse; epileptiform discharges, mouse) and on in clinico data (human seizures recorded with depth-EEG electrodes). Under these various conditions, results showed that the proposed reverse engineering method provides a reliable estimation of the average excitatory and inhibitory post-synaptic potentials at the origin of the measured LFPs. They also indicated that the method allows for monitoring of the excitation/inhibition ratio. The method has potential for multiple applications in neuroscience, typically when a time tracking of local excitability changes is required.Author summary The measurement of excitatory and inhibitory processes which control brain excitability is a key issue in brain research. Indeed, alteration of excitation/inhibition balance is involved in both normal (like plasticity-related changes in response to learning) and pathological (like epileptic seizures) conditions. We report a novel method to estimate the two main components of the local field potential, namely the excitatory and inhibitory post-synaptic potentials. This method starts from extracellularly-recorded signals and uses a physiologically-relevant neural mass model to constrain the local field potential decomposition into post-synaptic potentials. To our knowledge, this neural mass model-based reverse engineering method is novel as it constitutes the first attempt to combine neural mass modeling with optimization techniques to extract hidden excitatory and inhibitory components from local field potentials recorded by extracellular macroelectrodes. A second progress beyond the state-of-the-art is the integration of more accurate physiology in the neural mass model as it now accounts for specific input-output functions that characterize glutamatergic principal neurons and GABAergic interneurons. Results demonstrated that the proposed method enables real-time tracking of neural excitability with essential information about dynamic changes of the excitation/inhibition ratio possibly impaired in the epileptic tissue. New epilepsy treatment can benefit from this method as it can help finding optimal control strategies based, for instance, on deep brain stimulation.