Exploration of endoscopic neurodynamic activity via computational modelling

Abstract

Calcium imaging techniques are at the forefront of modern neuroscience research, enabling single-cell resolution imaging of large neuronal populations. However, current research utilising calcium imaging techniques focuses primarily on quantitative analysis. This study investigated the feasibility of developing computational neural circuit models directly from calcium imaging recordings, which would enable deeper quantitative analysis of the underlying mechanisms that drive such neural circuits. Initially, I applied a state-of-the-art calcium imaging pipeline to identify cell bodies and extract neuronal activity from recordings of a rat hippocampal Cornu Ammonis 1 (CA1) region. Next, I implemented a computational model of the CA1 using a modified leaky integrate and fire formulation. Finally, I applied optimisation techniques to calibrate the model to retrieve the original network connectivity and neurodynamics observed in the calcium imaging recordings. While I was successful at developing a simple model of the CA1 capable of replicating known CA1 dynamics, as well as calibrating my model to synthetic data, I was ultimately unable to reconstruct the observed biological activity. My analyses indicated that the main limitation of my pipeline was the limited neuronal activity extracted from calcium imaging, which fails to describe the in-vivo neurodynamics.