NeuroART: Real-time analysis of neuronal population activity for informed closed-loop experiments

Two-photon (2P) calcium imaging allows for the activity readout of large populations of neurons at single cell resolution, and holographic optogenetics makes the population accessible for targeted stimulation. However, significant pre- and post-processing is required to interpret signals from large datasets, which leads to the selection of stimulation targets based only on rudimentary qualitative criteria. We have developed NeuroART (Neuronal Analysis in Real Time), a software that accesses microscope data streams to provide real-time readout of neuronal activity, downstream analysis, and photostimulation during image acquisition.

Utilizing NeuroART, experimenters gain knowledge of the data quality in real time and can identify neurons of interest for targeted photostimulation based on quantitative analysis of the ΔF/F traces of each neuron. Multiple analysis techniques can be employed, for example: pairwise correlation-based analysis to construct functional networks and identify the most correlated neurons in the field of view.

NeuroART includes automatic neuronal cell identification routines (e.g., CaImAn, Cite-on) to facilitate real time neuronal activity analysis. The software is linked to a fast switching Spatial Light Modulator (Boulder Nonlinear Systems) that enables holographic optogenetic stimulation at millisecond timescales. The downstream analysis within the real-time loop runs rapidly enough to be computed on a laptop (Intel(R) Core i7-8565U CPU @ 1.80 GHz 1.99 GHz, 12 GB RAM, 64-bit Windows 10). Our software has been tested on a range of different imaging systems, imaging software and formats and operating systems. With these capabilities, NeuroART enables model-guided experiments for a wide range of microscope platforms.