Decoding Brain States Based on Microcircuits

Decoding brain states based on single-trial neural activity data is important for understanding neural codes. Microcircuits are local functional connectivity among ensemble neural dynamics, and play a vital role in cognition and behavior. Calcium imaging can examine microcircuits in behaving animals. We propose a computational framework called microcircuit-based brain state decoding (MBSD) to distinguish brain states based on microcircuits generated from single-trial data. We validated MBSD using five calcium imaging datasets in which neural activities of DI medium spiny neurons (MSN) in the dorsal striatum were recorded. MBSD was used to distinguish high/low velocity based on DI-MSN neural activities; and achieved classification accuracy greater than 0.9 in all five experiments.