Extracting Two-Dimensional Signal Correlation Maps via Gaussian Process Regression with Zernike Means

Analyzing the functional connectivity of the brain is an enormous challenge, as deciphering it requires knowledge of both functional responses and circuit structure. A promising strategy involves examining the spatial pattern of stimulus-driven activity correlations across neuronal populations. In the mouse primary auditory cortex (A1), while a global tonotopic organization exists, local frequency preferences in layer 2/3 exhibit a fractured, heterogeneous map. This raises the question of whether functional connectivity is spatially ordered or disordered. To answer this question, we developed an inference methodology based on a spatial Gaussian Processes with Zernike means (GPRZ) to estimate robust and statistically principled 2D SC maps from in vivo two-photon calcium imaging of A1 during presentation of pure tones (PTs) and harmonic two tone pairs (TTs) that are smooth and interpretable. We found that PT-driven signal correlation maps were restricted to local neurons, whereas TT signal correlations showed more global textures. Harmonic stimuli elicited spatially distinct patterns in SC maps, revealing precise functional connectivity between harmonically tuned neurons in TT presentations. Our findings suggest that while neighboring neurons in A1 exhibit diverse frequency preferences, their functional connectivity is precise and selectively aligned with harmonic relationships.