Spatiotemporal prediction of resting-state hemodynamics in awake mouse brain using deconvolution. From Left to Right are shown concurrent hemodynamically corrected Thy1-GCaMP6f ΔF/F, Δ[HbT], hemodynamic prediction using a deconvolution-based HRF, and the residual difference between the measurement and prediction. Both corrected GCaMP ΔF/F and Δ[HbT] were bandpass filtered from 0.02 to 2 Hz. Movie corresponds to the entire 60-s resting-state dataset for the epoch shown in Fig. 3B. The animal was not running during this trial. Patterns of increases and decreases in ∆[HbT] over the entire field of view can be seen to be mirrored between measured hemodynamic data and data modeled from simultaneously acquired neural activity recorded via Thy1-GCaMP fluorescence data.