Separating hemodynamic activity in mouse visual cortex into local and global components. A: experimental setup showing the camera for intrinsic imaging and the green LED (530 nm), the camera for eye tracking and the infrared LED (850 nm), the monitor showing the drifting bars, and the head fixation system. B: the spatial component F(s) of the separable model (). A, anterior; P, posterior, L, lateral, M, medial. C: the temporal component G(t) of the separable model (global component). D: map of preferred elevation obtained from the data in H. E: the response to moving flickering bars, expressed as a function of space (concatenated into a single dimension) and time, 10–40 s after stimulus onset. F: the baseline image in B expressed as a 2-D matrix. G: the global component in C expressed as a 2-D matrix. H: the local component. Pixels in E–H were sorted by their phase at the frequency of the stimulus.

Local activity, unlike global activity, reflects sensory responses. A: moving bar stimulus (4° wide drifting at 0.12 Hz for 5 cycles) and color bar mapping degrees of visual field to the position of the bar on the screen. B: map of retinotopy obtained through frequency analysis of the local component. Colors indicate preferred position in visual field (color scale in A). Brightness indicates signal amplitude. Scale bar, 1 mm. C: time course of global activity (ΔI/I) for a single stimulus presentation. D: same as C for the local activity measured in the representative pixel marked with a white cross in B. E and F: same as C and D but averaged across repeats. G and H: power spectral density of the global component and local component. A peak at frequencies of ∼2–3 Hz reflected respiration rate. Dotted line indicates the frequency of the stimulus. A peak was only present in the local component. The power spectral density of the local component was averaged across all pixels. I: average amplitude of the fluctuations for 17 experiments for the global (y-axis) and local (x-axis) components. Each circle represents an individual experiment. The square indicates mean across the population.

Relationship between pupil dilations and global hemodynamic activity. A: 3 consecutive recordings of pupil dilations. Traces are briefly interrupted between recordings. B: simultaneously measured global fluctuations of the hemodynamic activity (red). The prediction for the global hemodynamics (gray) is obtained by convolving the kernel (black trace in C) with the pupil radius trace. The kernel is computed over a training set of 80% of the data (red bar) but also predicts the test data (black bar). All traces in A and B are z-scored and therefore dimensionless. C: correlation between pupil radius and global hemodynamics (thin black trace) and kernel predicting global hemodynamic activity computed from the trials in A and B (thick black trace) or computed by shuffling the trial number in the eye tracking experiment (gray dashed trace). D: average correlation between pupil radius and global hemodynamics (thin black trace) and average kernel predicting global hemodynamic activity from pupil dilations (thick black trace), in 17 experiments (n = 10 mice). When we shuffled the trial numbers in each eye-tracking experiment, the peak in the kernel disappeared (dashed trace).