Video 1 from Go with the FLOW: visualizing spatiotemporal dynamics in optical widefield calcium imaging

Steps to compute a FLOW portrait for a synthetic two-dimensional Gaussian that grows, translates to the right, then shrinks. This synthetic example is depicted in Figure 1, and Supplemental Figures 2 and 4. (Raw Data) Raw spatiotemporal data for the two-dimensional Gaussian example. (Optical Flow) The optical flow vector field overlaid on the original data. Optical flow is used to convert frame-by-frame changes in pixel intensity into a vector field. (Particles) Uniformly spaced particles are propagated through space according to the optical flow vector field. The particles' po23 sitions are updated by adding a displacement that corresponds to the vector field multiplied by a unit time step. Regions where particles cluster will be captured as ridges in the backward time FTLE and regions where particles separate will be captured as ridges in the forward time FTLE. (FLOW Portrait) The final FLOW portrait for this dataset. Orange structures correspond to the forward time FTLE and show regions where activity is repelled; purple structures correspond to the backward time FTLE and show regions where activity is attracted. The FLOW portrait was computed using a XX frame integration length and a XXth threshold percentile