Figure 11: Single frame of a demixing video generated from real two-photon data.

In this frame, we see the various components of the demixing movie, presented in the exact same manner as Fig. 9. The lack of visible signal in the mean-subtracted residual suggests that our algorithm has successfully extracted all signals. Unlike Fig. 9, we subtract the minimum value from each pixel of the “Signal Estimates” and “Estimated Colored” frame to allow the reader to more clearly see the most active cells in each frame. Note that this dataset contains both somatic and dendritic signals; our algorithm is able to adequately identify both. We provide a demixing video here.