Extended Data Fig. 6: Deformation-based motion correction using displacement fields.
From: Long-term optical imaging of the spinal cord in awake behaving mice
a, Each motion correction method run on the movie (5,000 frames, 13.9 Hz) from a Phox2a-Cre; Ai162 (GCaMP6s) mouse displays 1: the mean of all movie frames, 2: combined numerical gradient in both lateral directions on the mean frame, 3: the standard deviation over all movie frames (hence visibility of neurons on left and right side of the spinal cord), and 4: ΔF/F frames. Arrows indicate areas of interest where differences between methods are most evident. b, 2D correlation coefficient of all frames to the mean frame of the movie (as in a) for displacement field motion correction compared to raw, TurboReg, and NoRMCorre. All movies (except raw) were spatially filtered to remove large magnitude, low-frequency changes in fluorescence, which artificially enhances correlations. c, Histogram of 2D correlation coefficients over all frames from b. d, Spearman’s rho of all frames to the mean frame of the movie (as in a) for displacement field motion correction compared to raw, TurboReg, and NoRMCorre. All movies (except raw) were spatially filtered to remove large magnitude, low-frequency changes in fluorescence, which artificially enhances correlations. e, Histogram of Spearman’s rho values over all frames from d.
