Extended Data Fig. 3: Spatial regression for correction of hemodynamic artifacts in mesoscopic data.
From: Spatiotemporally heterogeneous coordination of cholinergic and neocortical activity
a, GFP/mCherry fluorescence signals measured in V1 during presentation of drifting grating stimuli, illustrating different hemodynamic correction methods. Plots show averaged ΔF/F GFP (left) and mCherry (right) activity evoked by visual stimulation in one example session from a mouse co-expressing GFP and mCherry. Traces are for uncorrected fluorescence (blue) and images corrected using pixel-wise regression of 395 nm fluorescence data (purple) or 530 nm back-scatter data (light green), or spatial regression of 395 nm (orange) or 530 nm (dark green) data. b, ΔF/F GFP and mCherry activity across all parcels, evoked by visual stimulation. Population averaged (n = 3 mice) images are from uncorrected data and data corrected using different regression methods. c, Individual and population averaged (n = 3 mice) GFP and mCherry values in V1 for visually-evoked ΔF/F negativity (hemodynamic artifact), from uncorrected data. * indicates p < 0.05, two-tailed paired t-test (t(2) = −17.353, p = 0.003). d, Individual animal and population mean (n = 3 mice) values for visually-evoked ΔF/F GFP negativity for the different correction methods. * indicates p < 0.05, post hoc two-tailed paired t-tests following repeated measures ANOVA comparing uncorrected, pixelwise (395) and spatial (395) regression methods (F(2,4) =14.754, p = 0.014; uncorrected vs pixelwise: t(2) = −2.499, p = 0.130; uncorrected vs spatial: t(2) = −5.905,p = 0.028; pixelwise vs spatial: t(2) = −2.807, p = 0.107). e, as in d for mCherry data (F(2,4) = 6.270, p = 0.059). f, ACh3.0 signals measured in V1 during visual stimulus presentation (arrows), illustrating different hemodynamic correction methods. g, Average ΔF/F ACh3.0 activity evoked by visual stimulation in one example session with each correction method. h, Individual animal and population mean (n = 6 mice) values for visually-evoked ΔF/F negativity with each correction method. * indicates p < 0.05, post hoc two-tailed paired t-tests following repeated measures ANOVA (F(2,10) = 54.423, p < 0.001; uncorrected vs pixelwise: t(5) = −4.758, p = 0.005; uncorrected vs spatial: t(5) = -7.259, p = 0.001; pixelwise vs spatial: t(5) = −7.568, p = 0.001). i-k, as in (f-h) for jRCaMP1b data (n = 6 mice) in V1 (F(2,10) = 6.776, p = 0.014; uncorrected vs pixelwise: t(5) = −1.719, p = 0.146; uncorrected vs spatial: t(5) = −2.607, p = 0.048; pixelwise vs spatial: t(5) = −2.610, p = 0.048). l, Pixel-wise variance remaining from imaging of a GFP-expressing mouse following hemodynamic correction with pixel-wise (left) or spatial (right) regression of 395 nm fluorescence data. m, Average GFP fluorescence in V1 evoked by air-puff stimulus to the animal’s flank. Traces are for uncorrected fluorescence (blue) and images corrected using pixelwise (purple) or spatial (red) regression of 395 nm data. n, Individual animal and population mean (n = 3 mice) values for air-puff-evoked ΔF/F negativity for the different regression methods. * indicates p < 0.05, post hoc two-tailed paired t-tests following repeated measures ANOVA (F(2,4) = 36.002, p = 0.003; uncorrected vs pixelwise: t(2) = −2.583, p = 0.123; uncorrected vs spatial: t(2) = −6.821, p = 0.021; pixelwise vs spatial: t(2) = −8.203, p = 0.015).
