Figure 1.
Geometric transformation of activity in a single brain slice, in stratum radiatum, evoked by temporoammonic pathway stimulation.
(A) Schematic and VSDI camera frame showing the hippocampal anatomy. (B) Framework for transforming 3-D movie data (x,y,time) to 2D raster image (polygon position, time). An average temporal signal is obtained from the pixels enclosed by each polygon. (C) After transformation, a raster plot completely displays the spatiotemporal response in the stratum radiatum. Warmer colors indicate depolarization; cooler colors indicate hyperpolarization. Each row of the raster is a temporal trace from one polygon in B. From bottom to top, rows proceed from the Hilus, to CA3, to CA1. White rows indicate transitions between the anatomical regions. (D–F) Full VSDI camera frames, showing activity at (D) −10 ms (E) +10 ms, and (F) +36 ms (stimulus occurs at t = 0). For comparison, these time frames correspond to the black arrowheads that mark the columns in C. To aid visualization of the anatomy in D–F, ΔF/F values within 1.5× of the standard deviation of pre-stimulus noise were excluded from pseudocoloring (no points were excluded from pseudocoloring in the rasters). (G) Temporal activity at selected hilus and CA1 polygons. Spatial positions of the selected polygons are indicated with red (hilus) and blue (CA1) arrowheads in C.
Figure 2.
Statistical analysis of the response to temporoammonic stimulation in the stratum radiatum.
(A–B) Rasters from multiple recordings are averaged to show overall trends in activity in the (A) control and (B) mutant groups of rasters. Visual inspection suggests that activity is different between groups. (C) Heatmap showing the degree of difference in activity between groups, across space and time. Statistically significant p-values (p<0.05) are shaded purple. (D) To obtain a spatiotemporal map of the significant difference in activity in mutant hippocampus, the control raster A was subtracted from the mutant raster B. A threshold was applied to display only sites of significant difference (p<0.05). Significant differences were registered at 7499 of 13244 sites (57%). Color scale is the same in A, B, and D.
Figure 3.
Geometric transformation of activity in stratum radiatum, evoked by perforant pathway stimulation.
(A) Schematic and VSDI camera frame showing the hippocampal anatomy and the position of the stimulating electrode. Two electrodes were placed in the slice, but only the electrode labeled “PP stim” (visible at the right edge of the raw image) was used to deliver the stimulus. The remaining electrode, marked with an “X” in the schematic, was unplugged during this recording. (B) Polygonal geometry for transforming data to 2D. A temporal signal is obtained from each polygon. (C) After transformation, a raster plot completely displays the spatiotemporal response in the stratum radiatum. Warmer colors indicate depolarization; cooler colors indicate hyperpolarization. Each row of the raster is a temporal trace from one polygon in B. From bottom to top, rows proceed from the Hilus, to CA3, to CA1. White rows indicate transitions between the anatomical regions. (D–F) Full VSDI camera frames, showing activity at (D) −10 ms (E) +10 ms, and (F) +36 ms (stimulus occurs at t = 0). For comparison, these frames correspond to the black arrowheads that mark the columns in c. To aid visualization of the anatomy in panels D–F, ΔF/F values within 1.5× of the standard deviation of pre-stimulus noise were excluded. (G) Temporal activity at selected hilus and CA1 sites. Spatial positions of these signals are indicated with red (hilus) and blue (CA1) arrowheads in C.
Figure 4.
Statistical analysis of the response in stratum radiatum to perforant pathway stimulation.
This analysis was conducted in the same manner as shown in Figure 2. (A–B) Visual inspection of the averaged (A) control and (B) mutant rasters suggests that activity is similar in both groups. (C) Heatmap showing the degree of difference in activity between groups, across space and time. Statistically significant p-values (p<0.05) are shaded purple. (D) To obtain a spatiotemporal map of the significant difference in activity in mutant hippocampus, the control raster A was subtracted from the mutant raster B. A threshold was applied to display only sites of significant difference (p<0.05). Significant differences were registered at 669 of 13244 sites (5%). For α = 0.05, we expect 5% of sites to be identified as significantly different by chance. Therefore, these data indicate that perforant pathway evoked activity in the stratum radiatum is not significantly different between groups. Color scale is the same in A, B, and D.
Figure 5.
Spatiotemporal analysis, compared to conventional analysis, of the response to temporoammonic pathway stimulation in mutant (n = 8) and control (n = 10) slices.
(Same data as in Figure 2.) (A) In the first step of conventional VSDI analysis, a temporal signal is obtained for a region of interest (ROI) by averaging the data across a cluster of pixels and over a time interval of interest. Here, we have selected two regions: CA1 stratum radiatum and CA3 stratum radiatum. (B) The average response in the CA1 stratum radiatum ROI to four, 10 Hz stimuli delivered to the temporoammonic pathway in mutant (red) and control (black) slices in CA1 stratum radiatum. Conventional analysis proceeds by identifying time intervals of interest in these data. Here, we have chosen time intervals corresponding to the fast excitatory postsynaptic potential (EPSP) and the slow inhibitory hyperpolarization that follows the stimulus. Both of these time intervals are marked with magenta bars and these intervals are 6 ms long (3 camera frames at 500 frames per second). (C) The average response in the CA3 stratum radiatum ROI, in the same recordings of mutant and control slices. We have chosen to analyze the time interval corresponding to the fast EPSP for analysis in CA3; this 6 ms-long interval is marked with a cyan bar. (D–F) Traditional (ROI-based) statistical comparison of voltage-sensitive fluorescence in mutant and control slices in (D) CA1 during the fast EPSP, (E) CA1 during the slow inhibitory response, and (F) CA3 during the fast EPSP. Significant differences were observed in the fast EPSP in CA1 and in the fast EPSP in CA3 (t-test; ** P<0.01, * P<0.05; solid and dashed lines indicate mean±standard deviation). No significant difference was observed in the slow inhibitory response in CA1. (G–H) Rasters of average activity in (G) control and (H) mutant slices. Visual inspection suggests that activity is qualitatively different between groups across many sites. (I) Heatmap showing sites of significantly different spatiotemporal activity that were identified by the permutation test. The spatiotemporal sites that were analyzed using conventional VSDI analysis (described in A–F) are outlined in CA1 (magenta) and CA3 (cyan). The spatial (vertical) and temporal (horizontal) dimensions of these boxes match the spatial and temporal extent of the conventional ROI analyses performed in panels A–F. All boxes are 6 ms (3 samples) wide.
Figure 6.
Frequency of sites identified as significantly different in control data: response in stratum radiatum following perforant pathway stimulation.
(A–B) A total of 18 control slices were shuffled into two random groups, “Group A” and “Group B”. The average of each group is shown. (C) Heatmap, showing p-values obtained by comparing Group A to Group B at each spatiotemporal site. P-values less than 0.05 are colored purple. For the random Groups A and B, significant differences were registered at 4.98% of sites. (D–E) Histogram and cumulative probability distribution, showing the number of positive sites obtained from permutation test comparison of 1000 random groupings of the slices. Under the null hypothesis, the theoretical rate of observation of positive sites is predicted to be 5% for α = 0.05. In 1000 permutations of the actual data, 4.90% of sites were registered as significantly different.
Figure 7.
Frequency of sites identified as significantly different in control data: response in stratum radiatum following temporoammonic pathway stimulation.
(A–B) A total of 10 slices were shuffled and divided into two groups, “Group A” and “Group B”. The average of each group is shown. (C) Heatmap, showing p-values obtained by comparing Group A to Group B at each spatiotemporal site. P-values less than 0.05 are colored purple. For the random Groups A and B, significant differences were registered at 4.72% of sites. (D–E) Histogram and cumulative probability distribution, showing the number of positive sites obtained from permutation test comparison of all possible groupings of the data into 2 groups of 5 (126 unique combinations). Under the null hypothesis, the theoretical rate of observation of positive sites is predicted to be 5% for α = 0.05. In 126 permutations of the actual data, 3.97% of sites were registered as significantly different.
Figure 8.
In temporoammonic pathway evoked activity in the stratum radiatum, the statistical analysis output is similar regardless of the chosen width of polygonal segments for image segmentation.
To determine if the percentage of spatiotemporal sites identified as significantly different was a function of the width of polygonal segments used in image segmentation, the image segmentation procedure was repeated fifteen times, to generate rasters from polygonal geometries with segment widths ranging from 0.020 mm to 0.300 mm, in 0.020 mm increments. Permutation testing was conducted to compare experimental rasters to control rasters that were generated from geometries with each of the fifteen segment widths, to determine if segment width influenced the output of the permutation test. (A) Polygonal geometries are shown for the narrowest (0.020 mm) and the widest (0.300 mm) segment widths tested. (B) The percentage of sites identified as significantly different for each of the fifteen segment-width iterations. The percentage of sites identified as significantly different is similar across most segment widths. One notable exception occurs at segment width = 0.020 mm, where fewer sites of significant difference were identified. This is probably because the segment width (0.020 mm) is narrower than the pixel width (0.025 mm), so these segments each include few pixels and therefore the resulting signals from these narrow segments have poor signal-to-noise ratio. (C) The significant difference in activity, identified by permutation testing, for each of the fifteen segment widths tested (each panel corresponds to one tested segment width). The difference in activity is shown in the same format as in Figure 1D: (mutant activity – control activity), thresholded to show only sites of significant difference at α = 0.05. Qualitatively, the output of the permutation test is similar enough that the same conclusions about the underlying physiology could be drawn from consideration of any of the panels.
Figure 9.
In perforant pathway evoked activity in the stratum radiatum, the statistical analysis output is similar regardless of the chosen width of polygonal segments for image segmentation.
(This is the same test shown in Figure 8, applied to perforant pathway VSDI data.) To determine if the percentage of spatiotemporal sites identified as significantly different was a function of the width of polygonal segments used in image segmentation, the image segmentation procedure was repeated 15 times, to generate rasters from polygonal geometries with segment widths ranging from 0.020 mm to 0.300 mm, in 0.020 mm increments. Permutation testing was conducted to compare experimental rasters to control rasters that were generated from geometries with each of the fifteen segment widths, to determine if segment width influenced the output of the permutation test. (A) Polygonal geometries are shown for the narrowest (0.020 mm) and the widest (0.300 mm) segment widths tested. (B) The percentage of sites identified as significantly different for each of the fifteen segment-width iterations. The percentage of sites identified as significantly different is similar across most segment widths. For all segment widths, the percentage of sites identified as significantly different is similar to the percentage of sites expected to be identified as significantly different by chance (∼5% of sites at α = 0.05). (C) The significant difference in activity, identified by permutation testing, for each of the fifteen segment widths tested (each panel corresponds to one tested segment width). The difference in activity is shown in the same format as in Figure 1D: (mutant activity – control activity), thresholded to show only sites of significant difference at α = 0.05. Qualitatively, the output of the permutation test is similar enough that the same conclusions about the underlying physiology could be drawn from consideration of any of the panels.