Figure 9

Semi-automated skeletonization and puncta-finding. We illustrate here how we obtained skeletons for the dendrites (a, b), and determined the location of putative postsynaptic sites (c-k). (a) A set of masks is generated in Amira to differentiate dendritic regions (masked in red) from the cell body (blue) and background artifacts (green). (b) In order to map dendritic length, a dendritic skeleton was generated by first selecting a seed position (S) at an edge of the dendritic mask. This seed initiated a wave of activation that propagated through the dendritic mask, voxel by voxel, until the mask was filled (red shading). The mean positions of this wave front (1–6) were connected to generate the dendritic skeleton (blue). (c) Punctate regions of the YFP channel (green) were first identified by thresholding 8 bit images at every other gray value. For each threshold, the red rectangular box represents connected voxels that pass the intensity and size criteria (see Materials and methods). The volume of each punctum was defined by the full-width-half-maximum (blue shading) of the summed rectangles. (d) For each potential punctum identified in (c) (shaded blue), the brightness of the YFP channel was compared to the td-Tomato channel to obtain ΔF/F. Red and green values for the non-punctate region of the image were compared to determine a green value predicted by each red value (dark green line). ΔF for each punctum was then defined as the difference between the actual and predicted green value at that punctum, and was divided by the predicted green value to obtain ΔF/F. (e-k) Examples of proximal and distal dendrites passing through the user guided stage of puncta identification. (e) Dendritic fill with td-Tomato. (f) PSD95-YFP clustering. (g) PSD95-YFP displayed at high contrast to show dimmer puncta. (h) td-Tomato and PSD95-YFP. (i) Potential puncta identified with minimal thresholds applied. (j) User selection of potential puncta (yellow circles) and artifacts (magenta circles), which are then used to determine a template of optimal thresholds for defining puncta across the entire cell. (k) Final output of the dot finder program once optimized thresholds are applied.