Supplement 3: Table of Adjustable Parameters Used in the Software, and Their Descriptions

Supplement 3: Table of adjustable parameters used in the software, and their descriptions

Parameter
(section in paper) / Description / Typical range of values
Intensity threshold (Section 4.1) / Threshold value to identify pixels that are definitely part of the background that can be ignored for processing (use higher values for images with higher background intensities). / 2 – 7 (based on background range)
Connected components size (Section 4.1) / Used to remove objects that are too small to be spines (use higher value to eliminate larger false objects due to imaging noise). / 10 – 100
Anisotropic Diffusion conduction parameter k(Section 4.1) / Anisotropic diffusion conduction coefficient. (use higher value of this parameter to increase the amount of image smoothing). / 10 – 800
Anisotropic Diffusion smoothing iterations t (Section 4.1) / Number of iterations of anisotropic diffusion smoothing (use larger number to guarantee convergence if necessary). / 2 – 10
Gradient vector magnitude threshold for detecting critical points (Section 4.2) / This is a threshold value used to choose critical points. When the image intensity gradient magnitude is lower than this threshold, a critical point is detected (use higher value to detect more critical points, i.e., increase sensitivity of critical point detection). / 0.04 – 0.15
Threshold for detecting high curvature seed points (Section 4.2) / If the curvature at a point has a value above this threshold parameter, and it is also a local maximum of curvature values, we treat it as a high curvature seed point (use higher values to pick up fewer seed points) / 0 – 10
IW-MST edge range threshold (Section 4.3) / If the distance between 2 vertices on the initial graph exceeds this threshold, we assume there is no edge between them in the graph structure before generating the IW-MST (use higher value to obtain a more connected graph). / 5 – 15
Graph prune size (Section 4.4) / Threshold size used to prune the minimal spanning tree structure and eliminate trivial branches are shorter than this length (use a higher value for pruning longer branches). / 2 – 10
Number of iterations for the graph morphology (Section 4.4) / The number of iterations of graph theoretic erosion and dilatation steps to clean up the IW-MST to estimate the neurite backbones. (use a higher value for eliminating longer small branches). / 30 – 70
MDL weight factor α
(Section 4.4) / Optional parameter to adjust the tradeoff between coverage and conciseness in the MDL estimation for spines only. (use a higher value for more complex models). / 0.5 – 0.95
Extra spine detection offset (Section 5) / This is with reference to the method illustrated in Figure 3F. If the distance between the initially detected backbone (blue line) and the fitted spline (red line) exceeds this threshold, then we detect this missing spine (use a lower value for detecting more extra spines). / 1.5