Morphometric updates in NeuroMorpho.Org as of 24 June, 2013

The following 21 morphometrics are calculated for each reconstruction using L-Measure:
Soma_Surface, N_stems, N_bifs, N_branch, Width, Height, Depth, Diameter, Length, Surface, Volume, EucDistance, PathDistance, Branch_Order, Contraction, Fragmentation, Partition_asymmetry, Pk_classic, Bif_ampl_local, Bif_ampl_remote, Fractal_Dim.

Latest updates in L-Measure (v5.0) affect the following metrics in NeuroMorpho.Org:

1. Height, Width, Depth
All values (both whole and arbor specific selections) for width (w), height (h), and depth (d) are calculated after orienting the whole neuron based on major axes as computed from a slight variation of principal component analysis (PCA) or single value decomposition (SVD). This procedure first translates the entire neuron such that the soma is at the origin (0,0,0). Then, the direction that captures the greatest amount of variance in the data (i.e. that minimizes the variance of data points around a line in that direction running through the center) is the first principal component. The second and third principal component axes minimize the variance while being oriented orthogonal to the first. The h, w, and d are calculated as the minimum span encompassing 95% of the tracing points along the first, second and third principal components, respectively. Therefore, the values correspond to the relative height, width and depth of selected arbor types within the orientation of the whole neuron.

2. All metrics except Soma_Surface and N_stems are calculated with specificity arbor type > 1. Hence, these metrics will exclude compartments defined as soma.

3. Bif_ampl_remote
Metrics for remote angles were modified to include the terminal branches (Previous calculations were restricted only to branches between two bifurcation points).

4. Fragmentation
Fragmentation is modified to include the stem branches (Previously the stem compartments were not included in the results).

5. Fractal dimension
The fractal dimension is computed for every branch as the slope of the regression line between the log10 of path distance and the log10 of Euclidean distance of each tracing point from the beginning of the branch.

See L-Measure for further details.