Skip to content

Measure Object Sholl

Measure Object Sholl does a Sholl analysis (Sholl 1953) of skeletons: it draws concentric circles around a centre (usually the soma) at a fixed step and counts, per circle, how often the skeleton crosses it. This shows how much a neuron branches at which distance from its soma. The crossings per circle are a profile, a curve over the radius; the summary values follow Ferreira et al. 2014.

A skeleton around its soma and three circles at a fixed step: the dots mark where the skeleton crosses each circle; the crossings per circle make up the profile, the circle with the most is the critical radius, the last one crossed the enclosing radius

Use it on the skeletons of Skeletonize Objects. To centre the circles on the soma, link the skeletons to their somata with Relate Objects first and choose the soma class as Centre From.

In a project file it is MEASURE_OBJECT_SHOLL.

ParameterDescription
IDThe step’s ID, unique in the project. Its values’ names in the results start with it (e.g. sholl_01).
ObjectsThe skeletons measured (the skeleton class of Skeletonize Objects).
Centre FromThe class of the centres (e.g. the somata), linked to the skeletons as their parents by Relate Objects: the circles are around its centre. Unset (default): around the skeleton’s own centre.
Ring StepThe distance between two circles (0.1 to 100000, default 10).
First RadiusThe radius of the first circle (default 0: one step).
Last RadiusThe radius of the last circle (default 0: up to the farthest skeleton pixel). Set it to compare neurons at the same radii.
UnitUnit of the radii and the step: px (default) or nm. Without a pixel size, nm are taken as pixels.

A profile and five values per skeleton:

MeasurementUnitDescription
IntersectionscountThe profile: the number of crossings of each circle, one value per radius.
Max intersectionscountThe most crossings of any circle.
Critical radiuspxThe radius of the first circle with the most crossings.
Enclosing radiuspxThe largest radius still crossed.
Mean intersectionscountThe mean crossings of the circles from the first up to the enclosing radius.
Sholl regression coefficient-kk of the semi-log method: how fast the crossings per area fall with the radius (see below).

The radii are in pixels, also when Unit is nm. A skeleton that crosses no circle has 0 max and mean intersections and no other summary value.

A profile is a measurement with several values per object, each at a position: here the number of crossings at each radius. Each circle is one row in the results, its radius stored as the row’s position. It holds a value for every circle out to the last radius, zeros included, so averages over neurons stay right.

As a profile is no single number, the List View and the other charts don’t offer it as a column. The Profile chart of the results draws it: a line per class, image, well or plate, or the curve of the selected skeleton on top. In the measurements export it has a position column with the radius. The summary values are ordinary values, in the List View like any other. In the pipeline editor’s measurement preview, the profile shows the number of radii and objects instead of a value.

The skeleton is thinned once more first, so it is 1 pixel wide. The centre is the centre of mass of the skeleton’s parent of the Centre From class, or of the skeleton itself when Centre From is unset or the skeleton has no such parent.

The circles have the radii

ri=rfirst+i⋅Δr,ri≤rlastr_i = r_{\text{first}} + i \cdot \Delta r, \qquad r_i \le r_{\text{last}}

with Δr\Delta r the Ring Step.

Crossings are counted on the skeleton as a line, not by pixels in a ring (a 1-pixel ring could be jumped by a diagonal step). The skeleton is a set of steps between neighbouring pixels p,qp, q (straight or diagonal; a diagonal step that a straight neighbour of both pixels bridges isn’t one, as in Measure Object Skeleton). With dd the distance of a pixel’s centre to the centre, the circle rr crosses the step if

d(p)<r≤d(q)d(p) < r \le d(q)

Summary values (Ferreira et al. 2014), with N(r)N(r) the crossings of circle rr:

  • Max intersections is the highest N(r)N(r); the critical radius the first radius it is at.
  • The enclosing radius is the largest rr with N(r)>0N(r) > 0.
  • Mean intersections is the mean of N(r)N(r) over all circles up to the enclosing radius, circles without crossings in between included.
  • The Sholl regression coefficient kk comes from the semi-log method: a least-squares line through the circles with crossings,
ln⁡N(r)πr2=−k⋅r+m\ln \frac{N(r)}{\pi r^2} = -k \cdot r + m

It needs at least two circles with crossings.

  • Radii and crossings are 2D: circles, not spheres.
  • One centre per skeleton: a skeleton that spans several neurons (touching neurites) is analysed around one soma only. Trace Filaments splits touching neurons.