Written and maintained by CASRAI Editorial Board
Last updated
Short answer
Stereology is a set of sampling-based methods that estimate three-dimensional quantities, such as volume, surface area, length and particle number, from two-dimensional sections or other lower-dimensional probes. It does not rebuild the whole structure. It measures a carefully sampled part and extrapolates to the whole, and the sampling design determines whether the estimate is trustworthy.
Definition and scope
A widely quoted definition describes stereology as “the science of reconstruction of 3-dimensional structures from 2-dimensional sections of them.” In practice, a review in Glomerular Diseases clarifies that stereology estimates 3D parameters from lower-dimensional probes rather than reconstructing the object from serial sections.
The quantities it targets are size, shape and number. Two broad families exist:
- Model-based (“biased”) stereology relies on assumptions about the shape and size of the particles being measured.
- Design-based (“unbiased”) stereology uses sampling designs that avoid those geometric assumptions. The disector, described below, is the main example.
Design-based work depends on random sampling. The same review calls a system of random sampling “of paramount importance,” with equal probability for each part of the sample space. A vendor-run educational site on design-based stereology states the same rule: probes should always be used together with systematic random sampling.
How it is used in practice
Most stereology starts from tissue that has been cut into sections (see what a microtome does) and often stained, for example by immunohistochemistry. A grid, frame or virtual probe is then laid over the image or the section, and the analyst records how the probe meets the structure of interest. The main building blocks are:
- Cavalieri estimator. Volume is estimated as the sum of the section areas multiplied by the spacing between sections. It is a design-based method that dates from 1635.
- Point counting. A grid of points is projected on the image. Points falling on the object, compared with points falling on the reference space, give an area estimate.
- Disector. Two sections, a reference section and a look-up section, are compared. A particle that “uniquely first appears in one section, regardless of its size or shape” is counted, so counting needs no assumption about particle size or shape.
- Fractionator. This design directly estimates the total number of particles within a reference space.
Because the scale of an image matters for any area or length estimate, imaging settings should be calibrated first; see microscope calibration with a stage micrometer.
Probes and what each estimates
The table below maps each quantity to a probe named for it in design-based stereology.
| Quantity of interest | Probe or method | How it is described |
|---|---|---|
| Volume of a region | Cavalieri estimator with point counting | Section areas times section spacing; points on object versus reference space give areas |
| Number of cells or particles | Disector with fractionator; optical fractionator | Counts particles that first appear in a look-up section; the optical fractionator samples sub-volumes of thick sections and extrapolates to the whole |
| Volume of individual cells | Nucleator | Uses one to four rays from a point inside the cell |
| Length of filaments such as axons or vessels | Spaceballs | Counts intersections with a virtual sphere |
| Surface area of membranes | Isotropic fakir | Uses a triplet of line segments |
Where it matters for labs and core facilities
Stereology matters wherever a study reports a total, such as total cell number in a brain region or total volume of a compartment in an organ, and a count made on a single section cannot stand in for that total. Labs and imaging cores planning such a study should decide the following before sectioning, because the sampling scheme cannot be repaired afterwards:
- What the question needs. Absolute number, density, volume, length or surface area each call for a different probe (see the table).
- What the reference space is. The review warns of the “reference trap”: a density can change when the reference volume changes, even if the total number of particles is unchanged. Report the reference volume and, where possible, the total.
- How sections and fields will be sampled. Systematic random sampling gives each part of the structure an equal chance of selection.
- How counting rules will prevent double counting. Sections spaced too closely can count the same particle twice, and the disector counting rules exist to avoid overestimation.
- How sections will be cut. Thicker sections tend to inflate apparent area, and the effect is minimised by constant, thin sectioning.
Common misunderstandings
Counting profiles on a section gives particle number
It does not. The number of profiles on a section reflects particle size, shape and section thickness as well as number. This is why the disector counts a particle only where it first appears between two sections.
Thicker sections are harmless
According to the review, thicker sections tend to increase the apparent area, so sectioning should be thin and constant.
“Unbiased” means error-free
No. In stereology, “unbiased” describes a design that does not depend on shape or size assumptions. Random sampling still produces sampling variation, and the review also notes pitfalls such as Simpson’s paradox.
Density equals total number
Density depends on the reference space. If the reference volume changes, counts per section change although the total number may not.
FAQ
What is stereology in simple terms?
It is a way to estimate three-dimensional measures, such as volume or number, from samples taken through a structure, using sampling designs that make the estimate valid.
What is the difference between model-based and design-based stereology?
Model-based methods assume something about particle shape and size. Design-based methods use sampling designs that avoid those assumptions.
What is a disector?
A counting probe that compares a reference section with a look-up section and counts particles that first appear in one of them, so no assumption about particle size or shape is needed.
What does the Cavalieri estimator measure?
Volume, calculated as the sum of section areas multiplied by the spacing between sections.
Why does random sampling matter?
Each part of the sample space needs an equal probability of selection; without it the estimate loses the design-based guarantee.
For wider background on how samples are drawn, see sampling methods; for counting cells on a slide, see what a hemocytometer is.
Sources
- Lemley KV. “An Introduction to Stereology with Applications to the Glomerulus.” Glomerular Diseases 2021;1(4):294-301. DOI 10.1159/000519719. https://pmc.ncbi.nlm.nih.gov/articles/PMC9677735. Checked 2026-10-29.
- Stereology.info, an educational site sponsored by MBF Bioscience, a maker of stereology software. Used here only for definitions of design-based probes. https://www.stereology.info/. Checked 2026-10-29.
