Stereology in histology and pathology turns measurements from tissue sections into estimates of three-dimensional structure. Its value is practical: distinguishing a change in cell number from a change in cell size, measuring how much tissue a lesion occupies, or comparing structural damage across specimens without choosing fields by eye.
The starting point is not the microscope or software. It is the biological question. “More stained tissue,” “more cells,” and “a larger affected compartment” describe different outcomes and require different measurements. The methods introduced in stereology fundamentals help keep those outcomes separate.
This article focuses on choosing and interpreting stereological measurements in tissue research. It does not propose replacing diagnostic assessment with a single numerical result.
What Should a Histological Study Measure?
Before selecting sections, write the intended result in a sentence that includes its units and reference compartment. “Collagen volume fraction within renal cortex” is more useful than “fibrosis measurement.” “Total number of marker-defined cells in the sampled organ” is more informative than “cell count.”
| Research question | Suitable measurement | Main interpretive concern |
|---|---|---|
| How much of a compartment is occupied by a tissue component? | Volume fraction, often estimated by point counting | A fraction is not an absolute amount. |
| What is the total volume of a lesion? | Cavalieri volume estimate | Sampling must cover the lesion at known intervals. |
| How many cells or discrete structures are present? | Disector counting with a suitable sampling design | Section profiles are not individual objects. |
| How much tissue interface is present? | Surface estimation using suitable test lines | Section or probe orientation must be addressed. |
Number estimation deserves particular care. A large object has more opportunity to appear in a section than a small one, so counting visible profiles does not directly count objects in three dimensions. The original disector method established a three-dimensional counting rule that avoids assumptions about particle size and shape.
For a cell study, define the counting object too. A nuclear count should not automatically be called a cell count where multinucleated cells are part of the question.
Keep the Reference Compartment in View
A percentage needs a denominator. Collagen as a fraction of the entire section, collagen within parenchyma, and collagen within a lesion are different endpoints. Set the reference boundary before collecting measurements, including whether vessels, lumina, capsule, necrosis, or other compartments belong inside it.
Consider a hypothetical experiment. In one specimen, collagen occupies 10% of a reference compartment measuring 100 mm3. Its estimated collagen volume is 10 mm3. In another, collagen occupies 15% of a compartment measuring 50 mm3, giving 7.5 mm3. The second specimen has a higher collagen fraction but a lower absolute collagen volume.
Neither answer is wrong. They answer different questions. Report the fraction if tissue composition is the endpoint; estimate the reference volume as well if the question concerns total amount. Measurements multiplied together must refer to compatible tissue compartments and processing stages.
For equally weighted test points, the basic volume-fraction estimate is the number of points hitting the target divided by the number hitting the reference compartment. The point-counting guide covers the measurement procedure. In an application study, the harder decision is often what qualifies as target tissue.
Sample the Tissue, Not the Most Persuasive Image
Build the sampling plan from the specimen down: tissue slabs, blocks, sections, and microscope fields. For systematic uniform random sampling, choose a random starting position within the first interval, then proceed at a fixed interval. Preserve the relevant sampling fractions whenever the final calculation requires them.
This distributes observations rather than concentrating them in convenient locations. The methodological basis for efficient sampling across these levels is developed in Gundersen and Jensen’s study of systematic sampling in stereology.
Write rules for damaged sections, tissue folds, missing fragments, and ambiguous boundaries before reviewing treatment groups. If a selected field cannot be measured, document why and follow a predefined response rather than moving to the nearest attractive field.
For patchy disease, consider defining separate compartments, such as lesion center, lesion margin, and surrounding tissue. Report them separately or combine their estimates using appropriate weights. Selecting equal numbers of fields from unequal compartments does not, by itself, produce a whole-specimen estimate.
A biopsy also sets a boundary on interpretation. Ask whether the intended conclusion concerns the sampled tissue or the whole organ. A careful analysis of a selected biopsy should not be presented as though the specimen had been drawn uniformly from the entire organ.
Applications in Histology and Pathology
Kidney: Separating Glomerular Number from Size
Renal research illustrates why number and size should be treated as separate outcomes. A study can ask how many glomeruli an organ contains, how large they are, or how much cortex they occupy. Those questions should not be collapsed into a count of glomerular profiles per slide.
Fractionator sampling has been used to estimate glomerular number and size in human kidneys, including the human kidney study by Nyengaard and Bendtsen. This provides an organ-level application rather than a justification for extrapolating any biopsy count to the whole kidney.
When planning a renal study, state whether the endpoint is total glomerular number, glomerular volume, or a compartment fraction. If investigating interstitial fibrosis, define cortical boundaries and exclusions independently of the glomerular counting protocol. A convenient field for one measurement need not be an appropriate sample for another.
Liver: Distinguishing Tissue Fraction from Diagnostic Grade
For liver steatosis, the fraction of tissue occupied by fat vacuoles and the proportion of hepatocytes containing fat are different quantities. A cell containing one small vacuole and a cell occupied by a large vacuole may both count as affected cells, yet contribute very different amounts of fat-vacuole area.
A comparison of semiquantitative steatosis scoring and stereological point counting documented discrepancies between these approaches. The practical lesson is to name the measured endpoint rather than label every output “percent steatosis.”
Apply the same discipline to fibrosis research. Decide whether the measurement concerns all collagen-positive tissue or a defined compartment. If the question concerns interstitial deposition, do not allow routine inclusion of capsule or large vascular structures to determine the answer.
Keep numerical tissue measurements alongside architectural assessment rather than assuming one substitutes for the other. A study protocol should state how each contributes to the research question.
Lung: Volume, Surface, and Preparation Conditions
Lung stereology makes preparation conditions part of the measurement problem. Inflation and perfusion conditions affect the structural state being sampled. Alveolar surface area, tissue volume, and alveolar number also require different estimators; counting airspace profiles does not provide an alveolar count.
The ATS/ERS standards for quantitative assessment of lung structure address fixation, sampling, orientation, resolution, and reference volume. They also distinguish organ-level measurements from measurements made in biopsy material.
For a pulmonary study, define whether the endpoint is an absolute quantity or a quantity per unit reference volume. Keep inflation conditions comparable and use an orientation strategy appropriate to the surface or length measurement. An arbitrary section plane should not be assumed suitable for every structural endpoint.
Tumors: Measuring Burden Rather Than Selected Profiles
For tumor research, total lesion volume can answer a different question from lesion count or the diameter of the largest profile. Stereological sampling can distribute measurements through an organ instead of restricting assessment to its surface or one selected level.
A study of metastasis volume in mouse lungs used Cavalieri estimation across sampled tissue levels. It also showed why surface nodule counts could miss internal deposits and fail to account for differences in lesion size.
When designing a treatment-response experiment, decide whether viable tumor, necrosis, stroma, and the entire tumor bed need separate measurements. For example, a hypothetical reduction in viable-tumor fraction could accompany either a smaller or a larger tumor bed. Record the quantities needed to distinguish those possibilities rather than relying on a percentage alone.
Match Tissue Preparation to the Intended Measurement
Plan the estimator before committing all material to a processing route. For number estimation, ask whether the available preparation supports matched physical sections or optical sampling through sufficient tissue depth. For volume measurements, establish which processing stage the reported dimensions represent.
Do not substitute the microtome setting for a measurement of final section thickness where the calculation requires the latter. For optical counting, check visibility through the usable depth and justify any excluded surface zones. The guide to section thickness, guard zones, and tissue shrinkage covers these preparation issues.
Build identification checks into the pilot. Can observers distinguish the target from neighboring structures? Are the same boundaries recognizable across specimens? Is the staining adequate throughout the counting region?
For archived material, start with an inventory of what remains: blocks, section sequences, thickness records, sampling locations, and preparation notes. Choose an endpoint those materials can support. Do not retrofit a total-organ claim onto a collection whose original sampling history is unknown.
Digital Pathology and Automated Measurements
Use automation to carry out a defined measurement task, then test whether its output agrees with that task. A published evaluation of automated steatosis quantification against stereological point counting compared image-analysis methods with human assessments and examined disagreement between observers. Ambiguous vacuole boundaries were one source of disagreement.
For a new study, retain separate checks for sampling and classification. Does the image set represent the intended compartment? Does the algorithm label the correct tissue within those images? A convincing answer to the second question does not settle the first.
Keep training and evaluation specimens separate. Review difficult cases, not just clean examples, and compare specimen-level outputs as well as image-level agreement. If an automated result is an area fraction or a profile count, preserve that label unless the study design supports a three-dimensional interpretation. Software should not upgrade the meaning of the measurement by changing its column heading.
Report Results at the Level the Study Supports
Plan the reporting alongside the sampling. Readers should be able to identify the biological unit, reconstruct the route from specimen to field, and understand what each reported number represents. The guide to reporting stereological methods and results provides a broader framework.
For a histology or pathology application, include at least:
- The target structure, reference compartment, and measurement units.
- The specimen-selection process and sampling stages.
- The estimator, counting rules, and relevant sampling fractions.
- Preparation conditions, exclusion rules, and observer masking.
- Specimen-level estimates and an appropriate assessment of precision.
Keep the number of biological specimens distinct from the number of sections or fields. In an animal-level experiment, hundreds of images from a few animals should not be presented as hundreds of independent biological replicates.
Use the pilot to decide where further effort is worthwhile: more specimens, broader tissue sampling, clearer identification, or more measurements within each sample. Set that decision against the study’s actual question.
The final interpretation should remain as precise as the measurement. Report a volume fraction as a fraction, a local estimate as local, and a total quantity only when the sampling and calculation support it. That discipline is what makes stereology useful in pathology: not more numbers, but numbers with a defensible biological meaning.