A stereology study should begin with the quantity you need to estimate, not the slides already available or the software installed on the microscope. Define the target population, reference region, sampling design and measurement rules before processing specimens. Then test the complete workflow in a pilot study before committing the main study material.
This guide covers the decisions that connect study design, specimen preparation, microscopy, estimation and reporting. The aim is a practical protocol: one that another observer can follow, that preserves the sampling probabilities, and that produces an answer to the original research question.
Define the Outcome and Reference Region
Replace broad aims such as “assess tissue damage” with a measurable outcome. Suitable outcomes might include total neuron number in a defined brain region, the volume fraction of fibrosis within a tissue compartment, or total surface area within an organ. State whether the outcome is a total, a density, a fraction or a mean particle size.
These quantities answer different questions. A density can change because the target quantity changes, because the reference volume changes, or both. The ATS/ERS standards for quantitative assessment of lung structure establish the importance of an appropriate reference space and of relating density measurements to reference volume when estimating totals.
Consider a hypothetical region containing 100,000 cells in 10 mm3. Its numerical density is 10,000 cells/mm3. If the same cells occupy 8 mm3, density becomes 12,500 cells/mm3. That is a 25% density increase without an increase in cell number.
Write boundary rules before collecting measurements. Identify anatomical landmarks, treatment of cavities and lesions, and whether the target includes one side or both. For an archived biopsy, distinguish what can be estimated within the biopsy from what the sampling history permits you to infer about the organ.
Separate Experimental Replication from Tissue Sampling
The experimental unit is not automatically the microscope field. In an animal experiment, it might be an animal, a litter or a cage, depending on treatment allocation. Sections and fields sampled within that unit provide measurements of it; they do not create additional independent treatment replicates. The ARRIVE 2.0 guidelines for animal research require clear reporting of experimental units, sample size, exclusions, randomization and blinding.
For planning purposes, keep two separate budgets: the number of independent units and the measurement effort within each unit. Discuss the intended comparison and sample size justification with the person responsible for statistical analysis before specimen collection.
Write the allocation and masking procedures into the protocol. Decide who retains the group codes, who processes specimens and who assesses images. Where practical, distribute groups across processing batches and measurement sessions rather than processing one group completely before starting another.
Set exclusion criteria in advance. Describe what happens if a specimen lacks the target region, a stain fails, or sections are damaged. Keep exclusions and their reasons in the study record, including cases that never reach the microscope.
Choose the Estimator Before Choosing the Preparation
Match the estimator to the outcome and then determine what material it requires. This decision should precede embedding, cutting and imaging.
| Intended outcome | Possible approach | Planning requirement |
|---|---|---|
| Volume of a defined region | Cavalieri estimation | Sections spanning the region with known sampling intervals |
| Volume fraction of a component | Point counting | Defined component and reference space, with suitable spatial sampling |
| Total cell or particle number | Disector counting with fractionator sampling, or numerical density combined with reference volume | Reliable object identification and the measurements required by the chosen design |
| Surface area or length | Appropriate intersection probes | A compatible orientation design and adequate resolution |
For particle number, the Acta Stereologica paper on particle number estimation distinguishes disector and fractionator designs and their practical requirements. A physical disector needs identifiable corresponding objects in paired sections. An optical disector needs reliable observation through section depth. Counting profiles in a single image is not an interchangeable substitute.
Record the intended unit of recognition too: a cell, nucleus or another identifiable feature. If the counted feature stands in for a cell, state and justify that relationship rather than leaving it implicit.
Map the Sampling Route Through the Specimen
Draw the sampling sequence from the independent unit down to the final observation. A tissue study might proceed through organ, slabs, blocks, sections and microscope fields. At every stage, specify how selection occurs and what information must be retained to calculate the estimate.
Systematic sampling distributes observations across the material rather than concentrating them in convenient locations. Its efficiency across stereological sampling stages is developed in Gundersen and Jensen’s research on systematic sampling. For a simple section series, choose a uniform random start within the first interval and then sample at the fixed interval throughout the reference region.
As a worked planning example, sampling every tenth section after a random start of seven gives sections 7, 17, 27 and so on. Keep the original section numbering and document losses. Do not silently substitute the nearest attractive section when a selected section is damaged.
Specify field placement independently of what the fields contain. Also record orientation requirements: randomizing position does not randomize direction. Surface and length measurements may need orientation procedures that a volume estimate does not.
Use the Pilot to Test the Whole Procedure
Run pilot specimens through the actual fixation, sectioning, staining, mounting and imaging sequence. Include material expected to challenge the method, such as a small target region, sparse labeling or altered tissue architecture. Do not optimize only on the easiest specimen.
Record practical outputs: usable sections, sampled fields, countable events, measurement time, boundary disagreements and technical failures. Test whether the exported data contain enough detail to reproduce the estimate. Discovering a missing sampling fraction after the main count is an avoidable problem.
Use the pilot to choose section intervals, field spacing, probe dimensions and counting effort. Treat these as tested settings for this preparation, not universal settings for an organ or species.
For optical counting, measure thickness after processing and examine object visibility through depth. Research on section compression, lost particles and optical disector bias demonstrates why particle distributions through the section should be checked before choosing guard zones. A clear surface image alone does not validate the usable counting depth.
Finish the pilot with a dated protocol version and a decision record. If pilot observations will enter the final analysis, decide that before inspecting group differences and confirm that their procedures remain compatible with the main study.
Control Preparation and Preserve Specimen Identity
Create a specimen tracking sheet that follows each unit through collection, fixation, embedding, sectioning, staining and storage. Record processing times, batch identifiers, section order and departures from the protocol. Photographs or diagrams of the sampled specimen can help connect the physical material to the sampling record.
Use the specimen preparation, sectioning and staining guide when developing the laboratory procedure. At the workflow level, require documented acceptance checks before releasing material for measurement: recognizable boundaries, an interpretable target label, intact sampled regions and traceable section identities.
Keep reserve material where feasible, but define its intended use. A reserve series can support troubleshooting; it should not become a pool from which observers select whichever sections produce the most convenient result.
Treat Thickness and Guard Zones as Measured Decisions
Keep nominal cutting thickness separate from measured mounted thickness. Record the thickness observations used by the estimator, the disector height and the placement of any guard zones. Specify the response to a site that is too thin or cannot be read reliably through the intended depth.
The detailed procedures belong in the guide to section thickness, guard zones and tissue shrinkage. For the study protocol, retain the evidence behind the chosen settings and require a fresh check when preparation conditions change.
Do not apply a single shrinkage correction by habit. Identify which dimensions matter to the estimator, which processing stage the measurements describe, and whether the available observations support a correction.
Qualify the Imaging System and Train Observers
Before collecting study data, check spatial calibration, probe dimensions and the ability to resolve the counting feature. For optical methods, include the depth measurement system. Save the instrument configuration and software version alongside the protocol.
Choose microscopes, imaging systems and stereology software around the measurement requirements rather than a feature list. Require exports that preserve counts, sampled locations, relevant dimensions and sampling settings, not just the final result.
Train observers on shared practice material. Ask them to draw the same boundaries and classify the same difficult objects independently, then resolve disagreements before the main study. Save annotated examples of inclusion, exclusion and ambiguous cases. Establish a review procedure for uncertainty rather than asking observers to improvise.
Run Measurements with a Visible Audit Trail
Use coded specimen identifiers and a predefined measurement order. Begin each session with the agreed checks and log any instrument or software changes. Preserve the sampling locations even when a field contains no eligible objects.
Keep technical failure separate from a genuine zero. For each unreadable site, record what failed and apply the predefined response. Do not move the frame to a nearby region simply because it contains clearer cells.
Schedule repeat assessments of selected coded material to check observer consistency. Review discrepancies while the original images and specimens remain accessible. If the counting rule needs revision, document the change and decide whether earlier material needs reassessment.
Back up raw observations, contours, image identifiers and settings during collection. A spreadsheet of final estimates is not a substitute for the observations that produced them.
Review Estimates and Precision Before Comparing Groups
Calculate an estimate for each independent unit using its recorded sampling information. Check identifiers, units, missing records and the estimator implementation before looking at group differences. Where sampling fractions vary, use the appropriate weighting rather than treating all observations as interchangeable.
Choose a coefficient of error calculation suited to the estimator and sampling design. Research on variance estimation under systematic stereological sampling shows that the appropriate procedure depends on sampling density. A software output labeled “CE” still needs a named method and an explanation of its use.
Review precision alongside counts, spatial coverage and technical quality. A small CE does not establish that the reference region was correctly drawn or that the target was consistently recognized.
Use the guide to bias, precision and coefficients of error to distinguish sampling uncertainty from variation among independent units. Then apply the planned statistical analysis at the correct level. Keep any additional analyses clearly separate from the original plan.
Build the Report from the Study Record
Write the methods from the working protocol and deviation log, not from memory. Describe the reference region, independent units, preparation, selection procedures, estimator, recognition rules and measurement settings. Include how damaged material, uncertain objects and missing observations were handled.
Present results with units and a clear account of what each value represents. Distinguish estimated totals from raw counts, and sampling precision from variation between specimens. Explain departures from the original plan without hiding them inside a general methods paragraph.
The guide to reporting stereological methods and results provides the next level of detail. Archive the protocol version, sampling records, raw observations and analysis files together so that the reported estimates can be reconstructed.
A useful final check is straightforward: can another researcher follow one specimen from its selection to its reported result? If the answer requires guessing which sections were used, which boundaries applied or how counts were expanded, the study record is not yet complete.