The optical disector is a three-dimensional counting probe used to estimate the number of cells or other discrete particles in thick tissue sections. Instead of counting every profile visible in one image, the observer focuses through a measured depth and records objects when a defined feature first appears. A counting frame controls inclusion in the horizontal plane; two focal planes control inclusion through the section depth.
The distinction matters whenever cell size, shape or tissue volume may differ between specimens. More profiles in a microscope field do not necessarily mean more cells. The optical disector addresses that problem, but its performance depends on specimen quality, sampling and consistent counting rules. It is an optical implementation of the disector principle for counting particles, not a replacement for a complete study design.
How the Optical Disector Works
A single section can intersect a large particle over a greater range of positions than a small particle. Counting the resulting profiles therefore gives larger particles more opportunities to appear in the sample. The optical disector changes the event being counted: an eligible particle contributes once when its designated counting feature first becomes recognizable within the sampled depth.
The probe has a known cross-sectional area, a, and a measured height, h. Its volume is a × h. Within that volume, the observer follows a consistent focusing direction and applies rules that prevent repeated counting. The use of two optical planes inside a sufficiently transparent physical section is part of Gundersen’s original treatment of particle number estimation.
“Optical” does not mean that a confocal microscope is compulsory. The defining operation is observation through successive focal depths. The microscope must provide enough axial resolution to distinguish the counting event and enough working distance to inspect the required tissue depth.
Nor does the method require each counted cell to fit completely inside the probe. A nucleus can extend above or below the counting volume. What determines eligibility is its counting event and its relationship to the counting frame, not complete containment of the cell.
Define the Object Before Setting the Probe
Write down what qualifies as an object before collecting counts. “Count neurons” is not a sufficient operational definition. The protocol needs to describe the identifying stain or marker, the morphological criteria and the feature used to register one counting event.
A nucleus may provide a suitable counting unit, but a nuclear count is not automatically a cell count. If the target population includes multinucleated cells, that relationship needs separate justification. A nucleolus can also be problematic if cells contain more than one and the counting rule does not resolve that multiplicity.
For a hypothetical study of marker-positive cells, separate two decisions: whether the cell belongs to the target population, and whether its nuclear counting event qualifies for inclusion. A bright patch of cytoplasmic label should not become an extra object simply because it appears at another focal depth.
Preparation determines whether those decisions are practical. If the available material consists of thin serial sections rather than optically accessible thick sections, the physical disector may be the more appropriate implementation. Forcing an optical probe into unsuitable material does not make the estimate more rigorous.
Apply Counting Rules in All Three Dimensions
The counting frame
An unbiased counting frame has inclusion boundaries and exclusion boundaries, commonly displayed as two pairs of differently colored lines. Eligible profiles inside the frame or touching an inclusion line can be counted. Profiles touching an exclusion line, including its prescribed extension, are rejected.
Apply the frame rule to the designated counting unit, such as the nucleus, rather than switching between nuclear and whole-cell outlines. A profile touching both an inclusion and an exclusion boundary is excluded.
The counting height
With a convention that excludes the upper focal plane and includes the lower plane, reject objects already recognizable at the upper plane. Focus downward and count eligible objects first recognized after that plane, up to and including the lower plane. The Kreutz and Barger optical counting protocol provides a practical application of frame boundaries and counting through tissue depth.
The following hypothetical decisions assume that the object meets the identification criteria:
| Observation | Decision | Reason |
|---|---|---|
| A nucleus is already recognizable at the upper exclusion plane | Do not count | Its counting event is outside the accepted depth interval |
| A nucleus first appears within the interval and lies inside the frame | Count | Both depth and frame conditions are satisfied |
| A qualifying nuclear profile touches only an inclusion line | Count | The inclusion boundary permits acceptance |
| A qualifying nuclear profile touches an exclusion line | Do not count | The exclusion boundary overrides inclusion |
| A previously counted nucleus remains visible in later focal planes | Do not count again | Continued visibility is not a new counting event |
Train observers on difficult examples, including weakly labeled nuclei, overlapping profiles and boundary contacts. Resolve disagreements before the main count. A written rule that two observers interpret differently is still an unfinished rule.
Place Disectors Without Choosing the Cells
A correctly applied probe cannot rescue selective field placement. Choosing fields because they contain clear, isolated or strongly stained cells changes which objects can enter the sample. The same problem arises when an observer moves a frame slightly to avoid an awkward cluster.
Use a defined sampling design across sections and within the region of interest. Systematic uniform random sampling combines a random start with fixed sampling intervals. Establish the region boundaries independently of where the most convenient cells happen to be.
For example, if a predetermined grid position lands in a sparse part of the region, keep that position. A zero count is a valid observation. Replacing it with a nearby crowded field would turn a sample of tissue into a selection of cells.
Set procedures for damaged sections, folds and unmeasurable sites during the pilot stage. Record failures rather than silently deleting them. If poor visibility repeatedly affects one anatomical compartment or experimental group, the issue is not just lost counting time; it may change the population represented by the estimate.
Choose Guard Zones and Height From the Actual Tissue
Guard zones are the portions of section depth outside the counting interval. They are commonly used to avoid cut surfaces where particle fragments may be missing, damaged or difficult to recognize. The upper guard zone separates the tissue surface from the first counting plane; the lower guard zone separates the end of the disector from the opposite surface.
Do not adopt a guard-zone depth simply because another laboratory used it. Experimental work on differential shrinkage and guard-zone placement found that particle distributions through section depth can be uneven. Removing the surfaces does not automatically make the remaining tissue representative.
During a pilot, record counting-event depths across the section and inspect their distribution. Consider surface damage, uneven compression, staining penetration and optical visibility before choosing the interval. A central deficit in observed cells deserves investigation, not an automatic adjustment that makes the graph look flatter.
Use the thickness measured in the processed, mounted specimen. The microtome setting describes the cutting operation, not necessarily the tissue available for counting. The relationship is:
Upper guard zone + disector height + lower guard zone ≤ local section thickness.
For an illustrative site measuring 24 µm, a 3 µm upper guard zone and a 15 µm disector leave 6 µm below the probe. At a site measuring 19 µm, the same arrangement leaves only 1 µm. Whether either arrangement is acceptable depends on the validated surface exclusion requirements, not the arithmetic alone.
Plan how to handle such variation before data collection. Do not shorten probes informally while continuing to use the original sampled volume. The related guide to section thickness, guard zones and tissue shrinkage covers the measurement and estimator issues in greater detail.
Verify Visibility Throughout the Counting Depth
A cell that cannot be recognized cannot contribute a valid counting event. Check the preparation through depth rather than judging it from its best-looking focal plane. This applies to the object’s identity as well as its physical visibility.
Incomplete antibody penetration is a demonstrated risk. In experiments on thick brain sections, labeling could be weak or absent centrally, and extending secondary-antibody incubation corrected penetration for some markers but not others. The experimental study of immunolabeling penetration supports validating each labeling protocol rather than assuming that surface staining represents the section interior.
For multiple-label studies, inspect each channel used to classify the target population. Strong nuclear counterstaining does not prove that a phenotype marker penetrates equally well. Otherwise, a nucleus may be visible but wrongly classified as marker-negative.
A useful acceptance test is practical: can the observer recognize the smallest or weakest eligible objects at the least favorable depth used for counting? If not, revisit staining, optics or specimen preparation before enlarging the study.
For image-stack workflows, preserve the depth sequence and calibrated spacing. A projection that collapses the stack into one image removes the information needed to determine first appearance. Review uncertain events in the original focal planes rather than deciding from the projection.
Calculate Numerical Density Without Confusing It With Total Number
For equal-probability sampling with correctly measured probe volumes, numerical density is estimated from the accepted count divided by the sampled reference volume:
NV = ΣQ− / ΣVdis
Here, ΣQ− is the sum of accepted counting events, and ΣVdis is the summed disector volume within the defined reference compartment. For m identical probes lying entirely within that compartment, the denominator becomes m × a × h. Density and reference-volume estimation are combined in the Wu and colleagues neurostereology protocol.
A worked example
Suppose a hypothetical study uses 120 disectors, each with a frame area of 2,500 µm² and a height of 12 µm. All probes lie within the reference compartment, and 180 eligible nuclei are counted.
Each probe samples 30,000 µm³. The total sampled volume is therefore 3,600,000 µm³, equivalent to 0.0036 mm³. Dividing 180 by 0.0036 gives an estimated numerical density of 50,000 nuclei/mm³.
That result is a density, not the number of nuclei in the entire region. If a compatible reference-volume estimate were 8 mm³, the density-times-volume calculation would give 400,000 nuclei. The volume must refer to the same anatomical compartment and a compatible processing state.
The distinction can change interpretation. In a hypothetical region containing 100,000 cells, reducing volume from 2 mm³ to 1 mm³ doubles density from 50,000 to 100,000 cells/mm³ without adding a single cell.
The optical fractionator takes a different route to total number: it combines optical disector counts with known sampling fractions. The disector supplies the counting probe; the fractionator supplies the framework for expanding those counts to the whole region. The terms are related, not interchangeable.
Record Enough Detail to Reproduce the Count
Keep a specimen-level record that connects raw observations to the reported estimate. At minimum, document:
- The reference region, target population and counting feature.
- Section selection, field placement, frame area and disector height.
- Upper and lower guard zones, measured thicknesses and depth-validation results.
- Optical configuration, calibration and the inclusion/exclusion convention.
- Accepted counts, sampled volumes or fractions, and handling of failed sites.
- The estimator, precision assessment and observer quality checks.
Preserve counts by section and sampling site rather than retaining only a final total. Those records make it possible to investigate an unusual result, identify a thickness problem or check whether a small part of the region dominates the estimate.
The guide to reporting stereological methods and results addresses the wider study report. For the optical disector itself, the practical standard is straightforward: another trained observer should be able to identify the same target population, place the same kind of probe and apply the same counting decisions. A final number without those details is difficult to audit, however many decimal places the software supplies.