3D Confocal Imaging Exposes Hidden Bias in Cerebral Microhemorrhage Counts

Cerebral microhemorrhages, the pathologic substrate of cerebral microbleeds, have long been quantified with a method that researchers now show is less reliable than assumed. A team at the University of California, Irvine, has demonstrated that standard two-dimensional histological sectioning systematically misses a substantial fraction of these lesions, and that the size of the error depends directly on how often the tissue is sliced. The findings, published as a methods article in Frontiers in Neuroscience, offer a quantitative framework for interpreting microhemorrhage counts across studies that use different sampling protocols.

The conventional approach relies on Prussian blue staining of thin tissue sections to visualize cerebral microhemorrhages. Because this method provides a snapshot of lesions within a brain rather than a comprehensive count across the entire brain, investigators typically examine a small number of sections. The Irvine group, led by Bernard Choi of the Beckman Laser Institute and Mark Fisher, set out to measure exactly how much of the true lesion population such snapshots overlook.

Their solution was to bypass sectioning altogether. Using tissue optical clearing combined with confocal microscopy, the researchers imaged intact, optically cleared 1-millimeter-thick sections of brain tissue in three dimensions. This allowed them to visualize cerebral microhemorrhages throughout large tissue samples without physically cutting the brain into thin slices, providing a comprehensive three-dimensional reference against which simulated histological sampling could be judged.

When the team simulated a sparse sampling scheme — analyzing one 40-micrometer-thick section out of every six — the simulated protocol omitted 38 percent of individual microhemorrhages. The losses were not distributed evenly: smaller lesions were disproportionately likely to be missed. In other words, standard histology does not merely undercount; it preferentially erases the smallest lesions from the record.

Increasing the sampling frequency improved the picture but introduced a new problem. When one in every two or three sections was analyzed, the miss rate fell to between 13 and 20 percent, yet the likelihood of counting the same lesion multiple times rose. The authors report this trade-off as a measured observation of their simulated protocols, in which denser sectioning raises the chance that a single lesion is registered more than once, inflating the apparent lesion burden in the opposite direction.

These opposing errors — omission of lesions at low sampling frequency and repeat counting at high frequency — frame the quantification of cerebral microhemorrhages as a sampling problem with a genuine trade-off. The authors report that using 40-micrometer sections at a frequency of one in every four sections offers a balance between missed lesions, repeated counts, and the practical workload of analysis. They emphasize that this recommendation may change for different study designs or experimental models, and that the optimal protocol is not universal.

The technical significance of the work lies in its validation of tissue clearing as a quantitative tool rather than merely a visualization technique. By comparing comprehensive three-dimensional confocal counts with simulated two-dimensional sampling drawn from the same tissue, the study converts a methodological concern into concrete numbers. Researchers who have compared microhemorrhage burdens across animal models or treatment groups can now ask whether differences in sectioning frequency, rather than biology, might explain part of the observed variation.

The implications extend beyond the laboratory. If standard histology undercounts lesions by tens of percent and skews the size distribution toward larger lesions, then cross-study comparisons built on different sectioning protocols warrant careful scrutiny. The authors position their results as guidance for interpreting existing and future histological counts, not as a replacement for established methods.

Limitations of the study are worth noting. The work is a methods article, and the reported percentages derive from comparisons within the tissue samples the team imaged; the precise miss rates could differ in other models, brain regions, or disease contexts. The authors themselves note that their recommended sampling frequency is a starting point that may require adjustment for particular experimental designs.

Nevertheless, the message is clear and likely to resonate with laboratories quantifying vascular pathology: the number of microhemorrhages reported in a study is not an absolute quantity but a function of how the brain was sliced. With three-dimensional confocal imaging of cleared tissue now offering a way to see the whole lesion population, the field has both a tool and a quantitative benchmark for making histological counts of cerebral microhemorrhages more comparable and more transparent.

Subject of Research: Quantitative sampling bias in histological counting of cerebral microhemorrhages assessed by 3D confocal imaging of optically cleared brain tissue

Article Title: 3D confocal imaging reveals quantitative bias in cerebral microhemorrhage counts from standard histology

Article References: Xie, D., Crouzet, C., Lau, W. L., Cribbs, D., Fisher, M., & Choi, B. (2026). 3D confocal imaging reveals quantitative bias in cerebral microhemorrhage counts from standard histology. Frontiers in Neuroscience, 20, 1940336. https://doi.org/10.3389/fnins.2026.1940336

Image Credits: AI Generated

DOI: 10.3389/fnins.2026.1940336

Keywords: cerebral microhemorrhages, cerebral microbleeds, confocal microscopy, tissue optical clearing, quantitative histology, sampling bias, Prussian blue staining, neuroscience methods, brain imaging, histology, confocal, imaging