Quantitative anatomy using design-based stereology

Abstract

In 1961, scientists from biology, geology, and material sciences convened a small meeting at a mountaintop retreat in Germany. Their purpose was to discuss an obscure problem common to all of their research: how to quantify the morphological features of arbitrary-shaped 3-D objects based on their 2-D appearance on cut surfaces. To describe this topic, they selected the term stereology, from the Greek stereos for “the study of objects in three dimensions,” and a year later formed the International Society for Stereology (ISS). After more than a decade of debating the relative merits of different stereology approaches in their respective fields of study, the ISS categorically rejected older methods based on assumption-and model-based Euclidean geometry in favor of innovative new methods designed to avoid methodological bias. Today unbiased or design-based stereology is the sine qua non for the reliable quantification of structure in many fields of biological and biomedical research. This chapter examines key concepts of design-based stereology using systematic random sampling and objective geometry probes in comparison with biased methods. To exemplify an application to neurosciences (neurostereology), we review the results and discussion from our recently published study to quantify differences in postmortem brains from autistic and normal children. Continuing advances in computer-assisted microscopy, image segmentation, and whole slide scanning support the growing trend toward higher throughput and greater efficiency in computerized stereology, with the promise of fully automatic and accurate stereoanalysis of tissue sections and in vivo images in the near future.