An automated system for counting silver grains in autoradiographs

D. F. Wann, J. L. Price, W. M. Cowan, M. A. Agulnek

Research output: Contribution to journalArticlepeer-review

23 Scopus citations


A system, incorporating a microscope, a television camera, and a small digital computer, has been developed for automated counting of silver grains in autoradiographs. The system has been designed specifically for analysis of experiments using the autoradiographic method of demonstrating axonal connections, although it should be adaptable to other uses. The system is capable of counting grains over stained sections up to a density of 300-500 grains/1000 sq. μm, depending on the grain size. Throughout most of this range, the accuracy of the counts is better than ± 10%. The counting algorithm is based upon the direct recognition of individual grains rather than on photometric measurements. This makes the counts relatively independent of grain size or shape, and permits accurate counts even though the grains occupy a layer of the emulsion thicker than the depth of focus of the objective. This is achieved by collating the counts obtained at different planes of focus on a grain-to-grain basis. Several automatic features, notably automatic focusing and stage movement, permit the counting of large areas of a section with minimal operator intervention. The system also includes features for the convenient identification of morphological landmarks in the section, and their correlation with the count data. There are several options available to the user which influence the accuracy and spatial resolutions of the output data and conversely the time required to count a large area. Because of these options the time required to count a 1 sq. mm area can vary from 2 to 8 h.

Original languageEnglish
Pages (from-to)31-58
Number of pages28
JournalBrain Research
Issue number1
StatePublished - Nov 29 1974


Dive into the research topics of 'An automated system for counting silver grains in autoradiographs'. Together they form a unique fingerprint.

Cite this