Efficient determination of axon number in the optic nerve: A stereological approach

Abstract

Quantifying the number of axons in the optic nerve is of interest in many research questions. Here, we show that a stereological method allows simple, efficient, precise and unbiased determination of the total axon number in the murine optic nerve. Axons in semi-thin optic nerve cross sections from untreated eyes (n = 21) and eyes subjected to retinal damage by intravitreous NMDA injections (n = 32) or PBS controls (n = 5) were manually identified, counted and digitally labeled by hand. A stereological procedure was empirically tested with systematic combinations of different sampling methods (simple random sampling without replacement, systematic uniform random sampling, stratified random sampling) and sampling parameters. Extensive numerical Monte Carlo experiments were performed to evaluate their large-sample properties. Our results demonstrate reliable determination of total axon number and superior performance compared to other methods at a small fraction of the time required for a full manual count. We specify suitable sampling parameters for the adoption of an efficient stereological sampling scheme, give empirical estimates of the additionally introduced sampling variance to facilitate experimental planning, and offer AxonCounter, an easy-to-use plugin implementing these stereological methods for the multi-platform image processing application NIH ImageJ.

Keywords: Axon number estimation; ImageJ; Monte Carlo; Optic nerve; Sampling; Stereology.