A method for producing unbiased histograms of neuronal profile sizes

An important goal in neuroscience is to produce frequency distribution curves or histograms that relate numbers of cells to their sizes. Unfortunately such histograms, which are extremely common, are biased. Some of the reasons are inadequate attention to sampling paradigms, the lack of assurance that the section through the center of the cell is measured, and that large cells will have more profiles than small cells. The first goal of this paper is to illustrate these biases for dorsal root ganglion cells by showing significant statistical differences between histograms prepared the classic way and histograms produced by unbiased methods. The differences are particularly dramatic for plastic embedded material. The second goal of the paper is to describe an unbiased method for obtaining these histograms. The procedure is 2-fold. Cells are chosen in an unbiased way (unbiased in this sense means that every cell has an equal chance of being chosen), and then the largest profile of the cell is measured. We further suggest that these histograms be accompanied by analyses of cell volume, since volumes of cells rather than diameters or areas of cell profiles, will be the future measure of choice when considering the sizes of cells (or any other particles of biologic interest).