Sources of variation in the computer-assisted motion analysis of rat epididymal sperm

Random and nonrandom factors associated with sample preparation and the automated analysis (CellSoft) of rat cauda epididymal sperm motion were studied. Random factors included inherent system variation at both the individual cell level and at the multiple cell level. Repeated analyses of identical tracks across grey level revealed a statistical interaction between grey settings and curvilinear velocity. However, in multiple track analyses, grey level was seen to be a factor only at higher settings. Nonrandom factors included time after sample preparation, dilution medium, and sample preparation procedures. Using a nicked preparation of the entire cauda epididymis from Long-Evans rats, the effects of time were studied on sperm suspended in 1) phosphate-buffered saline + 10 mg BSA/mL, 2) TEST yolk buffer, and 3) Medium 199. In PBS/BSA, the percent motile sperm estimate decreased (50% to 30%) over an hour, while the curvilinear velocity increased (127 to 142 microns/sec). Both sperm motion parameters were maintained in the TEST yolk buffer and in the Medium 199, although at lower values for the latter. Evaluation of the relative contribution of several factors, nested within sample, to the overall variance of three separate motion endpoints revealed that there was a large variation from field to field, negligible variation between overall CellSoft analyses of 200 cells or more, low variation at the preparation aliquot level, and moderate variation at the animal level. In planning experiments to test for effects on sperm motion endpoints, consideration of the relative contribution of the individual study factors to the overall variance of the parameter estimates will result in more sensitive experimental designs.