This column is, perhaps, a departure from many techniques papers because the accompanying image chosen as “representative“ really is representative, rather than what is seen in many published papers in which the representative image is the one perfect image among hundreds of those that are biologically messy. The image used in this article to illustrate methods is not the best to use when creating regions of interest with Photoshop toots, for it contains elements which make the separation into regions of interest difficult to accomplish without reacquiring the image using better microscope techniques. I'm assuming that an image of this sort CAN bereacquired; if that isn't possible, some head-scratching and, perhaps a shift from a position of absolute accuracy to statistical accuracy might occur, and many would argue, should occur (since the whole matter is statistical anyway). Hopefully the investigator understands that, in some instances, human error can be as confounding as image processing error, except that image processing can create consistency. In other instances, artifacts in the images themselves cart cause too many wrong features to be measured, leaving humans as the best instrument.