Also – how many objects/cells do you expect to see? You’re going to approach your sampling differently if you’ve got a sparse population vs a common population!Structure difference: e.g., rostral to caudal, add more coronal sectionsSection difference: e.g., medial to lateral difference, add more sites to your coronal sectionWork on your SN example description…big effect, you can accept greater sampling error, or be less conservative in your sampling parameter design. (so you can use a preferential section + a isotropic probe or vertical sections and a probe that is random in at least one dimension etc). 2nd ed., Bios, Oxford, 2005Sum up by sayinghtat remind the viewer that tissue + object + probe = isotropic to be unbiased. Geometric probes are required to quantify 3D data Howard CV, Reed MG: Unbiased Stereology. Section collection must be regularly sampled and available for analysisMaximize post-processing section thicknessMinimize damage due to processingStaining must penetrate the entire section thickness. Reduction of the sampling area of a section by a known area since counting all cells is prohibitive.Grid spacing is systematic, placement of grid is randomFor lung, often a technique call the Smooth Fractionator is performed instead of systematic random…. The goal of stereology is to ensure that the individual sampling error does not overshadow the difference due to experimental manipulation. 
Sampling design can yield high accuracy but low precision so that each estimate varies from each other, yet are clustered at the true number…Or it can be highly precise but low in accuracy – so replication can yield similar estimates which are not close to the true numberOr…it can be both precise and accurate with the estimates clustered together near the true number. With sampling, a given estimate of a population will vary from the true (and unknown) number.
0 kommentar(er)
