I have '.tif' image stacks that I am analyzing as volumes. For every object, I can get the Volume, surface area:volume, 'sphericity' and Euler number. For every one of those features, the objects I am interested in falls within a normal distribution.
Doing a PCA reveals that 99.8 of the variance can be explained by the volume. But given the number of objects being in the thousands I still need to use the other features to eliminate non-relevant items.
Having those 4 different features/normal distributions, what statistical analysis can I do to isolate my objects of interests with a certain degree of confidence. I come from a purely biological background so apologies if my question is not very clear.