Minor bug with Aperio Images with fractional objectives

[Derek Magee]

>Downsampling will affect the MPP, but shouldn't change the objective power, since that's a property of the optical path.  (See, for example,
>MIRAX/CMU-1-Saved-1_16.)

Depends how you define what ' objective power' means. Downsampling by a factor of two is (roughly) equivalent to using an objective of half the power, so I've always found it helpful to thing of say a downsampled 20X image as a 'virtual' 10X image. However, if it is defined as the ORIGINAL objective before processing then it should be left as is.

> I'd say even if you objected to non-integer objectives (I don't see 
> why), the Aperio strategy of Rounding at least gives some clue.  I 
> personally use it to work out the ratio between original and 
> downsampled versions (although I could just use mpp-x, or mpp-y 
> equally as well).  It's not a major problem for me now I know it happens, but it may trip up others.

>`openslide.objective-power` is documented to be optional.  We do typically add workarounds for issues likely to be seen in the wild, but I can't tell whether that applies in this case.  Thus my question: are the >affected slides created by vendor code, or by a custom image pipeline of some sort?

It's the output of image analysis in this case. However, I'm working with some people doing high-res blockface imaging at the moment and their highest resolution is 10X, and they often image at lower resolution. I'm not sure what selection of optics they have, but it's conceivable that there might be a non-integer lens/zoom (I'm not an expert on optics). At the end of the day a) it's not a huge issue for me personally, b) Does it hurt to parse non-integer objectives correctly?

Derek