Determining the actual scale factor ...
hauke.heibel at googlemail.com
Mon Dec 6 07:56:13 EST 2010
After reconsidering the topic I think it is in fact OK to compute the
scale based on the pixel spacing by using the reference values from
the DICOM (Aperio?) spec. I am quite sure this is fine since
a) the zoom is typically only in user interfaces of interest and used
in order to provide functionality which is keeping the software close
to the well known microscope based workflow
b) quantifications in histological images are relying on the actual
pixel spacing and not the zoom
That means one could compute the zoom factor as
zoom = 40 * 0.25 / pixel spacing
(well, assuming the pixel spacing is square). This leads us to the
point where we need to pixel spacing more desperately than the zoom
On Fri, Dec 3, 2010 at 2:32 PM, Adam Goode <adam at spicenitz.org> wrote:
> On Thu, 02 Dec 2010 17:27 +0100, "Hauke Heibel"
> <hauke.heibel at googlemail.com> wrote:
>> On Thu, Dec 2, 2010 at 4:00 PM, Mitko Veta <mitko.veta at gmail.com> wrote:
>> > Image up-sampling increases the spatial resolution (although no new
>> > information is added). Yes, my example was for the opposite situation.
>> > I would expect that if you use low magnification and fine sampling, a
>> > lot of the information in the image will be redundant (the image will
>> > be blurry). Again, I might be wrong.
>> And again, true, the scanner vendors could implement such
>> functionality. Disregarding for a moment the fact that this would be
>> scary and bad behavior, it would also mean that the actual objective
>> power were not corresponding to the 0th pyramid level and thus
>> indicating the need to store this additionally required information
> Part of the reason this was never implemented in OpenSlide was because I
> never found the metadata stored in the files to be terribly accurate or
> trustworthy. I think Trestle was particularly bad with being off by a
> factor of 2 with this measurement. I think this was related to the fact
> that while the objective magnification was recorded, the "virtual
> eyepiece" was not. Or something. It's all a little vague.
> Also, according to DICOM (really Aperio, I think),
> 1X == 10 MPP. 2X == 5 MPP, etc.
More information about the openslide-users