Pixel size at various layers

[Benjamin Gilbert]

On 03/13/2013 04:38 PM, ajwood at mta.ca wrote:
> At lever 0, we have square pixels:
> MICROMETER_PER_PIXEL_X = 0.369047619047619
> MICROMETER_PER_PIXEL_Y = 0.369047619047619
>
> At hight layers, we don't:
> MICROMETER_PER_PIXEL_X = 2.95018096657441
> MICROMETER_PER_PIXEL_Y = 2.94944618006248
>
> This seems wrong, no?

In the MIRAX file format, it's difficult to see how downsamples other 
than powers of two could make any sense.  Currently, 
openslide_get_level_downsample() will always report power-of-two 
downsamples on MIRAX slides, regardless of MICROMETER_PER_PIXEL_{X,Y}.

So why the discrepancy?  We could have an incomplete understanding of 
how MIRAX files are generated.  The MICROMETER_PER_PIXEL values could be 
wrong.  Either possibility is credible.

(For example, the vendor's software may ignore the stored MPP values for 
all levels except level 0, as OpenSlide does, and so an incorrect 
calculation may not have been noticed.  This would not be the first 
piece of write-only metadata in the MIRAX format.)

--Benjamin Gilbert