1) Is it legal to have a GSPS object for a US image if Photometric
interpretation="RGB" and  "Ultrasound Color Data Present"= 00
( which says the pixels are grayscale encoded as RGB) ?

2).  Is it legal to have a VOI LUT in a GSPS object for  US IOD, since
the US IOD description says that the VOI LUT IE is not a component of
the US IOD?

Hi Del

DelFarmer wrote:
> 1) Is it legal to have a GSPS object for a US image if Photometric
> interpretation="RGB" and  "Ultrasound Color Data Present"= 00
> ( which says the pixels are grayscale encoded as RGB) ?

No.

You could use a color presentation state object, but that provides
no control over contrast.

You could make a derived copy of the RGB US image that was MONOCHROME2,
and then create a grayscale presentation state that applied to
that, but then your "no color" images would appear different in
terms of grayscale contrast from those that actually do have
a little color in them.

> 2).  Is it legal to have a VOI LUT in a GSPS object for US IOD, since
> the US IOD description says that the VOI LUT IE is not a component of
> the US IOD?

If the referenced US SOP Class is MONOCHROME2 or MONOCHROME1, then
yes, since the presentation state is independent of the IOD of the
referenced image, and replaces the entire grayscale or color
display pipeline ... all it depends on from the referenced image
instance is the stored pixel data values.

Some time ago, there was an attempt by some ultrasound folks to
convince WG 6 and WG 11 that the GSDF should apply to the
luminance component in color ultrasound images, and early
versions of CP 399 reflected that. The CP was initially
called "GSDF in Color Photometric Interpretations", however,
WG 6 did not agree and the CP was narrowed down to address
only the matter of color display calibration, and not color
photometric interpretations.

The wording in early versions of that CP talked about the
"luminance component", because it was trying to address the
difficult problem of how to address the displayed contrast
of the grayscale component of color images, whether color
information was present or not. I.e., regardless of whether
the use is looking at an RGB image with no color present
or with color Doppler superimposed, the grayscale structural
information should have the same perceived contrast.

The bottom line is that DICOM has no way to handle the persistence
of grayscale contrast transformations for mixed grayscale and
color ultrasound color images. Not to say of course that a
clever viewer cannot separate the two somehow and provide
the function to the user, just that you can't save it as
a presentation state.

The answer in the long term of course is to create better
ultrasound images that convey what the signals mean not what
they look like in a pseudo-screenshot, but that doesn't
help with the installed base.

David

On Nov 18, 8:13=A0am, David Clunie <dclu...@dclunie.com> wrote:
>
> The answer in the long term of course is to create better
> ultrasound images that convey what the signals mean not what
> they look like in a pseudo-screenshot, but that doesn't
> help with the installed base.
>
> David


Thanks David,

We've got some legacy US objects that are photo interp=3DRGB and do not
display properly because of an associated GSPS.  The GSPS objects have
bogus values for VOI LUT in them e.g. Window Center=3D0 and nothing
else.  We were trying to determine if we could throw these particular
GSPS objects away.

I think we can make a good decision now.

Thanks again.