US IOD: Is it possible to encode multicolored (paletted) overlay?

[Yuri Volokhov]

Hi!

Indeed, I have two questions:

1. Now I'm trying to add some DICOM functionality to a couple of US
modalities. Each modality has multicolored overlay above the US image. Is it
possible to encode such overlays without loss of color information and how
can I do it?

My ideas:

a) burn overlay into image (but it's pretty bad that such "overlay" can't be
turned off in viewer;

b) give an overlay plane per each overlay color (can't be > 16 planes, right?
and, further, I haven't an idea how to encode the color of an overlay plane);

c) reduce my overlay to 1 bpp;

d) give the reduced or burned overlay for common use and encode multicolored
overlay in some private way for ours.

Have I missed something?

2. Do (60xx,****) overlay tags mean (6000,****) for the first overlay,
(6001,****) for second etc?

Thanks in advance.

Yuri

[David Clunie]

Hi Yuri

There is another alternative, to create a Grayscale Presentation
State, and encode the overlays in that, assigning each a value
for Graphic Layer Recommended Display RGB Value that is the color
you want it to be displayed.

That will work fine, if the referenced image is a grayscale image;
but in theory if the referenced images is a color image, then GSPS
doesn't apply. However, one could argue that this approach is still
the best of other alternatives even if the referenced image is a
color image, and either ignore the grayscale consistency information
in the GSPS, or apply it to the luminance component of the referenced
image ... this would work quite well with US color images, which are
generally grayscale anyway with a little extra color only for the
Doppler signal.

Going forward, GSPS is much more likely to be properly supported by
image viewing applications than is any creative variant of overlays,
or a purely private approach.

There was once in the standard a half-completed version of giving
overlays a particular color, but it was removed in CP 90 because
it didn't work (or wasn't clear how it should work).

david

[Doug Sluis]

I differ with David's recommendation.
In general, US images are color and GSPS does not apply
as David correctly points out.
The use of GSPS is a pretty heavy weight mechanism when
the only value is not to burn the graphics in the image.

The real value of GSPS is the use cases where presentation
related infomration (LUT, zoom, annotation, etc.) is routinely
created subsequent to image acquisition. This is simply
not the case for routine workflow in ultrasound or echo and
for this reason there is little enthusiasm for GSPS in ultrasound.

The simplest approach is to use attributes of the
Graphics Presentation Module. The only downside is that the image
becomes a Standard Extended SOP Class rather than a purely Standard one.
But Standard Extended SOP Class are frequently encountered.

- Doug
David Clunie <dcl...@dclunie.com> wrote in message news:<3D429147...@dclunie.com>...

[David Clunie]

What I was suggesting was a way to use standard SOP classes
to achieve the objective.

Using a standard extended SOP class as Doug suggests would
mean that image viewers that don't know about this extension
(which is effectively proprietary) would ignore the information,
which is presumably not the effect that you want - the information
might as well be in private attributes. I assumed you were looking
for a way to encode the information that other vendors would
be able to make use of.

Neither approach is perfect, and I agree with Doug that the
GSPS is certainly pretty "heavy weight", but on the other
hand that is the direction that the other modalities are
going, for what that is worth, as is IHE.

Of course the application of the GSPS to a color image is
also non-standard, so I certainly can't claim the moral high ground
here.

Either way, the standard needs to deal with such issues and no
doubt we will have such a discussion at upcoming working group
meetings as to the best way to handle it. There has been much
angst over the issue of color presentation states that are hung
up in the complexity of color consistency, and an intermediate
step may be necessary.

david