PET image are not displayed properly

[mokudo]

PET images may not display properly. Coarse image is displayed. DCMTK 3.6.0 is used. someone help me?
http://dl.dropbox.com/u/1794792/test.dcm

[J. Riesmeier]

What do you mean by "not display properly"? After applying the Modality LUT transformation, the pixel value range is [0..18] and the stored VOI window says WindowCenter=2.5 and WindowWidth=5.0 - so there's not much contrast in this image.

[mokudo]

Thank you for your reply.

"test.dcm"
To convert the BMP image.

DCMK3.6.0 is used.
http://dl.dropbox.com/u/1794792/bad.bmp

Other tools is used.
http://dl.dropbox.com/u/1794792/good.bmp

A bug report from the hospital.
Why use the DCMTK, good.bmp will show?

test.dcm tried to look at.
I did not understand it.

This tag is the impact that I think have been
(0028,1053) DS [0.01] # 4, 1 RescaleSlope

Please help me.

[J. Riesmeier]

Then you should probably disable the modality transform (dcm2pnm -M) and select an appropriate VOI window (dcm2pnm +Ww <center> <width>). The problem is that DCMTK maps the result of the modality transform to integer values, i.e. 0, 1, 2, ... 18.

[mokudo]

Thank you for your reply.

To disable the modalities have converted (dcm2pnm -M). and (dcm2pnm +Ww 200 400). This was a process. Good.bmp, and the same image.

However, this process can be automated?
The float process DCMTK plans to change?

Please tell me.Part of the problem? Source code.

[J. Riesmeier]

The float process DCMTK plans to change?

No, there are currently no plans to add support for an intermediate pixel data representation based on floating point values. However, it should be possible to extend the code accordingly since all transformations are based on C++ templates.

The problem with your DICOM image is - as explained before - that the modality transform scales the stored pixel values down to a value range that only has 19 different (integer) values. One solution is to switch off this modality transform (and to use an appropriate VOI window), another solution could be to extend the code of dcmimgle for using floating point values ...

[mokudo]

Thank you provide information.