1

I'd like to convert a set of DICOM instances (CT, MR, X-Ray basically) to JPEG (regular 8bits lossy grayscale). I have been starring at the options from dcmj2pnm:

I am quite confident that --use-window 1 is the right option when the input DICOM instance provide a single value for (0028,1050) Window Center & (0028,1051) Window Width. What I fail to understand is what is the "right" option to choose from (when WC/WW is missing):

  1. --min-max-window,
  2. --min-max-window-n (what does extreme values mean ?),
  3. --histogram-window 5 (found ref here)

Are there any good rule of thumb when processing DICOM CT (Pixel Padding Value) images ? MR images ? X-Ray images ?

malat
  • 12,152
  • 13
  • 89
  • 158

1 Answers1

4

Being the author of this DCMTK tool, I should be able to answer your questions.

I am quite confident that --use-window 1 is the right option when the input DICOM instance provide a single value for (0028,1050) Window Center & (0028,1051) Window Width.

This is at least true if the stored VOI window is correct (i.e. appropriate for the stored pixel data).

What I fail to understand is what is the "right" option to choose from (when WC/WW is missing):

As a rule of thumb, --min-max-window (compute VOI window using min-max algorithm) usually creates good results. If there are extreme values in the image (very low and/or very high pixel values) that do not belong to the medical content, they can be ignored: option --min-max-window-n ignores the minimum and maximum values (and use the next lower/higher value in the pixel data as the borders of the VOI window), option --histogram-window [n] computes a histogram on the pixel values used in the image and ignores "n" percent of both the low and the high values (when computing the VOI window).

J. Riesmeier
  • 1,641
  • 10
  • 14