Using DICOM attribute information to determine the range of bone pixel values

Question

I'm working with DICOM CT images to segment bones. I was wondering if I can use DICOM attribute information to determine the range of bone pixel values.

Answer 1

Chapter 3, Annex A.3 of the DICOM standard [1] specifies the IOD for the CT dataset. It consists, among others, of the Image Pixel Module (C.7.6.3). The Pixel Module has two attributes "Smallest Image Pixel Value" (0028,0106) and "Largest Image Pixel Value" (0028,0107). However, these are Type 3 attributes and might not be present in your dataset.

[1] http://medical.nema.org/medical/dicom/current/output/pdf/part03.pdf

Answer 2

You should never rely on Type 3 attributes. In your specific case (CT) it is just as trivial as reading the Rescale Slope and Rescale Intercept. Those attributes allow to transform the pixel values to HU or other units, as specified in the attribute Rescale Type.

For CT images, the unit should be HU (Hounsfield). So after the transformation you can then lookup the well know values for bones:

http://en.wikipedia.org/wiki/Hounsfield_scale#The_HU_of_common_substances