NIfTI header conversion to Hounsfield units


I have converted a series of CT images that were originally DICOM files into NifTI, which I am using to derive intensity values. I have a range of intensity values that I am looking to convert to Hounsfield units. I understand that the linear transformation to convert from intensity to Hounsfield is typically stored in a DICOM header under slope and intercept. However, in the NiFTI file, these header categories are set to NaN (not a number.)

scl_slope : nan
scl_inter : nan

Is there a way to find the slope an intercept from a NifTI file to convert from intensity to Hounsfield units?

What converter did you use to convert the DICOM file to NIfTI, and what tool are you using to read the values from the NIfTI header? I would suggest you use dcm2niix - it should not only preserve the scaling values but will also deal with the common situation that CT scans often have slices with unequal thickness as well as the fact that the slice direction is often not orthogonal with the imaging plane. See the dcm2niix manual. A correctly converted image should have air near -1000, water (CSF) near 0, soft tissue near 30 and bone near 1000. You can use fslhd to show the scale intercept and slope values, or a tool like MRIcroGL should show you the image intensity in the titlebar when you click on a voxel in the 2D views.

Hi Chris,

Thanks for the tips. I was using a commercial software (similar to ImageJ) provided by my home institution (BioImageSuite) for the conversion, which was done at the same time as a linear registration to an MRI. I am worried that the scaling values were altered and I am not able to verify this within the software so I will repeat this again with dcm2niix and verify with MRIcroGL.