search

spine-generic / data-multi-subject

Multi-subject data for the Spine Generic project
Creative Commons Attribution 4.0 International
22 stars 15 forks source link

Address scaling issues #58

Open jcohenadad opened 3 years ago

jcohenadad commented 3 years ago

Following on https://github.com/spine-generic/data-multi-subject/issues/36#issuecomment-688535998:

Some scans (notably on Philips) might have different signal quantization between the MTon/off and the T1w scan:

MT on/off:

    "PhilipsRWVSlope": 1.69499, 
    "PhilipsRWVIntercept": 0, 
    "PhilipsRescaleSlope": 1.69499,

T1w:

    "PhilipsRWVSlope": 1.85592, 
    "PhilipsRWVIntercept": 0, 
    "PhilipsRescaleSlope": 1.85592,

Here is a relevant comment from Guillaume Gilbert:

It is normal and expected that the Philips scale slope can be different between the MT on/off and T1w sequences. Some sites probably used a research sequence to keep it fixed, but it could in principle vary between the different sequences. Assuming that you have Classic (not Enhanced) Dicom images as the starting point, the easiest way to take the different scaling into account is to use dcm2niix for conversion and to make sure that the –p option is set to yes (which is the case by default). In that case, the scaling will be automatically applied and it should then be possible to compare the signal intensities of the two sequences. In the case of Enhanced Dicom, I have seen the scaling step fail with dcm2niix. In that case, it would be possible to apply the scaling manually by dividing the stored signal intensity by the scale slope.

This is problematic for computing MTsat, and the effect needs to be accounted for in the analysis pipeline.