Multi tissue CSD and pooled b0s across shells

[jdtournier]

Hi all,

Sorry I’ve not responded to this thread so far, I’ve had other issues to deal with.

First off, this is an issue we’ve faced fairly routinely in my lab of late (mostly to deal with historical data), and it’s a trivial matter to scale each sequence so as to match the b=0 signal across separate series. Something like this ought to do the trick (although this is from memory, it might need amending):

scale=$(dwiextract -bzero dwi1.mif - | mrmath - -axis 3 - | mrstats - -mask brain_mask.mif -output median)
mrcalc dwi1.mif 1000 $scale -div -mult dwi1_scaled.mif

For each input series, following by mrcat as suggested previously:

mrcat -axis 3 dwi1_scaled.mif dwi2_scaled.mif ... dwi_corrected.mif

Sometimes that’s all you can do… And as mentioned above, it’s OK as long as the echo times match, but maybe not so cool if they don’t (although I have to admit, we’ve had to do that for some of our data too… ).

@Romain_Valabregue

I am working on siemens, and I am not sure that the receiver gain change from series to series (if FOV and sequence type is identical). I will ask our siemens engineer. Does any one know the dicom parameter to look at for a retrospective check on the data ?

Sure, look for RescaleSlope and RescaleIntercept in the dcminfo -a dicom_file output. Otherwise, just look at the output of mrinfo DICOM_folder/, the ‘Intensity scaling’ entry is lifted directly out of the DICOM headers.