Dear Tracy,
Regarding your first question you have to note that SPM uses a similar approach to DICOM to store intensities in nii files when using int16 datatype. The analyze/nifti file also store slope/intercept data in the header to get the proper intensities while using only 16 bit.
Regarding your second question it's pretty clear that you're experiencing something that shouldn't be happening. The values are not being stored correctly. Most likely the slope/intercept are being computed on the first frame and then applied to the other ones, which have higher intensities.
I suggest you either force the data to be stored in float32 when converting from 3D to 4D.
Otherwise I hope that somebody else that experienced the same problem might know why it's happening.
Have a nice day,
Luca
________________________________________
Da: SPM (Statistical Parametric Mapping) <[log in to unmask]> per conto di Tracy Ssali <[log in to unmask]>
Inviato: martedì 7 giugno 2016 21.13.04
A: [log in to unmask]
Oggetto: [SPM] DICOM datatype and precision
Hello SPM users,
I am working with PET data and am having some trouble with interpreting the intensities. I suspect it is related to the precision of the datatype. To my understanding, the RescaleSlope and RescaleIntercept need to be applied to get the correct intensities.
I noticed that the SPM DICOM importer, saves the data as the datatype: int16. However, in SPM Display, the intensity reported is not a int16 datatype (i.e. there are intensities greater than 32768 displayed in the image). How does this work? Is the datatype actually float32 and incorrectly being displayed as int16?
This becomes a problem when I convert from 3D to 4D NIFTY. All the intensities in the 3D NIFTY image which are greater than 1822.09 (Which is 2^15, the maximum intensity for int16, scaled by the RescaleSlope (~.055)) a reported as 1822.09 in the 4D NIFTY image. All Intensities below that value are identical to the 3D NIFTY. During the 3D to 4D conversion, is the image saved as an int16 datatype and then rescaled?
Is it possible to DICOM import the data as float32 instead of int16 to avoid the loss of precision? Or are there any other ways around this loss in precision?
Any clarification would be appreciated.
Thanks,
Tracy
Rispetta l’ambiente: non stampare questa mail se non è necessario.
Respect the environment: if it's not necessary, don't print this mail.
IL TUO 5XMILLE AL SAN RAFFAELE DI MILANO
Devolvi il tuo 5 per mille all’Ospedale San Raffaele: perché al centro della Ricerca ci sei TU. CODICE FISCALE: 07636600962, nel riquadro RICERCA SANITARIA. Non c’è cura, senza ricerca. Non c’è ricerca, senza il tuo 5xmille. Scopri come su http://www.5xmille.org
[http://www.5xmille.org/img/731x129.jpg]
|