 |
 |
Dear John and SPMers, I tried normalizing FDG PET images of two subjects using the new normalisation, and the results are pretty bad. My settings were: Bias FWHM: no corretion, Tissue probability map: spm12/tpm/TPM.nii, affine regularisation: European Brains, warping regularisation (default values), sampling distance: 3. I've tried using smoothness of 8 or 6 and the results are pretty identical. Basically, if I do a "check reg" loading the new normalisation, the old one and the TPM I can see that the old one gives results that overlap quite well over the whole brain (I used an FDG template). The new normalization instead is off by as much as 6 mm in many landmarks. Do you have any suggestion regarding how to check intermediate steps or how to set parameters differently in the new normalization? (BTW, the results are pretty identical applying the normalization to traditional images and to the ones reconstructed with resolution recovery algorithms, where the gray matter is quite well recognizable) Thank you very much, Luca ________________________________ Da: PRESOTTO LUCA Inviato: mercoledì 27 aprile 2016 18.40 A: John Ashburner Cc: [log in to unmask] Oggetto: R: [SPM] Normalizing PET in SPM12 Ok, thank you very much. I will be trying it then, but I think it's not going to be trivial and it's going to require a different pipeline from the one we're used to. I'm not probably going to do it quickly but I'll let the list up to date with the eventual problems. Thank you! Luca Inviata dal mio Windows Phone ________________________________ Da: John Ashburner<mailto:[log in to unmask]> Inviato: 27/04/2016 18:23 A: PRESOTTO LUCA<mailto:[log in to unmask]> Cc: [log in to unmask]<mailto:[log in to unmask]> Oggetto: Re: [SPM] Normalizing PET in SPM12 There is nothing to link to as far as I'm aware. There are relatively few PET users of SPM these days (on our SPM courses anyway), so support for PET is a bit less extensive than it used to be. I was reluctant to totally get rid of the old normalisation code yet (so it's there as a toolbox) as it may still prove to be better for PET. If users generally find that the new approach works better, then the old legacy code will be removed in a future release. Best regards, -John On 27 April 2016 at 16:49, PRESOTTO LUCA <[log in to unmask]<mailto:[log in to unmask]>> wrote: Thank you very much John Could you provide me with some links regarding how this new method should work with PET? Does it still attempt a segmentation even in PET? Should I be using the same tissue probability maps used for MRI? Did anybody try this method and compared it with the old normalization? I haven't found it cited yet. By the way, what should be my bias field correction regularization setting, to have it not running? Thank you again Luca ________________________________ Da: John Ashburner <[log in to unmask]<mailto:[log in to unmask]>> Inviato: mercoledì 27 aprile 2016 17.25.53 A: PRESOTTO LUCA Cc: [log in to unmask]<mailto:[log in to unmask]> Oggetto: Re: [SPM] Normalizing PET in SPM12 You are right. The settings for the default SPM12 spatial normalisation are not ideal for PET because of the intrinsic smoothness of PET images. One of the model assumptions is that noise is independent from voxel to voxel, which generally is not the case for PET. You can use the default normalisation for PET though, but it involves changing one of the settings on the user interface. This is the "smoothness", which you could try changing to (eg) 8 mm. Best regards, -John On 27 April 2016 at 15:12, PRESOTTO LUCA <[log in to unmask]<mailto:[log in to unmask]>> wrote: Hi there! I'm starting to use SPM12, and I'm working on PET images. However, if after opening SPM in the PET modality, I selected "normalise (estimate & write)". However, in SPM12, if I'm not mistaken, this module is not meant for PET and I have to use "old normalise". Did I get this right? Or can PET benefit also from the new normalise? (Unless I'm not mistake I'd say that the parameters are absolutely not meant for PET images) Thank you, Luca 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] 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] 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]