Dear Helmut,
I've tried running a number of segementation today. The gray matter gets segmented well, I'd say. Except that it's estimated to be super thick (~1 cm) instead that of the real thickness. But given FDG PET image proprieties I'd say that this isn't wrong from the point of view of a segmentation algorithm. That's for the first 3 classes, bone is segmented way too wide, including generally also the surrounding tissues, and instead class 5 is generally assigned to outside air.
Despite this fact that gray,white matter and CSF are segmented reasonably, the normalization isn't good at all. Generally it looks like the image is very under-warped, so to say (towards the apex the gray matter "finishes" 6 mm before it should, the division between the cerebellum and the brain is also off by ~5 mm etc...). I've tried playing with the warping parameters but to no avail, they result in lots of small scale changes but no large scale ones. To the best of these unorganized attempts it looks like the "smoothness" parameter has no effect.
Are there any suggestions regarding how to look more systematically into this?
Thank you,
Luca
________________________________________
Da: MRI More <[log in to unmask]>
Inviato: sabato 30 aprile 2016 14.57.59
A: [log in to unmask]; PRESOTTO LUCA
Oggetto: Re: Normalizing PET in SPM12
Dear Luca,
You have access to a larger number of parameters if you go via "Segment" instead of "Normalise". During "Segment", write out the flow fields and then apply them onto the images you want to normalise via a "Normalise: Write" step.
Best
Helmut
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]
|