Hi Gwenaëlle, here there is a resume of the DICOM characteristics for every T1 image... I would like to know if, looking at this more specific information, you could tell me if going ahead and which are the parameters I should include in the GLM to be fair with the analysis and get reliable results.

I have been looking to the paper you sent it to me about VBM analyses with ADNI-1 and ADNI-2 series with SPM8 and how there were differences between GM and WM stimations depending on the acquisition parameters.

Having selected only the same machine and model with more or less similar acquisition parameters...my sample has been reduced from 154 to 69. And I still have to look for any other problems that could arise by checking image by image individually.

Here there are the characteristics:

Acquisition plane: SAGITTAL

Sequence: MPRAGE

Field Strength: 3T

Pulse sequence: GR/IR

Slice thickness: 1

Flip Angle: 9

TE: mainly 2.98, some of them 2.52 or 2.27

TR: mainly 2300, some of them 1900

TI: 900

Pixel spacing: 1 x 1

Matrix: mainly, 240 x 256 x 176...others 256 x 256 x 176

I would be more than grateful if I could hear about your experience, always helpful in this forum.

Yours sincerely,

Rosalia.

2016-02-23 13:26 GMT+01:00 Gwenaëlle DOUAUD <[log in to unmask]>:

Hi Rosalia,

If you have high enough a sample size, then yes, it is definitely preferable.

As you say, it is also important to have the same protocol, not just the same machine, as clearly demonstrated in this study: A voxel-based morphometry comparison of the 3.0T ADNI-1 and ADNI-2 volumetric MRI protocols. http://www.ncbi.nlm.nih.gov/pubmed/25092796

Cheers,
Gwenaëlle

------------------------------------------------------------------------------------------------------------------------
Gwenaëlle Douaud, PhD
Associate Professor & MRC Career Development Fellow

FMRIB Centre, University of Oxford
John Radcliffe Hospital, Headington
OX3 9DU Oxford UK
Switchboard: +44 (0) 1865 222 493
Fax: +44 (0) 1865 222 717

www.fmrib.ox.ac.uk/team/principal-investigators/gwenaelle-douaud
www.fmrib.ox.ac.uk/research/fmrib-interface-analysis-clinical-neuroscience-group
-----------------------------------------------------------------------------------------------------------------------

De : Rosalia Dacosta Aguayo <[log in to unmask]>
À : [log in to unmask]
Envoyé le : Lundi 22 février 2016 18h48

Objet : Re: [FSL] Intensity problems after running fslvbm_2_template..Need for helping.

Hi Gwenaëlle thank you a lot for the link.

I think I will take a decision regarding using just one machine and discard the other ones. Just the same machine, model, and acquisition parameters...I think this is the best solution I can take for now, and the results will be valid then, as far as I know.

Best wishes,

Rosalia.

2016-02-22 15:52 GMT+01:00 Gwenaëlle DOUAUD <[log in to unmask]>:
Hi Rosalia,

In general, we would advise against using FSL-VBM in a multi-centric analysis, but in any case, you should absolutely add the information of each centre in your GLM model (for more on these issues, see Multi-site voxel-based morphometry--not quite there yet: http://www.ncbi.nlm.nih.gov/pubmed/21324367).

It is not surprising that you should see some differences in the GM maps coming from different scanners at different field strengths. In particular, subcortical GM is highly myelinated, and can appear, depending on the contrast-to-noise ratio, as GM or WM, or a mix of the two.

Cheers,
Gwenaëlle

------------------------------------------------------------------------------------------------------------------------
Gwenaëlle Douaud, PhD
Associate Professor & MRC Career Development Fellow

FMRIB Centre, University of Oxford
John Radcliffe Hospital, Headington
OX3 9DU Oxford UK
Switchboard: +44 (0) 1865 222 493
Fax: +44 (0) 1865 222 717

www.fmrib.ox.ac.uk/team/principal-investigators/gwenaelle-douaud
www.fmrib.ox.ac.uk/research/fmrib-interface-analysis-clinical-neuroscience-group
-----------------------------------------------------------------------------------------------------------------------

De : Rosalia Dacosta Aguayo <[log in to unmask]>
À : [log in to unmask]
Envoyé le : Jeudi 18 février 2016 9h53
Objet : Re: [FSL] Intensity problems after running fslvbm_2_template..Need for helping.

Hi Paul,

I can not attach any screenshot because of the size.....:(

Could I sent you the some of the problems in a pdf file to your email directly?...if you need to see the first template generated I can attach you too.

Thank you,

Rosalia

2016-02-18 10:14 GMT+01:00 paul mccarthy <[log in to unmask]>:
Hi Rosalia,

Could you attach a screenshot which shows your problem?

Cheers,

Paul

On 17 February 2016 at 11:22, Rosalia Dacosta Aguayo <[log in to unmask]> wrote:
Dear FSL experts,

I am worried about one thing that I have noticed it is happening in my images: when I visually check every individual file that contains my 4D GM template before running fslvbm_3_proc.

I see that subcortical structural grey matter structures are not all with the same "color" what means the following: caudate, for example, in some files appears with a grey color whereas in others appears in white color...and sometimes middle in white and middle in grey...this is worrying me a lot. As I am working with different kind of images, from different centers and different teslas (1.5 and 3)...I wonder if this will affect the final results.....

I have checked patient 123 (anterior email) and I have discarded it definitively because a big enlargment of the ventricles.... but what about regarding the other issue intensities?

Any help would be more than grateful,

Kind regards,

Rosalia