Phillips gradient | Diffusion Toolkit Usage

	You must be logged in to post	Login Register

Lost Your Password?

Forum Diffusion Toolkit Usage Phillips gradient

Phillips gradient

Page: 1

User

Post

6:04 am
July 1, 2011

TeresaML

Member

posts 4

Print this Post

Post edited 10:12 am – July 1, 2011 by TeresaML

Hi,

I'm newbie in TrackVis and I don't know what are the parameters I must use in Diffussion Toolkit. I'm use a DTI data from Philips 3T with 2040 images, 60 slices and 34 gradients. I try and I have obtain the following captures

6:07 am
July 1, 2011

TeresaML

Member

posts 4

Print this Post

The other one is

10:20 am
July 2, 2011

r.navarra

Member

posts 13

Print this Post

Post edited 1:38 pm – July 4, 2011 by r.navarra

Hi, as in http://trackvis.org/blog/forum…..sues-help/ with Philips data, tracking need "Invert X" (data from PARREC) or "Invert Z" (data from DICOM). Anyway, You said "34 gradients" but I know "high" sequence from Philips is 32 directions: First one (or 33-th, it depends on release) is B0, the last one is an average (this is not needed). So (32+1+1=) 34 volumes but 32 directions ;) Is it correct for you or are you using a non-standard Philips sequence?

To get correct gradients you have to ways: 1) to extract gradients from DICOM (MRIConvert) or 2) to use DTI_gradient_table_creator_Philips_RelX with informations from PAR/REC.

Bye.

6:12 am
July 4, 2011

TeresaML

Member

posts 4

Print this Post

Hi,

Thank you very much for your post. I don't know if it is a non-standard Philips sequence or not. I am newbie and I am a little lost… I used the MRIConvert and I obtanied the following gradient:

0:0,0,0
1:-0.641306,0.332612,-0.691444
2:0.77781,0.628299,-0.0158954
3:0.71532,0.211127,-0.66614
4:-0.668097,0.540402,-0.51148
5:-0.410149,0.748844,-0.520586
6:0.605044,0.577236,-0.548379
7:0.737654,0.390674,-0.550673
8:-0.628247,0.777928,0.0115192
9:-0.637902,0.548926,0.540149
10:0.767853,0.399197,0.501042
11:0.559395,0.65735,0.504944
12:-0.627997,0.77797,0.0195754
13:0.723172,-0.0691564,0.687197
14:-0.569243,0.495362,0.656185
15:-0.625729,0.777208,0.0664148
16:-0.198059,0.739319,0.643568
17:0.384224,0.677319,0.627384
18:0.777732,0.628292,-0.0195724
19:0.752264,0.221554,0.620494
20:-0.626085,0.777395,0.0606232
21:0.129322,0.704322,0.698002
22:0,0,0
23:-0.464544,0.544372,-0.698468
24:-0.423957,0.55583,0.715062
25:0.77753,0.628247,-0.0274187
26:-0.641114,0.33252,-0.691667
27:-0.156431,0.68817,-0.708485
28:0.07382,0.695027,-0.715184
29:0.458354,0.514413,-0.72477
30:0.68186,0.0752559,-0.727601
31:-0.235002,0.72889,-0.643035
32:0.348827,0.66603,-0.659336
33:0.542956,0.423073,-0.725402

I used the "DTI_gradient_table_creator_Philips_RelX" too but I dont' obtained the same results:

0: 0.000, 0.000, 0.000

1: -0.500, -0.500, -0.707

2: -0.500, -0.500, 0.707

3: 0.707, -0.707, 0.000

4: -0.653, -0.271, -0.707

5: -0.209, -0.676, -0.707

6: 0.020, -0.707, -0.707

7: 0.421, -0.568, -0.707

8: 0.690, -0.155, -0.707

9: -0.653, -0.271, -0.707

10: -0.293, -0.707, -0.644

11: 0.295, -0.706, -0.644

12: 0.515, -0.486, -0.706

13: 0.707, -0.293, -0.644

14: -0.707, -0.472, -0.526

15: -0.472, -0.707, -0.526

16: 0.555, -0.644, -0.526

17: 0.707, -0.472, -0.526

18: -0.707, -0.707, -0.000

19: -0.707, -0.472, 0.526

20: 0.707, -0.472, 0.526

21: 0.472, -0.707, 0.526

22: -0.707, -0.707, 0.008

23: -0.636, -0.425, 0.644

24: -0.706, -0.706, 0.055

25: -0.293, -0.707, 0.644

26: 0.293, -0.707, 0.644

27: 0.707, -0.707, 0.008

28: 0.707, -0.293, 0.644

29: -0.706, -0.706, 0.049

30: 0.035, -0.706, 0.707

31: 0.707, -0.707, 0.012

32: 0.707, 0.000, 0.707

33: 100.000, 100.000, 100.000

Is it correct? What of two ways is suitable? Sorry for my questions. Best Regards.

9:37 am
July 4, 2011

r.navarra

Member

posts 13

Print this Post

Post edited 2:02 pm – July 4, 2011 by r.navarra

Standard sequences (on Philips 3T Achieva) have only 1 b0 (at first or second last row) and 6, 15 or 32 gradients (low, medium and high; another one, "experimental", has 60dirs). You sequence has 2 b0 (first and 23-th row), it's not a standard sequences so, if U want to use DTI_gradient_table_creator_Philips_RelX, you need to specify the gradients (who has entered gradient table in the scanner?), then, in applet (or matlab code) use informations required: Patient Position, Preparation Direction, Angulation midslice and so no. I think it's better for you to get gradients from DICOM ;)

9:15 am
July 5, 2011

Ali Amad

Member

posts 6

Print this Post

Post edited 1:21 pm – July 5, 2011 by Ali Amad

Hi,

I am newbie in brain imaging and I try to work with trackvis.

All my datas are par/rec files, I convert them to nifti, anf I've obtained exactly the same picture as TeresaML.

I've dl MRIconvert to extract gradient from DICOM, but it seems that MRIconvert doesn't support par/rec files.

Is it possible for you to help me ?

Thank you very much for your help.

10:14 am
July 5, 2011

r.navarra

Member

posts 13

Print this Post

MRIConvert need DICOM. Get DICOM files from CD burnt (of PACS/OsiriX server, if you use it) ;) Again,
if DICOM's missing and sequence's not standard (low, medium or high) you need to get gradient table (from technician?) then re-orient by using DTI_gradient_table_creator_Philips_RelX before run DTK…

For info: if you can, email me a single PAR file –> r.navarra@itab.unich.it
Tnx.

10:54 am
July 5, 2011

Ali Amad

Member

posts 6

Print this Post

Thank you very much for your help.

In fact, I have found the gradient file, but it doesn't work well yet, even if I apply a "invert x" as it was proposed with rec/par files.

This is the first picture I have :

And, I've got really few fibers :

I precise that I obtained nifti with dcm2nii, and that my raw datas were rec/par.

I sent you by email a rec/par file with the gradient file, if it is possible for you to help me.

Thank you very much for your support.

11:46 am
July 5, 2011

r.navarra

Member

posts 13

Print this Post

my pipeline:

1) convert PAR/REC to NIFTI (dcm2nii)

2) remove last volume from nifti (I use fsl)

3) Create gradient table by using DTI_gradient_table_creator_Philips_RelX with infos from .PAR ( http://godzilla.kennedykrieger…..UI.html )

4) run dtk (invert x) –>

I've seen worse ;)

Ask scanner technician for gradient table!

12:33 pm
July 5, 2011

Ali Amad

Member

posts 6

Print this Post

Thank you very much for your help again.

Your explications are very clear.

Could you explain briefly how to remove the last volume from nifti with fsl ?

4:31 am
July 6, 2011

r.navarra

Member

posts 13

Print this Post

[OT] first, see: http://www.fmrib.ox.ac.uk/fsl/…..index.html and http://www.fmrib.ox.ac.uk/fslc…../index.htm

fslroi file_input file_output 0 n-volumes [enter]

Ex: data has 41 volumes (0-40), you want to eliminate the last one –>

fslroi data dataminus41thvol 0 40 [enter]

[/OT]

11:22 am
July 6, 2011

Ali Amad

Member

posts 6

Print this Post

Fantastic !!! It works Laugh .

Thanks a lot for your help r.navarra.

8:12 am
July 12, 2011

TeresaML

Member

posts 4

Print this Post

I get the real gradient used in the scanner, but I obtained the following fiber tracking. What is it happening? Have I left something to consider? Thanks!

8:37 am
July 12, 2011

Ali Amad

Member

posts 6

Print this Post

hi,

if it is possible for you send me a rec/par files at : aliamad2@yahoo.fr

I'll have a look and help you if I can.

Page: 1