new sequence

[versacea]

Thanks for the update! 

I have a couple of questions. The 1st question is about the DSI sequence that we are considering. To recap, this is the sequence with 101 directions and a bval with multiple b value with a random-like distribution. Our physicist claims that this even distribution of b values avoids an 'overheating' effect (I am not sure this is the proper term, but that's what I understood) and this might make the sequence 'more stable'..? My major concerns are: 1. Will I be able to use REKINDLE  (I read the paper which looks great.... Looking forward to tray it) with this sequence?  2. Thinking ahead, will this sequence allow me to employ CSD in a DKI framework in the future?

The 2nd question is about the network analysis. I was hoping to find some hints for these analyses in this new release. So, is there any other source that I could consider? 

Thank you very much, Amelia

PS: Hope you had a great time at the conference! The program looked very interesting.

[Sjoerd Vos]

Hi Amelia,

Typical DSI sequences have a strongly different in their acquisition compared to the typical DTI/CSD/DKI acquisitions for the methods available in ExploreDTI. ExploreDTI supports DTI and CSD analyses, which typically rely on many diffusion directions acquired at a single b-value (single-shell, e.g., b=1000, or b=2000). We also have DKI analysis, where multi-shell acquisitions have been shown to be optimal (http://www.ncbi.nlm.nih.gov/pubmed/20199917). DSI on the other hand acquires its data not in shells, but in grids in q-space (see Fig. 1 on http://massive-data.org/methods/gradient-directions.html for difference). Nevertheless, DKI will work on DSI sequences, but the DSI acquisition is likely not optimal - which may affect the DKI analyses. You will be able to use REKINDLE for DKI.

CSD in ExploreDTI currently only supports single-shell acquisitions, so your DSI sequence would not allow for CSD. At this year's ISMRM there was a talk on including data from multiple b-values for CSD (Ben Jeurissen, p973). I don't know if, and if so when, this will come available in ExploreDTI.

Over the last few days, there's been a few posts on using network analysis, or connectivity analysis. I suggest your read through those, hoping that helps you on the way.

Hope this helps in your considerations.

Sjoerd

[Alexander Leemans]

Hi Amelia,

See Section 14 of the updated manual: details on how to perform a network based analysis have now been added. If something is not clear, let me know.

Cheers,
-A

[Versace, Amelia]

Thanks, Sjoerd !! This helps a lot, but I still need some inputs.

Is there a way to derive  ‘optimized’ gradient tables for DTI/DKI sequences? Let’s say that I want to acquire a 3-shell sequence (b=1000, 2000 and 3000) with about 100 directions in a Siemens machine.

Many Thanks, Amelia

--
You received this message because you are subscribed to the Google Groups "ExploreDTI" group.
To unsubscribe from this group and stop receiving emails from it, send an email to E_DTI+un...@googlegroups.com.
Visit this group at http://groups.google.com/group/E_DTI.
To view this discussion on the web visit https://groups.google.com/d/msgid/E_DTI/9179c134-e339-4c97-a138-4e3907725533%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Versace, Amelia]

Thank you so much!! Amelia

 

From: E_...@googlegroups.com [mailto:E_...@googlegroups.com] On Behalf Of Alexander Leemans
Sent: Monday, June 02, 2014 6:45 PM
To: E_...@googlegroups.com
Subject: [E_DTI] Re: new sequence

 

Hi Amelia,

--

You received this message because you are subscribed to the Google Groups "ExploreDTI" group.
To unsubscribe from this group and stop receiving emails from it, send an email to E_DTI+un...@googlegroups.com.
Visit this group at http://groups.google.com/group/E_DTI.

To view this discussion on the web visit https://groups.google.com/d/msgid/E_DTI/44e202b5-01b9-4eae-947e-af2ca4a58962%40googlegroups.com.

[Sjoerd Vos]

Hi,

There is a tool in ExploreDTI to generate gradient direction sets on a single sphere, e.g., for DTI or CSD acquisitions. For the multi-shell acquisition there are two tools that I know of:

1) http://massive-data.org/methods/gradient-directions.html

2) http://www.emmanuelcaruyer.com/q-space-sampling.php

The first one is the one I am most familiar with, and I think gives a bit more freedom in designing your preferred acquisition.

Cheers,

Sjoerd

[Chantal Tax]

Dear Amelia,

Also, you might find this article useful: http://www.ncbi.nlm.nih.gov/pubmed/10467296. Gradients are ideally uniformly distributed over the (hemi)sphere using an electrostatic repulsion algorithm.

Here, you can find an example of 30 gradient directions that are uniformly distributed: http://www.ncbi.nlm.nih.gov/pubmed/21279600.

However, there are also other schemes that you can think of, see for example Fig. 1a in http://www.ncbi.nlm.nih.gov/pubmed/23475834.

Hope this helps! :)

Op woensdag 4 juni 2014 12:57:47 UTC+2 schreef versacea:

To unsubscribe from this group and stop receiving emails from it, send an email to E_DTI+unsubscribe@googlegroups.com.

[Versace, Amelia]

Thank you so much, Chantal and Sjoerd !!

Actually I tried both tools and none of the two is generating what I was hoping.

 

The first tool (http://massive-data.org/methods/gradient-directions.html) actually gets closer to what I am looking for. It generates a 3 shell b-vector table, but it does not allow for more then 60 directions in each shell. Is there a physical reason why I cannot acquire more than 60 directions with multi-shell sequences?

The second tool (http://www.emmanuelcaruyer.com/q-space-sampling.php) generates a bvector table that looks more like a DSI table to me. Here is what I get:

 

#shell	u_x	u_y	u_z
3	0.652	-0.715	-0.254
2	-0.371	-0.008	-0.928
3	0.659	0.699	-0.277
3	0.427	0.205	0.881
2	0.928	-0.009	-0.371
3	0.235	-0.705	0.669
3	-0.565	0.077	0.822
1	-0.166	-0.936	-0.311
3	-0.971	-0.037	-0.234
2	-0.015	0.999	-0.046
3	0.147	0.872	0.467
3	-0.232	-0.414	0.88
2	0.74	0.597	0.312
3	-0.765	0.536	-0.356
3	-0.105	-0.986	0.131
2	0.389	-0.555	-0.735
3	0.149	-0.506	-0.85
3	-0.935	-0.142	0.326
2	0.19	0.628	-0.755
3	-0.702	-0.549	-0.453
3	-0.52	0.275	-0.809
1	0.966	-0.255	0.045
3	-0.372	0.927	-0.054
2	-0.782	0.584	-0.216
3	0.829	-0.325	-0.456
3	-0.331	-0.751	0.572
2	-0.234	-0.696	-0.679
3	0.025	-0.038	-0.999
3	-0.546	-0.821	-0.163
2	0.704	0.571	-0.422
3	0.319	-0.781	-0.537
3	0.798	0.047	0.6
2	0.873	-0.088	0.481
3	0.687	0.343	-0.64
3	-0.93	0.367	0.006
1	0.038	0.267	-0.963
3	0.323	0.589	0.741
2	0.356	0.031	-0.934
3	0.895	0.446	0.019
3	0.439	-0.803	0.403
2	-0.351	0.758	-0.549
3	-0.15	0.34	-0.928
3	-0.538	0.459	0.707
2	0.51	-0.816	-0.273
3	-0.081	0.977	0.197
3	-0.026	-0.88	0.474
2	-0.48	-0.876	0.043
3	-0.821	0.265	-0.507
3	-0.369	-0.133	0.92
1	-0.807	-0.52	0.279
3	0.993	-0.048	-0.106
2	-0.984	-0.166	-0.068
3	-0.431	-0.89	0.15
3	0.132	0.351	0.927
2	-0.009	-0.297	-0.955
3	0.699	-0.712	0.06
3	-0.659	-0.346	-0.668
2	0.816	-0.453	-0.36
3	-0.8	-0.008	0.6
3	0.436	0.776	0.455
2	0.335	-0.426	0.841
3	0.516	-0.544	0.662
3	-0.818	-0.455	0.351
1	-0.522	0.61	0.597
3	-0.054	0.737	0.673
2	0.011	-0.92	-0.393
3	-0.069	-0.636	0.769
3	0.894	0.324	0.308
2	0.663	0.222	-0.715
3	0.399	-0.88	-0.257
3	0.298	-0.235	-0.925
2	-0.674	-0.269	-0.688
3	-0.321	0.076	-0.944
3	0.51	0.558	-0.655
2	-0.168	-0.888	0.427
3	0.84	-0.488	-0.236
3	0.25	0.957	0.146
1	0.637	0.067	0.768
3	0.186	-0.942	0.28
2	0.476	-0.866	0.154
3	0.749	0.66	0.065
3	0.935	-0.253	0.25
2	0.658	-0.235	-0.715
3	0.611	-0.6	-0.517
3	0.616	-0.015	0.788
2	0.684	-0.381	0.622
3	0.099	0.204	-0.974
3	-0.974	-0.219	0.06
2	0.839	0.541	-0.061
3	0.292	0.886	-0.361

 

 

Is there something that I am doing wrong or that I am not considering?

 

Thanks again!! Amelia

To unsubscribe from this group and stop receiving emails from it, send an email to E_DTI+un...@googlegroups.com.

Visit this group at http://groups.google.com/group/E_DTI.
To view this discussion on the web visit https://groups.google.com/d/msgid/E_DTI/9179c134-e339-4c97-a138-4e3907725533%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

You received this message because you are subscribed to the Google Groups "ExploreDTI" group.

To unsubscribe from this group and stop receiving emails from it, send an email to E_DTI+un...@googlegroups.com.

Visit this group at http://groups.google.com/group/E_DTI.

To view this discussion on the web visit https://groups.google.com/d/msgid/E_DTI/32e753ab-9ebf-4c87-9b92-b011e9d09a94%40googlegroups.com.

[Sjoerd Vos]

The gradient table you showed still has to be modified for the b-value. I know on GE and Philips this is impractical, not sure about Siemens.

The limit in the first tool to 60 gradient directions is probably an arbitrary one, likely based on excessive scantime when multiple shells are acquired with 60+ directions. The author (of the first tool) is a colleague of ours, so I'll ask and see if he can modify it to increase the upper limit. Will keep you updated.

Sjoerd

[Versace, Amelia]

Thank you so much!!

Amelia

--
You received this message because you are subscribed to the Google Groups "ExploreDTI" group.

To unsubscribe from this group and stop receiving emails from it, send an email to E_DTI+un...@googlegroups.com<mailto:E_DTI+un...@googlegroups.com>.

Visit this group at http://groups.google.com/group/E_DTI.

To view this discussion on the web visit https://groups.google.com/d/msgid/E_DTI/0831747d-2dfd-4474-986e-93654d79d51a%40googlegroups.com<https://groups.google.com/d/msgid/E_DTI/0831747d-2dfd-4474-986e-93654d79d51a%40googlegroups.com?utm_medium=email&utm_source=footer>.

[Martijn Froeling]

I have changed the tool that it now can generate 128 gradient direction per shell in multi shell optimization.

Be aware though that the optimization will be less accurate when generating many directions and the tool might become unstable, thus put the quality on perfect (5000 iterations) or extreme (10000 iterations)

Martijn

On Wednesday, June 4, 2014 3:24:57 PM UTC+2, versacea wrote:

Thank you so much!!

Amelia

On Jun 4, 2014, at 9:10 AM, "Sjoerd Vos" <sjoer...@gmail.com<mailto:sjoer...@gmail.com>> wrote:

The gradient table you showed still has to be modified for the b-value. I know on GE and Philips this is impractical, not sure about Siemens.

The limit in the first tool to 60 gradient directions is probably an arbitrary one, likely based on excessive scantime when multiple shells are acquired with 60+ directions. The author (of the first tool) is a colleague of ours, so I'll ask and see if he can modify it to increase the upper limit. Will keep you updated.

Sjoerd

--
You received this message because you are subscribed to the Google Groups "ExploreDTI" group.

To unsubscribe from this group and stop receiving emails from it, send an email to E_DTI+unsubscribe@googlegroups.com<mailto:E_DTI+unsub...@googlegroups.com>.

[Versace, Amelia]

Dear ExploreDTI experts,

 

I am having some issues when I create the .mat file with the new sequences that we are testing.

The DWIs look ok; however, the .mat file looks incomplete. I tried masking and w/o masking the dwi images, but I got the same results ( attached is the ‘fuse CV (I used DWIs) with FEFA’).  Have you seen this before? Am I doing something wrong?

 

Not really relevant, but please note that  I had the same issue with three sequences (all with 200 gradients+ 11 b0s) ran all on the same subject:

The 1^st one with   bmax=2500 (represented in the attachment), the 2^nd one with b=3000 and the 3^rd one with b=4000. (We were testing the effect of different b values).

 

Thanks a lot,

To view this discussion on the web visit https://groups.google.com/d/msgid/E_DTI/EC3487F5706D514EADE4F0995BD9AF472526DE92%40msxmbxnsprd17.acct.upmchs.net.

[Sjoerd Vos]

Hi Amelia,

The only thing I can think of based on this image is that the problem might be caused by low SNR. Physically, signal intensities should decrease with increasing b-values. Sometimes, due to noise, this is not the case. It could for instance be that along a certain gradient direction the DWI intensity at b=1500 is just a bit higher than at b=2500, due to noise. The fact that this happens in the lower portion of the brain is that SNR is generally lower there because of blood pulsatility.

If this is the problem, you could either try and do cardiac gating if you didn't do that. That should increase image quality. Or you could increase the voxel size a little bit to increase SNR.

I can't be certain of the cause without seeing the data, so if you're willing to share it I could have a look in more detail.

HTH,

Sjoerd

[Chantal Tax]

Hi Amelia,

Just a quick question: Have you performed motion/distortion correction on this data? Did this perform well?

Best,

Chantal

Op zondag 20 juli 2014 05:38:02 UTC+2 schreef versacea:

3

-0.331

-0.751

<p class="

...

[Versace, Amelia]

I did and turned out really bad, but I have to say that I used the FA image for registration purpose. I usually get better registrations with the FA images. However, in this case, while the FA is 'partial', the DWIs look fine. will try using the DWIs images.
Thanks!!
Amelia

[Versace, Amelia]

I found the issue. I am going to post it here because it might be usefully for somebody else.

While I use the ExploreDTI v484 for my analyses, at the very beginning of my analytic pipeline, I go back using the v481 for the very first step. In the original version, there was a "reordering_to_begin" or something like that, that automatically moved the b0s up-front.
This did always work fine with my 'mono-shell' sequence. I guess the multi-shell sequences are more challenging for that script, given the different intensities of the images. I have now used the proper "Shuffle/select..." and the images look fine!!

Thanks for your help!!
Amelia

-----Original Message-----
From: E_...@googlegroups.com [mailto:E_...@googlegroups.com] On Behalf Of Versace, Amelia
Sent: Monday, July 21, 2014 7:04 PM
To: Chantal Tax
Cc: E_...@googlegroups.com; sjoer...@gmail.com
Subject: Re: [E_DTI] Re: new sequence

--
You received this message because you are subscribed to the Google Groups "ExploreDTI" group.
To unsubscribe from this group and stop receiving emails from it, send an email to E_DTI+un...@googlegroups.com.
Visit this group at http://groups.google.com/group/E_DTI.

To view this discussion on the web visit https://groups.google.com/d/msgid/E_DTI/73815B60-25FE-4BE4-9307-93F345F7554A%40upmc.edu.