Population Average Template

[Evan Curtis]

Hello Frank and the DSI Studio Team,

I am attempting to set up an animal-based control average template for differential tractography. However, I am struggling. The program crashes when I run Tools>P1 to create a population average template. Included are the first two control mouse brain with in-tact skulls. I am not sure what I am doing incorrectly.

Thanks,

Evan Curtis

[Fang-Cheng Yeh]

The Updated DSI Studio has it included at [Step T3][C57BL6_mouse template]

--
You received this message because you are subscribed to the Google Groups "DSI Studio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dsi-studio+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/dc38acb9-a3b3-4441-ad48-47cd9078032bn%40googlegroups.com.

[Evan Curtis]

Thanks! I am having difficulty with the image co/registration. In reference to the ABA and step T3, the images I have are 90 degrees out of alignment from the atlas and need to be scaled to fit. Do you have any recommendations on how best to transform either the images or atlas? 

Thanks again,

Evan

[Fang-Cheng Yeh]

Have you checked https://dsi-studio.labsolver.org/doc/gui_t2.html ?

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/b31d5a40-2e88-425c-9049-1d60c1a0c229n%40googlegroups.com.

[Evan Curtis]

Yes. I am currently stuck at step T2. Is there a way to generally visualize/overlay a template and the images of interest? Otherwise, I am bouncing between Step T2 and Step T3. Thanks again. 

You received this message because you are subscribed to a topic in the Google Groups "DSI Studio" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/dsi-studio/DSzsKMVfIiw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to dsi-studio+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/CAG_QrtLWGqBf6A5zzpWNoRxHFBCDmu7Ge2zgan%2BJ5XvLea29MQ%40mail.gmail.com.

[Fang-Cheng Yeh]

Sorry I am not sure the purpose of " visualize/overlay a template and the images of interest"

It is likely not able to deal with the actual cause of the problem.

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/CAGbBXCcd9KBkzK_ym3e7TesZjzpz3EhtLdt7XzJ9zkU398Dkhw%40mail.gmail.com.

[Evan Curtis]

Ok. I am trying to figure out what orientation my animal images should be in based on Step T2. In Step T2, I can only see the red masking zone. I tried using the flip images options, but I am not sure how best to match the orientation of the templates. Is there a way to see the template at the same time as planning the reconstruction? 

[Fang-Cheng Yeh]

Easier if you can send me a sample of your data, and I will figure out the problem.

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/c77eaba8-3039-4eec-8ac6-0888aa7ae20en%40googlegroups.com.

[Fang-Cheng Yeh]

Hi Evan,

     I checked your data, and there are multiple issues.

(1) The tissues outside the skull needs to be removed by creating a mask. I can use threshold, smoothing, and defragment to create a mask. I also use [Crop background] to reduce volume.

(2) The orientation is very different from the template. I use swap YZ and flip Z to get to the right orientation.

(3) The slice is thick and the in-plane resolution is too high, leading to low SNR. The diffusion contrast is not enough to get orientation differences. The fitting with tensors is not good.

Best regards,

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/c77eaba8-3039-4eec-8ac6-0888aa7ae20en%40googlegroups.com.

[Evan Curtis]

Thanks, Frank!

With your suggestions/ recommendations, I was able to load in my 128x128 matrix size images (0.156 mm x 0.156 mm in-plane) with more success. Now I can auto-select regions of interest and get the DTI and GQI parameters for each brain region with more confidence thanks to the atlas. It is so much better than drawing manual ROIs on the MD or ISO maps.  Now, I am back on the path to trying to figure out differential tractography. 

I have one minor concern and then the bigger concern of step next. 

(1) The images were collected at a slice thickness of 1mm. However, DSI Studio states the thickness is 0.667 mm. How do I correct this issue? Is it something to be concerned about?

(2) While running the tractography, I get tracts that do not seem to make sense to me. For example, there seems to be streamlines that go from the corpus callosum to the cortex that only occur on one hemisphere. 

(3) I am trying to follow the human-based differential tractography example. Is it possible to do differential tractography based on the animal data that I have? I have 5 control-mice and multiple injured mice to compare. 

Thank you,

Evan

[Fang-Cheng Yeh]

Sorry for my late reply.

I have one minor concern and then the bigger concern of step next. 

(1) The images were collected at a slice thickness of 1mm. However, DSI Studio states the thickness is 0.667 mm. How do I correct this issue? Is it something to be concerned about?

DSI Studio extracts values from the source. It would be good to check and confirm how the error happened.

 

(2) While running the tractography, I get tracts that do not seem to make sense to me. For example, there seems to be streamlines that go from the corpus callosum to the cortex that only occur on one hemisphere. 

False results happen all the time.  Autotrack may help eliminate most of them.

 

(3) I am trying to follow the human-based differential tractography example. Is it possible to do differential tractography based on the animal data that I have? I have 5 control-mice and multiple injured mice to compare. 

Yes, the differential tractography document was recently updated. I would suggest you use "Type 4: mapping cross-sectional change in the template space" mentioned at https://dsi-studio.labsolver.org/doc/gui_t3_dt.html

 

Best regards,

Frank

[Evan Curtis]

Thank you, Frank!

 

(1) I cannot seem to find the issue between the thickness acquired and the thickness determined by DSI Studio. I will do a 0.5mm acquisition later and see what DSI Studio thinks the resolution will be.

(2) My concerns with the tractography are shown in Figures 1 and 2. Currently, I am seeing a “bed of nails” artifact of streamlines. I would not expect to see so many tracts to be going directly straight through the brain.  Any guidance is appreciated.

(3) Thank you! I will investigate these steps once I am ready to compare the tractographies.  

 

Thank you,

Evan

[Frank Yeh]

(2) My concerns with the tractography are shown in Figures 1 and 2. Currently, I am seeing a “bed of nails” artifact of streamlines. I would not expect to see so many tracts to be going directly straight through the brain.  Any guidance is appreciated.

Sorry I did not see the figure.  There are some troubleshooting figures at https://dsi-studio.labsolver.org/doc/gui_t3_whole_brain.html

[Evan Curtis]

Hello, I am still having issues with the overall slice thickness appearing way smaller than is actually reported. For example, in my original dataset the slice thickness was 1mm and in-plane resolution was 0.156mm. Since then, I have reacquired my samples at at larger in-plane resolution and smaller slice thickness. The new images are approximately 0.260 in-plane and 0.25mm slice thickness. Hopefully this will make for a more isotropic resolution for tractography. However, DSI Studio is reporting the slice thickness as 0.042mm. Furthermore, the images do not appear isotropic in the renderings in the software. As always, any guidance you have is greatly appreciated. Thank you, Frank!

Perhaps the issue is with my acquisition protocol. Do you know of any examples for Varian/ Agilent VnmrJ 4 ex vivo or in vivo mouse acquisition set up guidelines for DTI?

Thanks again. 

-Evan

[Fang-Cheng Yeh]

It is more likely that DSI Studio did not get the correct resolution.

I will check the data you uploaded and let you know what is the cause and solution.

Frank

--

You received this message because you are subscribed to the Google Groups "DSI Studio" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dsi-studio+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/fc489ce0-2f22-47f6-a8f8-da3c80856fb2n%40googlegroups.com.

[Fang-Cheng Yeh]

I opened the fdf file you send me in a text editor and its show that 

float  roi[] = {2.500000,2.500000,0.025000};

int    slices = 60;

float  matrix[] = {96, 96};

Thus DSI Studio thus calculates the in-plane as 25/96 and slice-thickness as 0.25/60 

[Evan Curtis]

Hmm... I am not sure how to correct for this issue. Thank you for looking into it. As a temporary workaround, I am now using a 3D spin echo sequence for the acquisition. It seems to be able to give the isotropic voxel without any issues at 25/96. However, I am still having issues with the over abundance of green tracts going through the brain. Is there a way to suppress these tracts? 

Thanks again,

Evan

[Fang-Cheng Yeh]

On Thu, Nov 10, 2022, 2:50 PM Evan Curtis <evan.t....@gmail.com> wrote:

Hmm... I am not sure how to correct for this issue. Thank you for looking into it. As a temporary workaround, I am now using a 3D spin echo sequence for the acquisition. It seems to be able to give the isotropic voxel without any issues at 25/96.

One quick solution is to open the SRC file in MATLAB and manually modify the voxel_size matrix.

https://dsi-studio.labsolver.org/doc/cli_data.html

However, I am still having issues with the over abundance of green tracts going through the brain. Is there a way to suppress these tracts? 

Likely due to slice movement or eddy current.

Do you have a screenshot of those false tracts?

Frank

You received this message because you are subscribed to a topic in the Google Groups "DSI Studio" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/dsi-studio/DSzsKMVfIiw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to dsi-studio+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/88adf241-256f-4494-aaac-83426d03424cn%40googlegroups.com.

[Evan Curtis]

Hello Frank and the DSI Studio Team,

Included below are a couple of ex vivo samples. As can be seen in the sagittal view, the fibers are running transverse/ axially through the brain (Figure 1). I would have expected the fibers to appear in a more radial pattern as show by other groups. I also am including a lower resolution image (Figure 2) for comparison.

The following are examples of tractography shown by a couple other studies that I am trying to replicate.

Examples:

-Whole brain connectomics: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6467764/

-dMRI connectome: https://pdfs.semanticscholar.org/79a9/eb8985275f6da5c25e905c78918b5813d208.pdf?_ga=2.30573244.2009020901.1668635500-1684921766.1668635500

Do you have any recommendations for suppressing the transverse fibers in the hope of seeing other features?

Thank you,

Evan

[Fang-Cheng Yeh]

It seems to me that using ROA or ROI to limit the findings would be a
quick solution, but it depends on what fibers you are going to
investigate.
Frank

> To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/08e9f74b-5a6b-43ef-ba73-96e025e9b785n%40googlegroups.com.

[Evan Curtis]

I attempted to generate tracts based on the ROI and ROA recommendations. The ROI based on the corpus collosum produced some curious results (Figure 1). For example, the tracks seem to stay only in the imaging plane and are not oriented anatomically. This is most obvious in Figure 2.

Thanks again,

Evan

[Fang-Cheng Yeh]

The tracking requires good-quality DWI data, and for most animal
scans, there are challenges.
I would suggest checking the "local fiber orientation" (a.k.a fixels)
at each voxel to make sure that they are oriented correctly.
If fixels are not correctly oriented, the problem is upstream (e.g.
acquisition, data processing); otherwise, the issue is downstream
(e.g. tracking algorithm, ROI/ROA combinations).
Frank

> To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/24eabe60-142e-43ee-9c05-c80d25578202n%40googlegroups.com.

[Fang-Cheng Yeh]

I checked the figures. All voxel show green color, meaning that the
fixels are all pointing in anterior-posterior directions.

There are data acquisition problems.

You may need to check out DWI troubleshooting videos about the
potential issues: https://www.youtube.com/watch?v=stL4GMeTC1I
If the local fiber orientations are not correct, then any analysis
downstream is not going to work out.

You may also check out the mouse template at Step T3 to see what a
good FIB file looks like.

Best,
Frank

[Evan Curtis]

Hi Frank,

I apologize for the slow response. Our MRI system went down and took a while before I could reacquire the sample images. I followed your advice and tried a couple of things. While the flipping the b-table did improve results around the brainstem, I am still seeing a dominant A-P green voxel issue. To see if the problem was originating from the acquisition, I acquired a 3D-Spin echo in the axial, coronal, and sagittal orientations. The intent was to determine whether the orientation had any impact on the tractography results. While there is some visible difference in the sagittal views between orientations, they all displayed the dominant A-P fiber pattern.

As always, any help you can provide is greatly appreciated.

Thank you,

Evan

[Fang-Cheng Yeh]

The results you sent me still does not look right.
I would suggest you check the raw DWI signals as what I have
demonstrated in a recent workshop (week4):
http://practicum.labsolver.org/
It is likely that the diffusion time or duration setting are not good.

> To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/ee1d357a-7a94-4ba4-9ba7-b7f6023bc622n%40googlegroups.com.

[Evan Curtis]

Hi Frank,

I have watched all of the training videos and tried to follow along to the best of my ability. However, when working with my dataset, the program is unable to conduct the EDDY adjustment. It begins to process, then states ‘Unknown Error.’ Do you have any idea what I am doing wrong?

I tried running before or after the other image adjustments and receive the same error message. I thought the issue might be with the using .fdf files instead of .dcm, but the .dcm files do not contain the bvec or bval information based on what Varian/ Agilent provides. 

Note: checking the ‘check b-table’ button did slightly improve the rendering and tractography.

As for diffusion time and duration, how much longer should I set the values to? Currently, I am using the default values given by the Varian/ Agilent software.

Included in the Word Doc are examples of the issues I am experiencing.

Thank you,

Evan

[Fang-Cheng Yeh]

The most likely cause is that the dataset does not have a btable eddy required.

You may upload a sample of dataset and I can help checking.

Best regards

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/b8178834-20ca-429e-bdae-fb32916db5a3n%40googlegroups.com.

[Fang-Cheng Yeh]

If possible, please send me the SRC file.

Frank

[Evan Curtis]

Hello Frank,

Included are the SRC files generated from Step T1. I was not sure which of the SRC files to attach, so I am sending you all of them for the sagittal 3D spin echo only.

·         slab001image001echo001.fdf.src.gz

·         slab001image001echo001.fdf.src.gz.bval

·         slab001image001echo001.fdf.src.gz.bvec

·         slab001image001echo001.fdf.src.gz.corrected.eddy_command_txt

·         slab001image001echo001.fdf.src.gz.index.txt

·         slab001image001echo001.fdf.src.gz.mask.nii.gz

·         slab001image001echo001.fdf.src.gz.nii.gz

·         slab001image001echo001.fdf.src.gz.odf.gqi.0.6.fib.gz

·         slab001image001echo001.topup.acqparams.txt

·         20221212_ExVivo_Control1_6mo_3D_96_orientation.zip

 

As a quick refresher, usually after opening the SRC, I do the following adjustments.

·         Edit

o   Flip YZ

o   Flip XY

o   Flip Y

o   Threshold 120

o   Defragment

o   Dilation

o   Crop Background Volume

·         Checked b-table box

Thank you,

Evan

[Fang-Cheng Yeh]

Thanks. I will check and get back to you 

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/b8634b00-4e1b-4a29-8ebd-569ae1840f2cn%40googlegroups.com.

[Fang-Cheng Yeh]

I checked your data. There is no visible eddy distortion, and there is no need to run FSL eddy.

If you still would like to run, then you would need to manually replace all non-zero b-values to 3000.

Best regards,

Frank

To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/b8634b00-4e1b-4a29-8ebd-569ae1840f2cn%40googlegroups.com.

[Fang-Cheng Yeh]

After flip Y , you will need to add a flip X. Otherwise, it will be a mirrored image.

Frank

[Evan Curtis]

Hi Frank, 

Thank you for the advice. Glad to know there is not any eddy current problems. However, I am still seeing a lot of anterior to posterior oriented tracts in the cortex. Do you have any additional suggestions on eliminating this issue?

If this is an acquisition issue, I am willing to try other methods or parameters.

Currently, I am using a 3D spin echo sequence with the following parameters…

TR:300 ms

TE: 21.26 ms (minimum TE)

Averages: 1

Gradient Spoil: On

Read, Phase, Phase2: 25 x 25 x 25 mm^3

Matrix: 96 x 96 x 96

Resolution: 260 x 260 x 260 um^3

 

Diffusion--

Amplitude: 33.21 G/cm

Separation: 12 ms

Duration: 6 ms

Directions: 1 b0, 30 target b-value (Jones30 scheme, according to Varian/ Agilent)

Target b-value: 3000 s/mm^2

 

Acquisition time: 23 hr 48 min 29 sec

 

Since, this is probably not an issue with DSI Studio, do you have any recommendation on who I should contact for guidance?

 

Thanks,

Evan 

[Evan Curtis]

Hello Dr. Yeh,

Just following up on the post from last week. Glad to know there is not any eddy current problems. However, I am still seeing a lot of anterior to posterior oriented tracts in the cortex. Do you have any additional suggestions on eliminating this issue? I can re-upload the images to dropbox if that helps. 

Since, this is probably not an issue with DSI Studio, do you have any recommendation on who I should contact for guidance?

Thanks,

Evan 

Evan Curtis
University of Nebraska-Lincoln
Lincoln, NE USA

[Fang-Cheng Yeh]

I am very sorry for the late reply due to my travel.
You may upload the SRC file again, and I will see if I can find a clue
about the cause.
One likely cause of the problem is the diffusion gradient duration and
diffusion gradient separation. The values may not be ideal.
You may refer to studies done by Duke CIVM, which I believe have
acquired one of the best animal ex-vivo images in the world.

Best regards,
Frank

> To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/a3c5144c-1f7d-4a3e-9662-89233279f660n%40googlegroups.com.

[Evan Curtis]

Hello Frank and Team,

I am running into difficulty with differential tractography based on the mouse template. I was able to get interesting results using the native space protocol presented in the workshop documents. However, I am not able to generate tracts using the recommended mouse template. I recently uploaded an example of my process and perhaps you can steer me in the correct direction.

The workshop tutorials and videos have been very helpful. 

Thank you,

Evan

[Fang-Cheng Yeh]

Received and will get back to you.
Frank

[Fang-Cheng Yeh]

Hi Evan,

I checked your data, and there are several issues (I am sorry it
took so long).

(1) The control data have the same demographics, and there is no need
to regress against the demo. You may create an averaged FIB from
control and use the average QA as the comparison target.
(2) The case study was reconstructed by GQI and it should use QSDR so
that the methods are consistent between control and case.

I would suggest you check out the workshop videos about
differential tractography at practicum.labsolver.org Although the
sample data were human subjects, similar principles applied.

Best regards,
Frank

> To view this discussion on the web visit https://groups.google.com/d/msgid/dsi-studio/b09c1c68-80e2-4f3c-9b50-403d17b2d260n%40googlegroups.com.