registration between T2 and FA

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

registration between T2 and FA

gabriele pasqua

Dear all,

 

I’m using 3dSlicer v.4.5 and I want to extract FA values from spinal cord MRI, using T2 MRI and spinal cord DWI.

 

In order to do this, I use the following image processing pipeline:

 

1)   Conversion of T2 and DWI DICOM datasets to NRRD format;

2)   Filtering of DWI using Joint Lician  LMMSE Filter with default parameters (Estimation Radius 2,2,1 and Filtering Radius 2,2,1)

3)   DWI mask creation with DWI Weighted Volume Masking module using 0,2 as Otsu Threshold (see image 1)

4)   DTI calculation using DWI to DTI Estimation module, using the created mask and baseline

5)   FA calculation using Diffusion Tensor Scalar Measurements module

6)   Registration between T2 image and FA using General Registration module with the following parameters:

a.     Percentage of samples 0,2

b.     BSpline Grid Size 3,3,3

c.      Slicer Linear Transform : none

d.     Slicer BSpline Trasform : create new

e.     BSpline >27 DOF

f.      Initial Transform Mode : none

7)   Finally, we make the ROI on FA image to extract FA values (image2)

 

Even if it seems that we get a good registration between T2 and FA, we are not satisfied with FA values obtained, especially where there is a compression on spinal cord (see image3) and we expect low FA values.

We get a mean FA value across the patients bigger than 0,6, but when we use GE software installed on MRI machine, we get FA values lower than 0,45, that is a result that fits with state of the art papers.

 

Is there something wrong in our processing steps?

 

Can you help us?

 

Regarding step 3), if I set a bigger or a lower value for Otsu Threshold I get the results showed in attached images mask_01 and mask_05,  using a threshold respectively of 0,1 and 0,5.



Regards


_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject
http://www.slicer.org/slicerWiki/index.php/Documentation/4.3/FAQ

image1.png (160K) Download Attachment
image2.png (245K) Download Attachment
image3.png (1M) Download Attachment
mask_01.png (713K) Download Attachment
mask_05.png (919K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: registration between T2 and FA

inorton
Two suggestions:

- Don't use the Joint Rician filter, because the noise model may be scanner-dependent.

- Don't register the FA to T2. Use the Baseline instead, because the contrast is most-similar to T2.

It is also potentially misleading to do such an ROI-based comparison using different software packages unless you are sure the masks are comparable. One sanity-check step would be to manually compare a small selection of FA values from corresponding individual voxels between Slicer-generated (unregistered! volume and the scanner-generated volume.


On Wed, Aug 3, 2016 at 3:53 AM, gabriele pasqua <[hidden email]> wrote:

Dear all,

 

I’m using 3dSlicer v.4.5 and I want to extract FA values from spinal cord MRI, using T2 MRI and spinal cord DWI.

 

In order to do this, I use the following image processing pipeline:

 

1)   Conversion of T2 and DWI DICOM datasets to NRRD format;

2)   Filtering of DWI using Joint Lician  LMMSE Filter with default parameters (Estimation Radius 2,2,1 and Filtering Radius 2,2,1)

3)   DWI mask creation with DWI Weighted Volume Masking module using 0,2 as Otsu Threshold (see image 1)

4)   DTI calculation using DWI to DTI Estimation module, using the created mask and baseline

5)   FA calculation using Diffusion Tensor Scalar Measurements module

6)   Registration between T2 image and FA using General Registration module with the following parameters:

a.     Percentage of samples 0,2

b.     BSpline Grid Size 3,3,3

c.      Slicer Linear Transform : none

d.     Slicer BSpline Trasform : create new

e.     BSpline >27 DOF

f.      Initial Transform Mode : none

7)   Finally, we make the ROI on FA image to extract FA values (image2)

 

Even if it seems that we get a good registration between T2 and FA, we are not satisfied with FA values obtained, especially where there is a compression on spinal cord (see image3) and we expect low FA values.

We get a mean FA value across the patients bigger than 0,6, but when we use GE software installed on MRI machine, we get FA values lower than 0,45, that is a result that fits with state of the art papers.

 

Is there something wrong in our processing steps?

 

Can you help us?

 

Regarding step 3), if I set a bigger or a lower value for Otsu Threshold I get the results showed in attached images mask_01 and mask_05,  using a threshold respectively of 0,1 and 0,5.



Regards


_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject
http://www.slicer.org/slicerWiki/index.php/Documentation/4.3/FAQ


_______________________________________________
slicer-users mailing list
[hidden email]
http://massmail.spl.harvard.edu/mailman/listinfo/slicer-users
To unsubscribe: send email to [hidden email] with unsubscribe as the subject
http://www.slicer.org/slicerWiki/index.php/Documentation/4.3/FAQ