I was trying to add noise to the phantom images generated by Fiberfox, save the 
.dwi image,
and try to *normally* process this image using the Diffusion module to 
regenerate the
original fibers (generated using Fiberfox).

This was my workflow:

*         Fiberfox:

1.       Choose tensor/ball model and set my image dimensions and spacing. The 
number of gradient
directions was 12

2.       Add Rician noise with a variance of 10

3.       Simulate to create a dummy image sample space (Signal Generation panel)

4.       Generate my fiber bundle (Fiber Definition panel)

5.       In the Fiber Processing module, create a binary envelope for my fiber 

6.       I've noticed that the envelope does not properly cover the span of the 
fiber bundle (http://imgur.com/a/QpPKy)

7.       In Fiberfox, in the Signal Generation panel, simulate the DW-MRI 
images choosing the fiber bundle and its envelope

8.       Save the generated image(s) as a .dwi image

9.       Close and reopen MITK Diffusion

*         Fiber tractography:

1.       Drag and drop the saved .dwi image into the Data Manager

2.       In DWI Denoising module, select the .dwi image and use Non-local means 
filter and
provide the same variance for Rician noise that we provided in Fiberfox. Apply 
the filter

3.       In Signal modeling toolbox, reconstruct the tensors and estimate the 
diffusion weighted images
from these tensors

4.       Used Gibbs tracking, streamline tracking and stochastic tracking to 
reconstruct the fibers.
The images are available here: http://imgur.com/a/dsAL1

5.       The tracked fibers do not match (not even coarsely) to the original 

Am I performing fiber tractography correctly here? In any case, I do see a 
problem with the generated binary

Can someone please comment?

I've uploaded a test file here: https://ufile.io/07108
As can be seen in the Coronal view, nearly half the fiber information is not 
captured in the DW-MRI image


Warm regards,

Warm regards,

mitk-users mailing list

Reply via email to