[Insight-users] Computing forces for deformation fields in MR data

[Luis Ibanez]

Hi Leila,

I loaded your MR data in the FastMarchingLevelSet application,
where you can easily experiment with the Sigma of the Gaussian,
and with the Alpha and Beta parameters of the Sigmoid.

Since your MR data is in the range of 16 bits, the values of the
gradient magnitude reach 1400..,1300 in the inferior part of the
kidneys, and just 400 in the superior part of the kidneys.

The central section of the kidneys have values in the range of
0 to 300.

With this numbers, we can estimate values for the sigmoid.
The values that matter most are the lowest values of the gradient
magnitude in the boundary of the kidneys. In this case, the 400.

In this conditions, a Beta value of 300 and Alpha value of -1
in the Sigmoid filter are convenient for producing a speed
image that has value 1.0 inside the kidney and values almost 0.0
in the boundaries.

Alpha values closer to -9 will produce a masking image with smoother
transitions that will be more appropriate for computing a Gradient
to be used as a Vector Field for guiding the movements of the
deformable model.


You may want to give it a try to the parameters above, and
verify again with Paraview what the values of the vector field
are around the kidneys.



Please let us know what you find,


    Thanks


       Luis