k-Wave

1. Songwriter45
   

   Member
   Joined: Sep '15
   Posts: 7

   Hello.

   I'm researching into ultrasound propagation through human skull, and firstly, I'm trying to define the emitter in a 3D medium.
   In our laboratory we have round and curved emitters, so I decided to create the emitter mask with the function "makeDisc" and then using the function "focus" to finally obtain the simulated round and curved focused emitter.

   The problem I found is that I obtained the pressure in the central axis, and I compared it to the analytical solution of O'Neil (Theory of focusing radiators), and the results are different. So, why are they different? The combination of the two functions "makeDisc" and "focus" is not realistic? Is there any other possibility of simulating a round and curved focused ultrasound emitter?

   Thank you.

   Posted 8 years ago #

2. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi Songwriter45,

   It might be possible to use a circular radiator with the appropriate time delays to mimic a focused bowl transducer near the focal region, but the radius and focus parameters will likely need to be optimised (e.g., see this reference, Fig. 1). Keep in mind that the two will not agree over the whole field.

   If you want to model focused emitters, the best approach would be to model the source geometry explicitly. We have written a new function called makeBowl (described in this paper) which will be included in the next k-Wave release that gives very good agreement. I'm happy to send you an early version of this function if that would be helpful.

   To mimic the field of an unfocussed plane piston transducer, just use makeDisc and drive it uniformly, i.e., don't use the focus function.

   Hope that helps,

   Brad.

   Posted 8 years ago #

3. ultrasoundlearner
   

   Member
   Joined: Dec '16
   Posts: 1

   Hi Bradley,

   I'm interested in using the makeMultiBowl function as introduced in your recent UFFC paper to simulate a HIFU phased array comprised of 256 circular elements. You mentioned that this function will be included in the next k-Wave release. May I ask, when will the next release be available. Forgive me if the information has already been posted on this website. I'm new to the forum.

   Thanks!

   Posted 7 years ago #

4. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi ultrasoundlearner,

   We are working on a new release at the moment, and hope to have it out sometime in the next few months.

   Brad.

   Posted 7 years ago #

5. lbruju
   

   Member
   Joined: Mar '18
   Posts: 19

   Hi,

   I want to mimic the evolution of pressure using an ultrasound probe. This probe has a bowl shape and is made of 256 'disc elements' with, each, different input signal.

   How it could be possible to model that? Define the index of all these elements in p_source_index is possible. But define the matrix p_source_input with a size np.size(p_source_index)xnp.size(input_signal) seems to be impossible.

   Thanks a lot,

   lbruju

   Posted 6 years ago #

6. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi lbruju,

   The function makeMultiBowl will also return a labelled matrix. You can then use this to decipher how to arrange the source terms in source.p. For example, let's say you have the driving signals for each physical element stored in a matrix called source_sigs which has dimensions num_bowls by Nt. You could then do something like the following:

   clearvars;
   
   % define grid parameters
   x_size              = 150e-3;
   Nx                  = 128;
   dx                  = x_size / Nx;
   grid_size           = [Nx, Nx, Nx];
   
   % create a Cartesian sphere with the x, y, z positions of the bowls
   sphere_radius       = 70e-3;
   num_bowls           = 32;
   bowl_pos            = makeCartSphere(sphere_radius, num_bowls, [1, 1, 1] * x_size / 2).';
   
   % convert the Cartesian bowl positions to grid points
   bowl_pos            = round(bowl_pos/dx);
   
   % define element parameters
   radius              = round(x_size / (2 * dx));
   diameter            = 13;
   focus_pos           = [1, 1, 1] * Nx/2;
   
   % create bowls
   [bowls_binary, bowls_labelled] = makeMultiBowl(grid_size, bowl_pos, radius, diameter, focus_pos);
   
   % create source signals (random in this example)
   Nt = 200;
   source_sigs = rand(num_bowls, Nt);
   
   % assign source mask
   source.p_mask = bowls_binary;
   
   % collapse the labelled bowl matrix to remove zeros
   bowls_labelled = bowls_labelled(bowls_labelled ~= 0);
   
   % assign source signals, looping through each grid point that forms part of
   % the source
   source.p = zeros(sum(bowls_binary(:)), Nt);
   for loop_index = 1:length(bowls_labelled)
       source.p(loop_index, :) = source_sigs(bowls_labelled(loop_index), :);
   end

   Hope that helps,

   Brad.

   Posted 6 years ago #

7. lbruju
   

   Member
   Joined: Mar '18
   Posts: 19

   Hi Brad,

   Thanks for your answer.
   I am simulating with the C++ code and I am creating the hdf5 file with python, so I don't have the function makeMultiBowl. That's why it seems to be hard to obtain the index of all the points defining all the 256 disc elements.
   If there isn't an easier way to simulate this, I will simplify the issue and try to simulate the pressure field with a transducer made of 8 rings, with a different input signal for each ring.

   I am a little bit lost concerning the index. If I know that a point which has coordinates (x,y,z) in the 3D grid is a source point, how could I calculate the index associated to this point (to define correctly p_source_index) ? (Knowing that I am generating that in python)

   Thanks a lot,
   lbruju

   Posted 6 years ago #

8. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi lbruju,

   You could do it manually fairly easily, or use numpy.unravel_index with the column-wise flag.

   Brad.

   Posted 6 years ago #

9. lbruju
   

   Member
   Joined: Mar '18
   Posts: 19

   Hi Bradley,
   thanks a lot for your help.
   lbruju

   Posted 6 years ago #

RSS feed for this topic

Reply

You must log in to post.