k-Wave

1. sepand
   

   Member
   Joined: Sep '14
   Posts: 19

   Hi

   Dear all , Bencox and Bradley Treeby

   I am using source and sensor class in 2D for ultrasound synthetic aperture imaging. I have defined 32 grid points as transducer elements with no spacing between them. For each emission, just one of the source points is active but all 32 sensor elements receive the echoes. Source and sensor are at the same place and with same geometry. The scattering medium is a homogeneous medium with a point target at the middle of medium (a single pixel with a little bit more density than background). As I know from ultrasound, this point target should reflect the echo signals to the sensors (receiving elements). But after plotting the sensor_data, the echo signals arranged for each transmission, so that, the nearest element to the emitter has the smallest delay. While, as you know, for synthetic aperture imaging (for a point target at the middle of medium) it should have the smallest delay for middle element for all transmissions regardless of location of emitter. So, I think that, for each transmission, all sensor points just sense the transmitted signal not echo signals from point target. How can I get to the true echo signals from point target (PSF)?
   Best regards
   Sepand

   Posted 8 years ago #

2. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi Sepand,

   In k-Wave, sources are implemented as the injection of force or mass into your medium. This means there is no physical interface at the transducer surface. Similarly, the sensor mask just tells the simulation where to sample the pressure field at each time step (so there is no physical sensor in the field). Consequently, what you are seeing in your simulations is the input signal propagating directly to the sensor positions. As you would expect, this arrives at the closest element first. To avoid this, you need to time-gate out the beginning of the received signal. This should allow you to see the reflections from the scatterer. Indeed, the same is true in a real ultrasound machine, as the elements cannot both transmit and receive at the same time.

   Hope that helps,

   Brad.

   Posted 8 years ago #

3. sepand
   

   Member
   Joined: Sep '14
   Posts: 19

   Hi Brad

   Thank you very much for your help. As you suggested I have eliminated the beginning of received signal and The echo signal could be seen in remained signal, but the big problem is some unknown oscillations in addition to echos. these oscillations have variable amplitudes and they sounds to have a constant frequency. I can not really understand the origin of them. Because,as you know, for a single scatterer in a homogeneous medium, we should have a echo signal similar to input signal with a specific time delay according to the distance from emitter. but I have gotten a long signal with unknown fluctuations. How can I get rid of these additional oscillations?

   In my 2D simulations the input signal is a 3MHz tune_burst with one cycle.

   thank you in advanced

   Sepand

   Posted 8 years ago #

4. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi Sepand,

   Without seeing your code it's hard to say. Have you tried running a convergence test, i.e., reducing dx/dy and the CFL number and seeing if you get the same answer?

   Brad.

   Posted 8 years ago #

5. sepand
   

   Member
   Joined: Sep '14
   Posts: 19

   Hi Brad
   My codes are as following:

   clear all
   close all
   clc
   data_cast = 'single';
   run_simulation = true;
   pml_x_size = 20; % [grid points]
   pml_y_size = 20; % [grid points]

   % set total number of grid points not including the PML
   sc = 1;
   Nx = 256/sc - 2*pml_x_size; % [grid points]
   Ny = 256/sc - 2*pml_y_size; % [grid points]

   % set desired grid size in the x-direction not including the PML
   x = 50e-3; % [m]

   % calculate the spacing between the grid points
   dx = x/Nx; % [m]
   dy = dx; % [m]

   kgrid = makeGrid(Nx, dx, Ny, dy);

   % define the properties of the propagation medium

   medium.sound_speed= 1540; % [m/s]
   medium.density = 1000; % [kg/m^3]
   medium.alpha_coeff = 0.75; % [dB/(MHz^y cm)]
   medium.alpha_power = 1.5;
   medium.BonA = 6;

   % create time array

   t_end = (Nx*dx)*2.2/medium.sound_speed; % [s]
   [kgrid.t_array, dt]= makeTime(kgrid, medium.sound_speed, [], t_end);

   point_magnitude =100*randn(Nx,Ny); % [Pa]
   x_pos = 3*Nx/4; % [grid points]
   y_pos = Ny/2; % [grid points]
   radius = 1; % [grid points]
   arc_angle = 2*pi; % [radians]

   Disc = point_magnitude.*makeDisc(Nx, Ny, x_pos, y_pos, radius, arc_angle);
   imagesc(double(Disc));
   %colormap(gray)
   density_map = 1000*ones(Nx, Ny);
   medium.density = density_map (:,:);
   medium.density = medium.density + Disc;
   imagesc(double(medium.density));
   colormap(gray)

   sound_speed_map = 1540*ones(Nx, Ny);
   medium.sound_speed = sound_speed_map (:,:);
   %medium.sound_speed = medium.sound_speed + Disc;
   imagesc(double(medium.sound_speed));
   colormap(gray)

   number_elements=64;
   element_width=1;
   element_spacing=0;
   transducer_width = number_elements*element_width ...
   + (number_elements - 1)*element_spacing;

   sensor.mask = zeros(Nx, Ny);
   sensor.mask(1, Ny/2-(number_elements/2-1):Ny/2+number_elements/2) = 1;
   %sensor.record_start_index=800;
   imagesc(double(sensor.mask));

   sensor_element_width=1;

   %%================================================
   %--synthetic aperture simulation------------------
   %=================================================

   %matlabpool('open',4); % start of Parallel Computing
   j=1;
   for i=Ny/2-(number_elements/2-1):Ny/2+number_elements/2;
   j
   source.p_mask = zeros(Nx, Ny);
   source.p_mask(1, i) = 1;
   imagesc(source.p_mask);
   colormap(gray);

   tone_burst_cycles = 1;
   tone_burst_freq=3e6;

   % % create the input signal using toneBurst
   input_signal = toneBurst(1/dt, tone_burst_freq, tone_burst_cycles);%,'Plot', true);

   source_strength=1e6;
   source.p =(source_strength./(1000*1540)).*input_signal;
   source.p_mode='dirichlet';

   input_args = {...
   'PMLInside', false, 'PMLSize', [pml_x_size, pml_y_size], ...
   'DataCast', data_cast, 'DataRecast', true, 'PlotSim', false};

   sensor_data{j,1} = kspaceFirstOrder2D(kgrid, medium, source, sensor,input_args{:});

   j=j+1;
   end

   thank you in advanced for your help
   Sepand

   Posted 8 years ago #

6. sepand
   

   Member
   Joined: Sep '14
   Posts: 19

   Dear Brad
   in addition to unknown fluctuations the mentioned echo signals are on an ascending curve around zero axis(y=0). I mean that the echo signals should be zero for other parts of homogeneous medium and they should have amplitudes for phantom position. But my received echoes are not on zero axis(y=0).

   Posted 8 years ago #

7. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi Sepand,

   Your source center frequency is very close to the maximum frequency supported by the grid. I would suggest reducing the grid spacing (dx/dy) or reducing the source frequency. For any simulation, it's always a good idea to run a convergence test by reducing dx and CFL and checking you get the same answer.

   Brad.

   Posted 8 years ago #

8. sepand
   

   Member
   Joined: Sep '14
   Posts: 19

   Hi Brad

   Thank you so much for your help. By reducing the grid spacing and the source frequency most of unknown fluctuations has been removed, But for removing all of them I have to change the boundary condition from dirichlet to additive(the default). Now it is a question for me that, is it true to use the additive boundary condition for ultrasound simulations?

   Thank you in advanced for your help
   Sepand

   Posted 8 years ago #

9. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi sepand,

   I would suggest using an additive source unless you specifically want to enforce a boundary condition (for example, to perform time reversal using experimental data). Using an additive source in k-Wave corresponds to the normal way mass and force sources are defined in the wave equation.

   Brad.

   Posted 8 years ago #

RSS feed for this topic

Reply

You must log in to post.