k-Wave User Forum » Topic: High Frequency Sensor for Photoacoustic Simulation

k-Wave User Forum » Topic: High Frequency Sensor for Photoacoustic Simulation http://www.k-wave.org/forum/topic/high-frequency-sensor-for-photoacoustic-simulation Support for the k-Wave MATLAB toolbox en-US Wed, 13 May 2026 08:14:30 +0000 http://bbpress.org/?v=1.0.2 <![CDATA[Search]]> q http://www.k-wave.org/forum/search.php Mukaddim on "High Frequency Sensor for Photoacoustic Simulation" http://www.k-wave.org/forum/topic/high-frequency-sensor-for-photoacoustic-simulation#post-7625 Wed, 17 Jun 2020 19:30:58 +0000 Mukaddim 7625@http://www.k-wave.org/forum/ Hello Brad, Thanks for your input. I had to set higher temporal sampling frequency thus reducing dt to resolve the issue. Regards, Rashid Bradley Treeby on "High Frequency Sensor for Photoacoustic Simulation" http://www.k-wave.org/forum/topic/high-frequency-sensor-for-photoacoustic-simulation#post-7604 Sat, 13 Jun 2020 13:05:20 +0000 Bradley Treeby 7604@http://www.k-wave.org/forum/ Hi Rashid, Quick check - if your point targets have a higher sound speed, try setting <code>medium.sound_speed_ref</code> to the background values (it defaults to the max). You may need to decrease your CFL / dt (i.e., increase the temporal sampling) to make the simulation stable. Brad. Mukaddim on "High Frequency Sensor for Photoacoustic Simulation" http://www.k-wave.org/forum/topic/high-frequency-sensor-for-photoacoustic-simulation#post-7589 Tue, 09 Jun 2020 20:54:47 +0000 Mukaddim 7589@http://www.k-wave.org/forum/ Hello Brad, Thank you very much for looking into the code. I followed your recommendation and performed the simulation again. Specifically, I am simulating 0.1 mm point targets with initial pressure magnitude of 3 Pa. To record sensor data, a 128-element linear array with center frequency of 21 MHz and 78% fractional bandwidth is being used and I am using a sampling frequency of 84 MHz to generate Kgrid time array. When I am setting dy = dx, my point targets are getting spread out when compared to the original settings with dx ~=dy. I am sharing two results with specific simulation parameters used for you here - <a href="https://drive.google.com/file/d/18E2dFq0ghuBAPBehtnglR41E7i0uc8fk/view?usp=sharing" rel="nofollow">https://drive.google.com/file/d/18E2dFq0ghuBAPBehtnglR41E7i0uc8fk/view?usp=sharing</a> In both cases, Delay-and-sum beamforming algorithm is used. I would really appreciate if you look into this and help me understand the reason behind the difference. Thank you very much for your help. Regards, Rashid Al Mukaddim Bradley Treeby on "High Frequency Sensor for Photoacoustic Simulation" http://www.k-wave.org/forum/topic/high-frequency-sensor-for-photoacoustic-simulation#post-7568 Sat, 06 Jun 2020 18:20:31 +0000 Bradley Treeby 7568@http://www.k-wave.org/forum/ Hi Mukaddim, Your simulation has a widely different maximum supported spatial frequency in each direction (i.e., dx ~= dy). This means that the bandwidth of the source will have a very strange directional response. I would recommend setting dx = dy. Then, the maximum frequency you can simulate for a given grid is given by c_min / (2 * dx). This is the maximum supported frequency that is printed to the command line. For the time step, I would choose something with a CFL around 0.3, where dt = cfl * dx / c0. In this case, you shouldn't notice much difference if you make dt smaller (this is a good check that the simulation has converged). Hope that helps, Brad Mukaddim on "High Frequency Sensor for Photoacoustic Simulation" http://www.k-wave.org/forum/topic/high-frequency-sensor-for-photoacoustic-simulation#post-7534 Fri, 29 May 2020 18:06:04 +0000 Mukaddim 7534@http://www.k-wave.org/forum/ Hello Forum, I am trying to perform 2D photoacoustic simulation of point target sources in a homogeneous medium. To record sensor data, a 128-element linear array with center frequency of 21 MHz and 78% fractional bandwidth is being used. Delay-and-sum beamforming is used for image reconstruction. I have noticed that K-wave allows us to set two sampling frequencies - spatial and temporal and changing either of them changes the reconstruction results. Higher spatial and temporal frequcy seems to provide better results (image enclosed here: <a href="https://drive.google.com/file/d/1hD-wVWQgKJN7id1qKPJQ0_My4JnVbg23/view?usp=sharing)" rel="nofollow">https://drive.google.com/file/d/1hD-wVWQgKJN7id1qKPJQ0_My4JnVbg23/view?usp=sharing)</a>. Can anyone help me to understand why this is happening? I am also providing my code below. Thank you very much for the help. %% Linear Array Photoacoustic Imaging for Point Initial Source % Define Probe Properties pitch = 90e-6; % pitch TransducerElem = 128; % Number of transducer elements f_cen = 21e6; % Transducer center frequency f_BW = 78; % Transducer Bandwidth imaging_aperture = pitch*TransducerElem; % aperture length for single quadrant imaging c0 = 1540; lambda=c0/f_cen; % wavelengsth [m] % Define K-space spatial grids Ny = 2500; % Ny = 256; Nx = TransducerElem; imagingdepth = 25e-3; % [m] imagingwidth = imaging_aperture; % [m] dy = imagingdepth/Ny; % grid point spacing in the y direction [m] --> Simulation Depth dx = imagingwidth/Nx; % grid point spacing in the x direction [m] --> Aperture is placed in this direction PML_size=[20,200]; % size of the PML in grid points kgrid = kWaveGrid(Nx, dx, Ny, dy); % Define Medium properties medium.density = 1000; medium.sound_speed = 1540; % [m/s] % medium.alpha_coeff = 0.3; % [dB/(MHz^y cm)] % medium.alpha_power = 1.5; % medium.alpha_mode = 'no_dispersion'; % Define K-wave Temporal Grid f_s = 42e6*2; dt = 1/f_s; % 1/sampling frequency dt_stability_limit = checkStability(kgrid, medium); while dt>dt_stability_limit f_s = f_s*2; dt = 1/f_s; end Nt = ceil((imagingdepth*f_s)/c0); % time steps to simulate kgrid.t_array = dt*(0:Nt); % Define K-wave sensor sensor.mask = zeros(Nx, Ny); sensor.mask(:,1) = 1; sensor.frequency_response = [21e6,78]; % [center frequency = 21Mz with 78% fractional BW] % Define initial pressure distrbution rad = 0.0001/dx; % radius = 0.1mm mag = 3; % magnitude in PA x_index = [round(Nx/2-Nx/6) round(Nx/2) round(Nx/2+Nx/6)]; xPos = kgrid.x_vec(x_index) + kgrid.x_vec(end); % x-location of point targets [m] yPos = [8e-3 12e-3 16e-3 20e-3]; % y-location of point targets [m] p0sum = zeros(Nx,Ny); for j = 1:length(xPos) x = round(xPos(j)/dx); for k=1:length(yPos) y = round(yPos(k)/dy); disc = mag*makeDisc(Nx, Ny, x, y, rad); p0 = smooth(kgrid, disc, true); p0sum = p0sum + p0; end end %% Run the Simualtion on GPU % input_args = {'PMLSize',PML_size,'PMLAlpha',4,'PlotPML',false,'PMLInside', false,'Smooth', false,'DataCast','gpuArray-single', 'PlotSim', true,'PlotLayout',false}; sensorData = kspaceFirstOrder2D(kgrid, medium, ... struct('p0',p0sum), sensor, input_args{:}); sensorData = gather(sensorData);