k-Wave

1. deblina
   

   Member
   Joined: Nov '12
   Posts: 13

   Hi,

   In K-wave, is it possible to use finite size sensor (for example circular sensor of 10 mm diameter) instead of using a point sensor?

   Posted 10 years ago #

2. bencox
   

   Developer
   Joined: Mar '10
   Posts: 285

   Hi deblina,

   Probably the easiest thing to do in this case is to make a sensor mask covering the region you want your (single) sensor, then sum the time series together afterwards. That will model the effect of the finite size of the detector.

   Hope that helps,
   Ben

   Posted 10 years ago #

3. deblina
   

   Member
   Joined: Nov '12
   Posts: 13

   Hi Ben,

   Thanks for your reply. In the post you have mentioned about 'time series'. Is it refer to the time domain signal (sensor_data)? If that is so, then the acquired signal from all the sensor point within that region should be summed or averaged?

   Thanks and regards,
   Deblina

   Posted 10 years ago #

4. bencox
   

   Developer
   Joined: Mar '10
   Posts: 285

   Hi Deblina,

   Yes, that's right. Sum the sensor_data so you only have one time series.

   Best wishes,
   Ben

   Posted 10 years ago #

5. deblina
   

   Member
   Joined: Nov '12
   Posts: 13

   Hello Ben,

   I understand your point but I am bit confused between sum and average because in one of the K-wave examples given as follows:

   clear all;

   % =========================================================================
   % SETUP THE GRID
   % =========================================================================

   % create the computational grid
   Nx = 180; % number of grid points in the x (row) direction
   Ny = 180; % number of grid points in the y (column) direction
   dx = 0.1e-3; % grid point spacing in the x direction [m]
   dy = 0.1e-3; % grid point spacing in the y direction [m]
   kgrid = makeGrid(Nx, dx, Ny, dy);

   % define the properties of the propagation medium
   medium.sound_speed = 1500; % [m/s]

   % define the array of time points [s]
   [kgrid.t_array, dt] = makeTime(kgrid, medium.sound_speed);
   Nt = round(0.75*length(kgrid.t_array));
   kgrid.t_array = (0:Nt-1)*dt;

   % =========================================================================
   % DEFINE A FOCUSSED ARRAY OF DIRECTIONAL ELEMENTS
   % =========================================================================

   % define a semicircular sensor centered on the grid
   semicircle_radius = 65; % [grid points]
   arc = makeCircle(Nx, Ny, Nx/2, Ny/2, semicircle_radius, pi);

   % find total number and indices of the grid points constituting the
   % semicircle
   arc_indices = find(arc == 1);
   Nv = length(arc_indices);

   % calculate angles between grid points in the arc and the centre of the
   % grid
   arc_angles = atan((kgrid.y(arc_indices))./kgrid.x(arc_indices));

   % sort the angles into ascending order, and adjust the indices accordingly
   [sorted_arc_angles,sorted_index] = sort(arc_angles);
   sorted_arc_indices = arc_indices(sorted_index);

   % divide the semicircle into Ne separate sensor elements
   Ne = 19;
   sensor.mask = zeros(Nx,Ny);
   for loop = 1:Ne

   % the indices of the grid points belonging to one element.
   % (There is a two grid point gap between the elements.)
   voxel_indices = sorted_arc_indices(floor((loop-1)*Nv/Ne)+2:floor(loop*Nv/Ne)-1);

   % add the element to the sensor.mask
   sensor.mask(voxel_indices) = 1;

   end

   % =========================================================================
   % SIMULATION
   % =========================================================================

   % Define an infinitely wide plane wave source. (This is achieved by turning
   % off the PML.)
   source.p_mask = zeros(Nx,Ny);
   source.p_mask(140,:) = 1;

   % set the display mask for the simulation
   display_mask = source.p_mask + sensor.mask;

   % define a time varying sinusoidal source
   source_freq = 1e6;
   source_mag = 0.5;
   source.p = source_mag*sin(2*pi*source_freq*kgrid.t_array);
   source.p = filterTimeSeries(kgrid, medium, source.p);

   % run the simulation with the PML switched off on the sides, so that the
   % source continues up to the edge of the domain (and from there infinitely,
   % because of the periodic assumption implicit in pseudospectral methods).
   input_args = {'PMLAlpha', [2 0], 'DisplayMask', display_mask, 'PlotScale', [-0.75 0.75]};
   sensor_data = kspaceFirstOrder2D(kgrid, medium, source, sensor, input_args{:});

   % split up the data, recorded on all the grid points, between the elements
   element_data = zeros(Ne,Nt);
   for loop = 1:Ne

   % the indices of the sensor grid points in the sensor mask
   sensor_indices = find(sensor.mask==1);

   % the indices of the grid points belonging to one element.
   voxel_indices = sorted_arc_indices(floor((loop-1)*Nv/Ne)+2:floor(loop*Nv/Ne)-1);

   % indices of sensor_data that refer to the data for this element
   data_indices = zeros(length(voxel_indices),1);
   for loop2 = 1:length(voxel_indices)
   data_indices(loop2) = find(sensor_indices == voxel_indices(loop2));
   end

   % for one element per loop, average the time series from each of the
   % element's grid points to give one time series for the whole element
   element_data(loop,:) = mean(sensor_data(data_indices,:),1);

   end

   take the mean of sensor_data (last line of the code) to get the time series of one element(combination of multiple sensor points).

   So I wonder which one will be suitable among sum and average (of sensor_data of multiple sensor points)to mimic a finite size sensor?

   Regards,
   Deblina

   Posted 10 years ago #

6. zengqw2021
   

   Member
   Joined: Jun '21
   Posts: 26

   Dear all,
   I want to use a finite size sensor (for example circular sensor of 10 mm diameter) instead of using a point sensor? According to Ben's suggestion:make a sensor mask covering the region you want your (single) sensor, then sum the time series together afterwards. I wonder whether the following code could achieve this goal?

   % run the simulation
   sensor_data = kspaceFirstOrder2D(kgrid, medium, source, sensor, input_args{:});

   %---model the effect of the finite size of the detector
   %--sum the time series together afterwards
   Num_point = 5 ; % Test points that need to be accumulated for a detector
   interval = 2 ; % interval between the two detectors
   N_elements = floor(num_sensor_points/(Num_point+interval)); %the total detectors
   New_sensor_data = zeros(N_elements,kgrid.Nt);
   for i=1:N_elements
   loop_point = (N_elements-1)*(Num_point+interval);
   New_sensor_data(i,:)=sum(sensor_data((loop_point+1):(loop_point+5),:),1);
   end

   Regards,
   Zeng

   Posted 2 years ago #

7. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,640

   Hi Zeng,

   If you want the signal amplitude to match the pressure amplitude (at least when the interaction is coherent), I would take the average instead of the sum.

   You could use the new kWaveArray class to define a finite-sized sensor and use the inbuilt methods to do the averaging for you (karray.combineSensorData(kgrid, sensor_data)).

   Brad.

   Posted 2 years ago #

RSS feed for this topic

Reply

You must log in to post.