k-Wave

1. puyang
   

   Member
   Joined: Jul '15
   Posts: 3

   Hi Dr. Cox and Dr. Treeby
   I am new to the PA field. I now build up my PA microscope. I use a focused transducer to do 2D scan of my sample and I see good results in the oscilloscope. I now have the PA data and I hope to use the “Example: 3D Time Reversal For A Planar Sensor" to reconstruct 3D image. However, I have problem to do it since I do not know how to replace the “variables” using my real data in the code.
   Can you give some instruction for this? For example, I now have a Nx by Ny by Nz raw data from the measurement, how to set the variable of “PML_size”, “kgrid”, “p0” “sensor.mask” and “sensor_data” etc to get start.

   Thanks a lot. Your support is highly appreciated.

   Posted 9 years ago #

2. puyang
   

   Member
   Joined: Jul '15
   Posts: 3

   In short question, what is the p0 input? Is p0 the measured voltage by oscilloscope?

   Posted 9 years ago #

3. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,643

   Hi puyang,

   Are you performing PA microscopy or tomography?

   If you are doing microscopy, then no image reconstruction is required (see this review paper for example). Just assemble your acquired time traces into a matrix, and visualise them using the appropriate axes scaling factors (i.e., converting time to depth using the sound speed).

   If you are doing tomography, then image reconstruction is required. In the k-Wave example you mention, the simulation is both generating the "experimental data" (where p0 is the initial pressure distribution), and then performing the reconstruction. You should comment out the first part of the example. The experimental data (measured voltage) is then used to replace sensor_data in this line:

   % assign the time reversal data
   sensor.time_reversal_boundary_data = sensor_data;

   Let me know if you're doing tomography, and I can give some more details on how to set the other variables.

   Brad.

   Posted 9 years ago #

4. puyang
   

   Member
   Joined: Jul '15
   Posts: 3

   Dear professor:
   Thanks a lot. I am doing microscopy. I am studying it.

   Posted 9 years ago #

5. lbc
   

   Member
   Joined: Aug '15
   Posts: 2

   Hi Dr. Treeby and Dr. Cox
   I am appreciated for your k-Wave toolbox, it indeed to facilitate the photoacoustic tomography reconstruction. here is my question. there is a photoacoustic tomography system in my lab. the original data acquisition equipment is an arc transducer, the arc angle is 3*pi/2. at present I launch an experiment and get the original tomography data. so I want to use the k-Wave toolbox to reconstruct the tomography image. I reference the example: 2D Time Reversal For A Circular Sensor and assign the time reversal data equal to my tomography data, then modify the sensor radius and number of elements, at last I run the m-file, but the result image is so far from the original tomography image.
   can you give me some advice for the present situation? when using the example: 2D Time Reversal For A Circular Sensor for processing the real tomography data, how to set up the corresponding parameters in order to get the relative satisfying image?
   your reply will be appreciated! thanks a lot!

   Posted 9 years ago #

6. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,643

   Hi lbc,

   There are only two parameters that you need to define correctly - the sensor geometry (in your case the radius of the arc angle), and the sound speed. If you get the latter incorrect, the image will be defocused.

   If you don't know the sound speed, you could try using the autofocus routine to estimate the sound speed that gives you the sharpest image (see the Automatic Sound Speed Selection Example).

   Keep in mind that the quality of the reconstructed image will also be affected by the size and frequency response of your acoustic detectors, etc.

   Hope that helps,

   Brad.

   Posted 9 years ago #

7. lbc
   

   Member
   Joined: Aug '15
   Posts: 2

   Hi Dr. Treeby

   Appreciate your reply and advice. it indeed to open my mind.

   first i try using the auto focus routine to estimate the sound speed by taking the example of Automatic Sound Speed Selection Example as a reference and furthermore read the paper Automatic sound speed selection in photoacoustic image reconstruction using an autofocus approach, as you said in the paper: `If the reconstructed image is rich in features and any image artifacts are comparatively small in magnitude, then the autofocus approach will work correctly to maximize the sharpness of the image feature, However, if the converse if true, then the autofocus approach may incorrectly attempt to maximize the sharpness or the prevalence of edges in the image artifacts. at present, my reconstructed image with the k-Wave toolbox is poor in features and image artifacts are comparatively large in magnitude, so this method do not work.

   here, i have another question about the radius of censor(sensor_radius), how do i assign the value of this parameter? should i set the value according to the radius of real transducer?

   now my original data is gathered in the vivo experiment, the sound speed differs among various of organs, so when i set the value of sound speed equal to a scalar, can i finally get the satisfying image?

   at last, you said the quality of the reconstructed image will also be affected by the frequency response of acoustic detectors, how do i implement this aspect in the code?

   thanks a lot, any your reply will be highly appreciated!

   Posted 9 years ago #

8. Bradley Treeby
   

   Developer
   Joined: Mar '10
   Posts: 1,643

   Hi lbc,

   Yes, the values of sensor_radius and sensor_angle should match the radius and arc-angle of the transducer used in the experiment.

   The sound speed in soft tissue only varies by a small amount (e.g. +/- 5%), so provided you use a sensible mean value, many experimental papers have shown you can still reconstruct excellent images (e.g., here).

   It is possible to correct for the frequency response of your detector over the detector bandwidth by deconvolving this before you run the reconstruction. You will need to be careful how you apply this so you don't amplify noise. Note, if you are using band-limited detectors (e.g., a PZT ultrasound array), you will likely not detect high and low frequency components of the signal, which will affect your reconstructed image (e.g., see here). Unfortunately, there is no way to correct for this if no signal is recorded at these frequencies.

   Hope that helps,

   Brad.

   Posted 9 years ago #

RSS feed for this topic

Reply

You must log in to post.