Tomographic image reconstruction using LabVIEW acquired data

[ZJ.Gu]

Hi All,

There is another simple example in the book . Although it does not include backprojection filter, it is another good example for learning fan-beam tomograph reconstruction:

You could download the code in

http://www.informit.com/content/images/0130474150/CDContents/0130474150_CDContents.zip . And find this example under exercises\exercises\Chapter_2_ImageGeneration.zip\Chapter_2_ImageGeneration.

Maybe you could try to improve it by incorporating backprojection filters.

Thanks!

ZJ Gu

[DonRoth]

Years ago, I think I remember this was included in the LabVIEW examples.    Because I know I have seen it before at some point

.....Don

[DonRoth]

rpursley8 wrote:

Here is an example (LV 8.6) that I have a slice of a 3-tube phantom.  The backprojection filters are written in Mathscript so you need that toolkit to execute that part of the code.  I placed that part of the code in a Diagram Disable Structure.  The other frame has a constant that contains a Cosine filter.  I also added a Wait function to the reconstruction loop so you can see the final image as it builds up.

Great work Rpursley8.  The cosine filter is a reasonable default as a  deconvolution filter.  I think another approach that would not require  mathscript would be use of  the signal processing window functions included in LabVIEW applied to the FFT result.  I slightly modified your example to show this for use of the LabVIEW general cosine window, but your cosine coefficients does a better deblurring job.

Don

[DonRoth]

Randall - I think you are right regarding the mis-application of the LabVIEW window functions for the frequency-based result because the FFT result is symmetric about 0 Hz and thus the filter must also be.  But I wonder if there is a way to make these windows useful for this application.  ZJ - can you comment on this?

Don

[ZJ.Gu]

Hi Don and Randall,

Thanks for discussing about the window function I checked with all the window functions. It seems no function match such a application requirement. But we could apply such a window using a simple code segment:

[DonRoth]

Thanks ZJ - this strategy works very well and should allow us to experiment with other filters outside of mathscript.

Sincerely,

Don

[pajogdand]

HI ALL FRDS,

From can i get online data of sets of 2D image

[Curt_Corum]

Dear All,

I am encouraged to see some ongoing interest in tomographic image reconstruction and processing with LabvVIEW

I have had an interest in using LabVIEW for non-cartesian MRI (Magnetic Resonance Imaging) signal processing and image reconstruction (for about the last 5-10 years)

In my current application, raw data is aquired on an MRI scanner, or Digital Reciever (non-NI) attached to an MRI scanner and then processed in LabVIEW.

For 3-D or 4D (3D + time series data) I had used LabVIEW with a gridding reconstruction (interpolation and fourier transform) algortihm which is much faster than the 3D inverse radon transform.

Currently I use a command line call to compiled multi-threaded fortran code, but would like to call 64-bit linux shared libraries. This is to have large matrix sizes and shared memory. At the moment the biggest difficulty is that there is no (to my knowledge) 64-bit linux LabVIEW. Is anybody using a beta?

Anyway, it sounds like folks are interested in 2D radon transform. I do have some old code from grad school for this (implemented in LabVIEW). THere is the example code that has been pointed out. I have not used the MathScript version of the inverse radon transform, but am interested in folk's experiences...

Best,

Curt

[pajogdand]

Sir,

        I am finding very hard for Developing Radon Transform and proceesing 2D image for improment of Image resolusion in labVIEW.Can u forward me some code related to CT imaging.

Thakns,

Pravin.

[mishklyar]

On my work I also collect the 3D MRI data with Labview. Tried to do the 3D filtered backprojection reconstruction on labview at first, but it

was quite impossible to debug the code. So I wrote a prototype with Matlab, and finally a C DLL for performance.