[Insight-developers] Direct Fourier Tomographic Reconstruction Filter

[Dominique Zosso]

Hi,

I am currently working on a prototype filter to reconstruct a 2D (3D) 
image from a series of 1D (2D) rotational projections, e.g. as used in 
tomographic volumetric reconstruction.

The filter uses a direct fourier reconstruction approach and makes 
extensive use of itk-FFT functions. It works quite well with VnlFFT but 
simply switching to FFTW routines causes the program to fail (not 
crash!). Any ideas in what these two routines might behave differently? 
[the Fourier transforms do not look the same at all, and 
backtransforming a 512x512 2D-Fourier-Image into signal space yields 
strange image dimensions 1022x512...]

Moreover, I just wanted to know whether there was a general interest in 
pushing this prototype further towards an Insight Journal publication 
and a more robust ITK contribution?

Thanks for your ideas

Dominique

-- 
Dominique Zosso, MSc
EPFL - STI - ITS - LTS5
Station 11
CH-1015 Lausanne