Section of Communication Systems, EPFL
The goal of this project was to come up with a forward model for X-Ray cone beam tomography, helping us reduce the number of angles used to reconstruct and thus reducing the damage done to the biological samples. Cone beam is very useful because it is the classical setup of laboratory light sources, unlike parallel beam or fan beam which demand more expensive light sources.We focused on differential phase contrast X-Ray tomography, giving use an excellent sensibility for soft tissues. By discretizing our object, using an isotropic function like the Kaiser-Bessel window function, we where able to express the cone beam geometry projection on the detector plane as a function of the projection of the centered basis function in parallel beam geometry, greatly simplifying our problem. Our implementation in Java and ImageJ, using multi-threading an optimized algorithms, can now permit us to calculate the projection of samples in this geometry, the first and hardest step in the reconstruction problem. Bellow we show the result of projection a bat cochlea, the first image being the input ( right ) and it's rendering ( left ). The second image is the result of the our algorithm.