Compressive-Sensing-Tomography

Here we explain the utility of Compressed Sensing in Tomographic reconstruction. We take 18 parallel beam projections from a MR volume of the brain and perform filtered back-propagation using Ram-Lak filter , individual compressive reconstruction and 2-3 slice cs recovery for utilising the consecutive capture redundancies.

How to reproduce the results ?

The code is present in /Code. Main code is present in TomographyCS.m. You will find the function handles used to mimic radon(idct2()) and iradon(dct2()) inside the @radonDCT, @radonDCT2, @radonDCT3 folders.

execute TomographyCS.m to reconstruct the results.

Results

Original Slice

FBP Ram-Lak reconstruction with 18 projection angles

CS based reconstructions(individual, 2-slice, 3-slice)

Forward Model Matrix for 2-slice and 3-slice reconstruction