ART for helical cone-beam CT reconstruction

We report on our first results on the use of Algebraic Reconstruction Techniques (ART) on helical cone-beam Computerized Tomography (CT) data. Two variants of ART have been implemented: a standard one which considers a single ray in an iterative step and a block version which groups several conebeam projections in calculating an iterative update. Both seem to produce high-quality reconstructions, although the number of cycles through the data to achieve those (between 15 and 20), while not huge, is larger than the number of cycles through the data needed for reconstructing volumes from data acquired from different modalities (1 iteration for PET data and 1 to 4 iterations for EM data). The reason for that maybe due to the uneven coverage of points by the data collection geometry, resulting in a slower rate of convergence.