Experimental Study on Multivalued Phantoms Using Different Filters in the DART Algorithm

In our previous paper [10] we proposed new variants of the Discrete Algebraic Reconstruction Technique with combined filtering technique and performed experiments on binary software phantoms within a new test framework to investigate the effect of the filters. Continuing our work, in this paper we extend our study to multivalued phantoms for a deeper investigation in this field. We create a new test phantom set with different intensity levels and perform a comprehensive experimental study. The results are evaluated with Relative Mean Error which is extended to multivalued discrete phantoms. We use a ranking system to create different views to our quantitative data. Finally, the achievements are discussed.