We present a FAST implementation of the Inverse Attenuated Radon Transform which incorporates accurate collimator response, as well as artifact rejection due to statistical noise and data corruption. This new reconstruction procedure is performed by combining a memory-efficient implementation of the analytical inversion formula (AIF , ) with a wavelet-based version of a recently discovered regularization technique . The paper introduces all the main aspects of the new AIF, as well numerical experiments on real and simulated data. Those display a substantial improvement in reconstruction quality when compared to linear or iterative algorithms. © 2011 IEEE.
|Title of host publication
|2011 IEEE International Workshop on Machine Learning for Signal Processing
|Institute of Electrical and Electronics Engineers (IEEE)
|Published - Sep 2011
Bibliographical noteKAUST Repository Item: Exported on 2020-10-01
Acknowledgements: This work was partially funded by KAUST and EPSRC. PEB was also sponsored by the Chinese Academy of Sciences.
This publication acknowledges KAUST support, but has no KAUST affiliated authors.