Comparison between MAP and postprocessed ML for image reconstruction in emission tomography when anatomical knowledge is available

Abstract

Previously, the noise characteristics obtained with penalized-likelihood reconstruction [or maximum a posteriori (MAP)] have been compared to those obtained with postsmoothed maximum-likelihood (ML) reconstruction, for emission tomography applications requiring uniform resolution. It was found that penalized-likelihood reconstruction was not superior to postsmoothed ML. In this paper, a similar comparison is made, but now for applications where the noise suppression is tuned with anatomical information. It is assumed that limited but exact anatomical information is available. Two methods were compared. In the first method, the anatomical information is incorporated in the prior of a MAP-algorithm and is, therefore, imposed during MAP-reconstruction. The second method starts from an unconstrained ML-reconstruction, and imposes the anatomical information in a postprocessing step. The theoretical analysis was verified with simulations: small lesions were inserted in two different objects, and noisy PET data were produced and reconstructed with both methods. The resulting images were analyzed with bias-noise curves, and by computing the detection performance of the nonprewhitening observer and a channelized Hotelling observer. Our analysis and simulations indicate that the postprocessing method is inferior, unless the noise correlations between neighboring pixels are taken into account. This can be done by applying a so-called prewhitening filter. However, because the prewhitening filter is shift variant and object dependent, it seems that MAP reconstruction is the more efficient method.

18F-FDOPA PET and PET/CT Accurately Localize Pheochromocytomas

“…Eight to 10 bed positions with a 58.3-cm transaxial field of view were measured. The images were reconstructed with iterative techniques: maximum a posteriori maximization (16) for the transmission scan and ordered-subset expectation maximization consisting of 2 iterations with 8 subsets (17) for the emission scan. Corrections for attenuation and scatter were applied.…”

18F-FDOPA Kinetics in Brain Tumors

“…The emission scan was acquired in the 3-dimensional mode. A total of 26 frames were acquired at 8 · 15 s, 2 · 30 s, 2 · 60 s, and 14 · 300 s. The images were reconstructed with iterative techniques: maximum a posteriori maximization (13) for the transmission scan and ordered-subset expectation maximization consisting of 6 iterations with 8 subsets (14) for the emission scan. Corrections for attenuation and scatter were applied.…”

The Use of Anatomical Information for Molecular Image Reconstruction Algorithms: Attenuation/Scatter Correction, Motion Compensation, and Noise Reduction

“…Note that guided image filter is much faster than NLM filter in computation. Although post-reconstruction filters are computationally fast, using regularization in iterative image reconstruction is preferable over post-reconstruction filtering for better image quality [48]. There have been many attempts to incorporate the anatomical image g in the penalized ML framework:…”

“…Ideas of using structural couplings between molecular and anatomical images for reconstruction have been studied a couple of decades ago [41][42][43]. Recently, interesting advances for noise reduction of molecular images using anatomical information have been introduced with state-of-the-art methods for post-reconstruction filtering [44][45][46][47] or regularization in inverse problems [48][49][50][51][52][53].…”