Optimal relaxation parameters of dynamic row-action maximum likelihood algorithm and post-smoothing filter for image reconstruction of dedicated breast PET

Background: High-resolution dedicated breast PET (dbPET) has higher photon sensitivity and spatial resolution than that of whole-body PET/CT, allowing for clearer visualization of breast lesions than whole-body PET/CT [1]. A three-dimensional list-mode dynamic row-action maximum-likelihood algorithm (LM-DRAMA) is applied to the image reconstruction of dbPET. The relaxation parameter, λ, in iterative calculations of DRAMA depends on the subset number. The noise propagation from the subset to the reconstructed image is suppressed as the subset number increases, resulting in fast convergence with a reasonable signal-to-noise ratio [2]. λ is defined by factor β, which determines the amount of reduction within one iteration. Further image smoothing using an appropriate smoothing filter is often applied after iterative reconstruction. Factor β and post-filter are important parameters that could influence the image quality of dbPET.

Purpose: To determine the optimal β-value of the relaxation control parameter and the post-smoothing filter in the LM-DRAMA , thereby detecting early-stage breast cancer with dbPET in phantom and clinical studies.

Methods: A breast phantom containing four spheres (5, 7.5, 10, and 16 mm in diameter) was filled with 18F-FDG solution (sphere-to-background radioactivity ratio, 8:1) and scanned on a dbPET scanner. The images were reconstructed using LM-DRAMA with different β-values (5, 20, or 100) and Gaussian post-filters (0, 0.78, 1.17, 1.56, 1.95, or 2.34 mm). The image quality was evaluated visually and by computing the coefficient of variation of the background (CV_BG), detectability index (DI), and contrast recovery coefficient. Parameters optimized in these phantom studies were applied to 25 clinical datasets (Table). Variabilities for different reconstruction methods in visual scores, the maximum standardized uptake value of breast cancer, and the tumor-to-background uptake ratio were estimated.

Results: The phantom images reconstructed with higher β-values and smaller post-filters yielded higher visual scores for detectability and DI, and lower smoothness and CV_BG scores (Figures 1 and 2). Based on the phantom study, the β-values and post-filters were optimized for clinical dbPET images except for β-5 and 2.34 mm. Applying the other reconstructions to clinical studies (Figure 3) showed that β-100 provided higher quantitative parameter values (Figure 4). The detectability of lesions was similar for β-100 and β-20 and decreased with larger post-filters.

Conclusions: Reconstruction using the relaxation coefficient factor β-20 and a 0.78- or 1.17-mm post-filter was optimal for dbPET imaging to detect early breast cancers. References: 1. Satoh Y, Motosugi U, Imai M, Onishi H. Comparison of dedicated breast positron emission tomography and whole-body positron emission tomography/computed tomography images: a common phantom study. Ann Nucl Med 2020;34:119-127. 2. Tanaka E, Kudo H. Subset-dependent relaxation in block-iterative algorithms for image reconstruction in emission tomography. Phys Med Biol 2003;48:1405-1422.