[Evaluation of commercial resolution recovery techniques in four state-of-the-art single photon emission computed...

This study aims to evaluate the fundamental performance of four leading advanced resolution recovery methods.

To evaluate the performance of the resolution recovery algorithm, we carried out the computer simulation with the cone/sphere digital phantoms. These phantoms were used to investigate the basic properties of those algorithms. The software of four packages (advance) were also tested, specifically Astonish(TM) (AST), Evolution(TM) (EVL), Flash-3D(TM) (FL3), and 3D-OSEM (3DOS). The performance was evaluated in the collimator systems (LEHR) reconstruction conditions using the full width at half maxi am (FWHM), aspect ratio (ASR), and artifacts of conical part.

In the "without BG," FWHM of the advance method indicated a true-FWHM with SI (subset×iteration)=20, 40. As SI increased, FWHM was composed with over estimate. Each advances of FWHM indicated only 5% of improvement as compared with reference FWHM in the "with BG." The ASR increased 20% to AST, FL3, and ASR of 3DOS remained in 10% in the outside. As for the reproducibility of the conical part, an artifact was caused by the FL3, EVL, and AST methods. This artifact did not occur in 3DOS.

An SI needs more than 150 to obtain an accurate compensation effect. As for the advance method, the major compensation effect was not demonstrated very much as compared with the OS-EM. The EVL, FL3, and AST overestimated values due to a Gibb's oscillation in the artifacts of the conical part.