Abstract
A New PET Image Quality Phantom, Acquisition, and Analysis Paradigm
IEEE conference record - Nuclear Science Symposium & Medical Imaging Conference, pp.1-1
11/01/2025
DOI: 10.1109/NSS/MIC/RTSD57106.2025.11287287
Abstract
A new PET image quality phantom, acquisition and analysis paradigm is presented. The new phantom is based upon the NEMA IQ phantom geometry but includes additional smaller spheres down to 6 mm and additional small spheres at an increased radius. The methodology utilizes a single bed- 40 minutes acquisition divided into multiple short duration realizations defined by the axial field of view of the system for fair comparison. Four PET systems from 3 manufacturers were assessed in a test-retest experiment. In this study, a discriminability metric d is proposed to evaluate system lesion detectability performance, using data from the multiple realizations. d is defined as the difference between the mean activity concentration of the multiple realizations of the sphere and the multiple realizations of the background concentrations, in units of the number "standard deviations" between the background and sphere distributions. To determine whether the detectability metric d has clinical relevance, each reconstructed realization from each of the four systems was reviewed to assess detectability of each of the 12 spheres according to a 4 -point Likert scale with 3= "Certainly Visualized", 2= "Probably Visualized", 1= "Equivocal", and 0= "Not Visualized". The Likert scores were then correlated with the discriminability metric d for each scanner. Results of the analysis were consistent with the Rose Criterion, suggesting that d may have utility as a meaningful lesion detectability metric.
Details
- Title: Subtitle
- A New PET Image Quality Phantom, Acquisition, and Analysis Paradigm
- Creators
- M. Zhang - SiemensY. Bercier - SiemensT. Deller - National Transportation Research CenterY. Wu - United Imaging HealthcareL. He - United Imaging HealthcareE. Leung - United Imaging HealthcareC. Schutze - Nephrologisches Zentrum GoettingenS. Siegel - SiemensG. Wells - Ottawa Heart InstituteT. Turkington - Duke UniversityJ. Sunderland - University of Iowa
- Resource Type
- Abstract
- Publication Details
- IEEE conference record - Nuclear Science Symposium & Medical Imaging Conference, pp.1-1
- DOI
- 10.1109/NSS/MIC/RTSD57106.2025.11287287
- eISSN
- 2577-0829
- Publisher
- IEEE
- Language
- English
- Date published
- 11/01/2025
- Academic Unit
- Radiology; Physics and Astronomy; Radiation Oncology
- Record Identifier
- 9985096041702771
Metrics
1 Record Views