| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
|
|||||||||
|
|
Instrumentation & Data Analysis: Data Analysis & ManagementQuantitative Methods |
1 National Institute of Radiological Sciences, Chiba, Japan; 2 Tokyo Metropolitan Institute of Gerontology, Tokyo, Japan
88
Objectives: We propose and evaluate a MAP estimation algorithm in likelihood based graphical analysis (MEGA) to reduce noise-induced bias and variance of the total volume of distribution (VT) using widely used Logan graphical analysis (GA) for neuroreceptor imaging.
Methods: In MEGA, a set of time–activity curves (TACs) was formed with VT varying in physiological range as a template, and then the most similar TAC was sought out for a given measured TAC in a feature space. In simulation, MEGA were compared with other three methods, GA, Multilinear analysis (MA1), and likelihood estimation in GA (LEGA) using 500 noisy TACs under seven physiological conditions (from 9.9 to 61.5 of VT). PET studies of 11C-SA4503, a ligand for 
1 receptors, were performed for three normal volunteers at baseline and after loading selective serotonin reuptake inhibitors. In clinical studies, the VT images of estimated by MEGA were compared with ROI estimates by nonlinear least square fitting (NLS) over 4 brain regions.
Results: In simulation, the estimated VT by GA had large underestimation (y = 0.27x + 8.72, r2 = 0.87). Applying the other methods (MA1, LEGA, and MEGA), this bias was improved (y = 0.80x + 4.04, r2 = 0.98; y = 0.85x + 3.05, r2 = 0.99; y = 0.96x + 1.21, r2 = 0.99, respectively). MA1 and LEGA increased variance of the estimated VT in simulation and clinical VT images. However, MEGA improved SNR in VT images with linear correlations between ROI estimates by NLS (y = 0.87x + 5.1, r2 = 0.96).
Conclusions: MEGA could improve estimates of VT in clinical neuroreceptor PET imaging.
| ||||||||||||||||||||||||||||||||||||||
