SNM Annual Meeting Abstracts
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     



J Nucl Med. 2008; 49 (Supplement 1):23P
This Article
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Google Scholar
Right arrow Articles by Chen, K.
Right arrow Articles by Reiman, E.
PubMed
Right arrow Articles by Chen, K.
Right arrow Articles by Reiman, E.

Instrumentation & Data Analysis: Data Analysis & Management

Quantitative Methods

An accurate and computationally efficient method for voxel-by-voxel estimation of the Pittsburgh compound B distribution volume ratio

K. Chen1, X. Liu1, Y. Zhou2, D. Bandy1, M. Yu1, J. Keppler1, W. Klunk3, C. Mathis3, N. Foster5, M. Weiner4, R. Koeppe6, W. Jagust7 and E. Reiman1

1 Banner Alzheimer's Inst, Phoenix, Arizona; 2 Johns Hopkins Univ, Baltimore, Maryland; 3 Univ of Pittsburgh, Pittsburgh, Pennsylvania; 4 Univ of Calif SF, San Francisco, California; 5 Univ of Utah, Salt Lake City, Utah; 6 Univ of Michigan, Ann Arbor, Michigan; 7 Univ of Calif Berkeley, Berkeley, California

91

Objectives: Distribution volume ratio (DVR) provides a quantitative index of fibrillar amyloid deposition in PIB-PET studies. We introduce a more computationally efficient method for estimating PIB DVR than non-linear fitting.

Methods: Data from 5 subjects were included. The generalized linear least square (GLLS) technique (Chen et al., 1998) was re-derived for the reference tissue model with 3 parameters (RTM3P) with a common k’2R (Zhou, et al, 2007), simultaneously estimating PIB DVR and R1 (tracer extraction relative to the reference region). DVR was estimated by GLLS and the RTM3P non-linear fitting over a number of cerebral ROI. The two approaches were compared by linearly regressing GLLS DVR against RTM3P DVR and examining the closeness of the regression slope/intercept to 1.0/0.0 and the correlation coefficient to 1.0. GLLS generated DVR parametric image were also compared to those using the Logan method.

Results: GLLS generated the DV/R1 parametric images in under a few minutes. GLLS-derived DVR was similar to that of the nonlinear RTM3P (regression slope 1.02, intercept=-0.074, correlation coefficient R=0.987, p=1.2e-11). The DVR parametric images by GLLS resembled those by Logan with the regression slopes ranging from 1.06 to 1.09.

Conclusions: GLLS is accurate and computationally efficient in estimating DVR and R1. Further studies are needed for its use in the differential diagnosis, early detection and tracking of AD.

Research Support: ADNI, Arizona Alzheimer's Consortium





This Article
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Google Scholar
Right arrow Articles by Chen, K.
Right arrow Articles by Reiman, E.
PubMed
Right arrow Articles by Chen, K.
Right arrow Articles by Reiman, E.