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


     



J Nucl Med. 2008; 49 (Supplement 1):383P
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 Tylski, P.
Right arrow Articles by Buvat, I.
PubMed
Right arrow Articles by Tylski, P.
Right arrow Articles by Buvat, I.

Instrumentation & Data Analysis: Data Analysis & Management

Data Analysis & Management Posters

A method to assign statistical significance to SUV changes in patient follow-up with FDG-PET

P. Tylski1, M. Dusart2, H. Necib1, B. Vanderlinden2 and I. Buvat1

1 UMR 8165 CNRS, Orsay, France; 2 Inst. Bordet, Bruxelles, Belgium

1614

Objectives: FDG PET is increasingly used for therapy monitoring in oncology. Yet, there is no consensus on how tumor SUV should be estimated and changes in SUV should be interpreted. We propose an approach to assign statistical significance to the observed SUV changes.

Methods: Two patients with lung cancer were scanned on a GE Discovery LS PET/CT system 5 and 6 times over the course of chemotherapy (up to 15 month follow-up). By considering each tumor individually, 17 pairs of images representing the tumor evolution between two successive scans were analyzed. For each tumor, 6 independent SUV estimates were obtained: maximum SUV in the tumor (SUVmax), and mean SUV in 5 tumor volumes (manual delineation; including all voxels with value > 40% of SUVmax; including all voxels with value > a contrast-dependent threshold; obtained from a fit; 15mm diameter cylindrical region). For each pair of images, percent change in SUV ({delta}SUV) was calculated using each SUV estimate. Using this sample of 6 {delta}SUV, the hypothesis that {delta}SUV was zero was tested using a non-parametric bootstrap approach.

Results: Averaging over the 17 cases, the mean difference (±1 sd) between the smallest and largest {delta}SUV observed for the 6 SUV estimates was 29%±21%, showing a high impact of the estimation method on the observed {delta}SUV. Using the bootstrap test, {delta}SUV was found significant in 16/17 cases (p<0.05). Considering these results to establish whether {delta}SUV was indeed significant, interpretation of a single SUV estimate would have led to a wrong conclusion in 2/17 cases, where at least one {delta}SUV was less than 10%, suggesting no significant change.

Conclusions: By taking advantage of the various methods used to estimate tumor SUV in FDG PET, we introduced a simple approach to test the statistical significance of change in tumor SUV between two scans.





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 Tylski, P.
Right arrow Articles by Buvat, I.
PubMed
Right arrow Articles by Tylski, P.
Right arrow Articles by Buvat, I.