Related Experiment Video
Updated: May 25, 2026
09:03
Radiotracer Administration for High Temporal Resolution Positron Emission Tomography of the Human Brain: Application to FDG-fPET
Published on: October 22, 2019
Voxel-based quantitative analysis of brain images from ¹⁸F-FDG PET with a block-matching algorithm for spatial
Christophe Person1, Valérie Louis-Dorr, Sylvain Poussier
1CRAN, CNRS UMR 7039, Nancy, France.
Clinical Nuclear Medicine
|February 8, 2012
Summary
Block-matching (BM) normalization improves brain FDG PET analysis accuracy compared to Statistical Parametric Mapping (SPM). BM reduces image distortion and false positives, especially for significant brain abnormalities.
Area of Science:
- Neuroimaging
- Medical Image Analysis
- Positron Emission Tomography (PET)
Background:
- Statistical Parametric Mapping (SPM) is standard for analyzing ¹⁸F fluorodeoxyglucose positron emission tomography (FDG PET) brain images.
- SPM's spatial normalization step can cause image distortion and artifacts, particularly with brain abnormalities.
Purpose of the Study:
- To evaluate a block-matching (BM) normalization algorithm for FDG PET brain imaging.
- To compare the performance of BM normalization against SPM normalization in minimizing artifacts.
Main Methods:
- Artificially simulated hypometabolic areas (large and small) in normal FDG PET images.
- Compared statistical analysis results using SPM versus BM normalization.
Main Results:
- BM normalization reduced errors in estimating large defect volumes by approximately 50% compared to SPM.
- BM normalization decreased false-positive rates when simulating numerous or extended abnormalities.
- Findings were corroborated using FDG PET scans from epileptic patients.
Conclusions:
- Block-matching (BM) normalization offers more precise and robust results for brain FDG PET analysis than SPM.
- BM normalization is particularly advantageous for cases with numerous or extensive brain abnormalities.
