Feasibility of estimation of brain volume and 2-deoxy-2-(18)F-fluoro-D-glucose metabolism using a novel automated image analysis method: application in Alzheimer's disease

• Published in: Hellenic journal of nuclear medicine Vol. 15; no. 3; pp. 190 - 196
• Main Authors: Musiek, Erik S, Saboury, Babak, Mishra, Shipra, Chen, Yufen, Reddin, Janet S, Newberg, Andrew B, Udupa, Jayaram K, Detre, John A, Hofheinz, Frank, Torigian, Drew, Alavi, Abass
• Format: Journal Article
• Language: English
• Published: Greece 01-09-2012
• Subjects: Aged; Algorithms; Alzheimer Disease - diagnostic imaging; Alzheimer Disease - physiopathology; Artificial Intelligence; Feasibility Studies; Female; Fluorodeoxyglucose F18 - pharmacokinetics; Humans; Image Enhancement - methods; Image Interpretation, Computer-Assisted - methods; Imaging, Three-Dimensional - methods; Male; Organ Size; Pattern Recognition, Automated - methods; Positron-Emission Tomography - methods; Radiopharmaceuticals - pharmacokinetics; Reproducibility of Results; Sensitivity and Specificity
• DOI: 10.1967/s002449910052
• ISSN: 1790-5427

The development of clinically-applicable quantitative methods for the analysis of brain fluorine-18 fluoro desoxyglucose-positron emission tomography ((18)F-FDG-PET) images is a major area of research in many neurologic diseases, particularly Alzheimer's disease (AD). Region of interest visualization, evaluation, and image registration (ROVER) is a novel commercially-available software package which provides automated partial volume corrected measures of volume and glucose uptake from (18)F-FDG PET data. We performed a pilot study of ROVER analysis of brain (18)F-FDG PET images for the first time in a small cohort of patients with AD and controls. Brain (18)F-FDG-PET and volumetric magnetic resonance imaging (MRI) were performed on 14 AD patients and 18 age-matched controls. Images were subjected to ROVER analysis, and voxel-based analysis using SPM5. Volumes by ROVER were 35% lower than MRI volumes in AD patients (as hypometabolic regions were excluded in ROVER-derived volume measurement ) while average ROVER- and MRI-derived cortical volumes were nearly identical in control population. Whole brain volumes when ROVER-derived and whole brain metabolic volumetric products (MVP) were significantly lower in AD and accurately distinguished AD patients from controls (Area Under the Curve (AUC) of Receiver Operator Characteristic (ROC) curves 0.89 and 0.86, respectively). This diagnostic accuracy was similar to voxel-based analyses. Analysis by ROVER of (18)F-FDG-PET images provides a unique index of metabolically-active brain volume, and can accurately distinguish between AD patients and controls as a proof of concept. In conclusion, our findings suggest that ROVER may serve as a useful quantitative adjunct to visual or regional assessment and aid analysis of whole-brain metabolism in AD and other neurologic and psychiatric diseases.