RT Journal Article SR Electronic T1 Evaluation of a quantitative analysis method using SUVRs for Flutemetamol PET JF Journal of Nuclear Medicine JO J Nucl Med FD Society of Nuclear Medicine SP 1641 OP 1641 VO 59 IS supplement 1 A1 Sara Pirozzi A1 David Mirando A1 Alex Kruzer A1 Aaron Nelson YR 2018 UL http://jnm.snmjournals.org/content/59/supplement_1/1641.abstract AB 1641Purpose: Previous studies have shown value in quantifying amyloid burden in PET brain scans in order to improve diagnostic confidence and accuracy of Alzheimer’s Disease. One method of quantification involving the computation of standard uptake value ratios (SUVRs) has demonstrated the ability to separate Amyloid positive scans from negative scans with a high degree of accuracy and consistency, using several approved tracers such as Florbetapir and Florbetaben1,2. The aim of this study is to explore the accuracy of a quantitative analysis method for Vizamyl PET scans using SUVRs to differentiate between patients with and without amyloid accumulation. Methods: Vizamyl PET imaging data for 109 clinically diagnosed healthy controls (M:F-47:62; mean age 72.1 years) and 18 AD subjects (M:F-10:8; mean age 74.2 years) were acquired from the Australian Imaging, Biomarkers & Lifestyle Study of Aging (AIBL) (www.aibl.csiro.au). Normal subjects were further classified as Amyloid negative by an expert reader. From the normal population 54 patients (M:F-22:32, mean age 72.3 years) were sub-selected as the training dataset and used to derive an average SUVR cutoff, computed as the average SUVR across all brain analysis regions (each normalized to the whole cerebellum) + 1.65 SD. Clinically significant brain regions analyzed included the Inferior Medial Frontal Gyrus, Lateral Temporal Lobes, Precuneus, Posterior Cingulate Gyrus, Anterior Cingulate Gyrus, and the Superior Parietal Lobule. The training dataset included roughly half the normal subjects from each age range so that there would be a similar age distribution between the training normals and testing normals. The average SUVR cutoff was then applied to the test dataset comprised of 55 normal healthy controls (M:F-25:30; mean age 72 years) and 18 AD subjects. Diagnostic sensitivity, specificity, and accuracy were computed across the entire test set, where subjects with average SUVRs under the cutoff were considered Amyloid negative (normal) and those above the cutoff were considered Amyloid positive. Results: Using the SUVR cutoff of 1.41 resulted in 97% accuracy across the entire test set with a sensitivity and specificity of 89% and 100%, respectively. All 55 normals tested had an average SUVR under the cutoff and therefore were correctly identified as normal. For the AD cases, 16/18 were correctly classified as Amyloid positive. Conclusion: With 97% classification accuracy, SUVR analysis for the Vizamyl tracer shows promise in separating Amyloid positive scans from negative scans using a cutoff derived from normal controls. Accuracy could further improve by comparing to the classification of Amyloid positive rather than clinical diagnosis, as there could be patients presenting with other forms of dementia who are misdiagnosed as AD. At 97% accuracy, this SUVR based approach has potential to be a helpful aid to visual interpretation of Vizamyl PT images. References: 1. Pontecorvo MJ, Arpra Al. Devine M. et al. Potential Value of a Quantitative Estimate of Cortical to Cerebellar SUVr in Aiding Visual Interpretation of Florbetapir PET Scans. Eur J Nucl Med Mol Imaging 2013; 40 (Suppl 2):S143. 2. Piper JW, Nelson AS, Javorek A. Evaluation of a Quantitative Method for Florbetaben (FBB) PET Using SUVR. Oral Presentation EANM Annual Meeting 2014.