Quantitative Evaluation of Beta-Amyloid Brain PET Imaging in Dementia: A comparison between both Hermes BRASS and Siemens Syngo.VIA Amyloid Plaque Quantification with the clinical report


Purpose: To compare commercially available image analysis tools Hermes BRASS and Siemens Syngo.VIA Amyloid plaque with clinical assessment in 18F-Florbetapir PET scans of patients with Dementia. Method: PET/CT scans of 225 patients injected with 370MBq 18F-Florbetapir were reported by three experienced clinicians and quantified using two software packages. To aid reporting, scans were classified into type A (typical features) or non-type A (atypical features) for both positive and negative scans. Both software packages produce regional amyloid uptake ratio relative to cerebellum (SUVr). For BRASS, scans with z-score (deviations from cortical ROI SUVr) ≥2, in at least 2 defined regions was classed positive. For Syngo.VIA a positive scan was indicated when average SUVr > 1.17. The software’s ability to correctly identify positive/negative subtypes was assessed. Results: Of the 225 18F-Florbetapir (Amyvid) amyloid scans, 182 (81%) scans were type A, and 43 (19%) scans were non-type A. The sensitivity of BRASS and Syngo.VIA for type A scans was 98.8% and 96.3%, and the specificity was 73% and 92%, respectively. A third threshold of identifiable levels of plaque (1.08 < SUVr < 1.17) was recommended for Syngo.VIA to increase detection of false negative scans. The false positive rate of BRASS significantly decreased when an alternative positive threshold value of SUVr ≥ 1.18 was introduced. The introduction of these two criteria resulted in similar improvements of sensitivity and specificity for non-type A scans but all results were taken with caution as a small sample size as used. A larger sample size is recommended for further analysis. Conclusion: The study results show that using the Hermes criteria for a positive scan leads to a high sensitivity but a low specificity (greater false positive rate). The Siemens Syngo.VIA criteria which is based on relative SUV gives a high sensitivity and specificity and agrees better with the clinical report. Alternative thresholds and classifications may help to improve agreement with the clinical report. Software packages may assist with clinical reporting of more difficult to interpret cases that require a more experienced read. In the majority of scans with typical features, the quantification toolkits agree with an experienced clinical report.