Added Value of Software-Assisted Analysis in FDG-PET for Neurodegenerative Disease Diagnosis: A Systematic Review and Meta-Analysis
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
In clinical practice, 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) evaluation of neurodegenerative diseases relies primarily on visual interpretation, which is inherently subjective. Although current international guidelines recommend incorporating quantitative tools to support visual reading, the magnitude of the incremental diagnostic benefit and the clinical contexts in which it is most pronounced have not been formally synthesized in a systematic meta-analytic framework.
Methods
Following the Preferred Reporting Items for Systematic reviews and Meta-Analyses for Diagnostic Test Accuracy (PRISMA-DTA) guidelines, we searched PubMed, EMBASE, Cochrane Library, and KoreaMed from inception to August 2025 for studies comparing visual-only versus visual-plus-quantitative FDG-PET interpretation within identical patient cohorts. Pooled sensitivity and specificity were estimated using random-effects models. Relative diagnostic performance was summarized as odds ratios (ORs), obtained by exponentiation posterior contrasts between visual analysis combined with quantitative analysis, and visual analysis. Subgroup analyses were conducted based on the clinical experience of the readers.
Results
Ten studies met the inclusion criteria. In the overall analysis (k = 9), visual analysis alone yielded a pooled sensitivity of 0.85 (95% Confidence interval: 0.77–0.91) and specificity of 0.78 (0.63–0.88), versus a sensitivity of 0.87 (0.82–0.91) and specificity of 0.88 (0.74–0.95) for the combined approach. The most pronounced gain was observed in differentiating Alzheimer’s disease (AD) from healthy controls, with specificity increasing from 0.69 to 0.94 (Bayesian OR 4.29 (95% Credible interval: 1.87–10.84)). Quantitative augmentation conferred a larger sensitivity gain among beginner readers (increasing from 0.75 to 0.87; Bayesian OR 2.39 [1.33–4.32]) than among expert readers, narrowing the performance gap between experience levels.
Conclusion
Adding quantitative analysis to visual FDG-PET interpretation yields modest overall improvements in diagnostic accuracy, with the largest gains observed in distinguishing AD from cognitively normal individuals and among less experienced readers. These findings are consistent with current international guidelines that position quantitative assessment as a complementary aid to visual interpretation rather than a replacement, with particular utility for less experienced practitioners and for specific differential-diagnostic scenarios.
