Diagnostic Accuracy of Radiomics Features from ^18F-FDG PET in Predicting Conversion from Mild Cognitive Impairment to Alzheimer’s Disease: A Systematic Review and Meta-analysis
Ahmed Moawed1, Sara Elayan2, Mayar Sinjilawi2, Mohamed ElMasry3, Mohamed Salem4, Abdallah Ibrahim5, Khaled M.H Mohamed6
1Faculty of Medicine, Mansoura University, Dakahlia, Egypt, 2Faculty of Medicine, Jordan University of Science and Technology, Irbid, Jordan, 3Faculty of medicine Suez university, Suez Egypt, 4Faculty of Medicine, Damietta university, Damietta, Egypt, 5Faculty of medicine, Alexandria University, Alexandria, Egypt, 6Visiting scholar, Pharmaceutical Sciences Department, College of Health and Human Sciences, North Dakota State University, Fargo, ND.
Objective:

To determine the pooled diagnostic accuracy of radiomics features derived from Fluorine-18 Fluorodeoxyglucose Positron Emission Tomography(^18F-FDG PET (in predicting the conversion from mild cognitive impairment (MCI)to Alzheimer’s disease (AD)

Background:

Radiomics applied to ^18F-FDG PET provides a new way to detect subtle brain changes linked to AD, as it reflects regional cerebral glucose metabolism and can reveal characteristic metabolic patterns in patients suspected of having AD. Predicting which individuals with MCI will progress to AD is still challenging

Design/Methods:

We conducted a comprehensive search of PubMed, Embase, Scopus, and Web of Science up to July 2025. We included observational studies assessing radiomic features extracted from ^18F-FDG PET in adults with MCI to predict progression to AD. Methodological quality was assessed using the QUADAS-2 tool, and data were independently extracted by two reviewers. We used a bivariate random-effects model to pool sensitivity, specificity, positive and negative likelihood ratios, and diagnostic odds ratio. I², τ², and Q statistics were used to explore heterogeneity

Results:

Eight studies were identified, including a total of 2,735 patients with MCI. The pooled sensitivity and specificity of the radiomics-based model across five independent cohorts were 0.72 (95% CI: 0.50–0.88; I² = 49.9%) and 0.94 (95% CI: 0.77–0.98; I² = 85.3%), respectively. The pooled  area under the curve (AUC) and c-index were 0.798 (95% CI: 0.745–0.851; I² = 77.9%) and 0.775 (95% CI: 0.702–0.848; I² = 98.6%) across five and six cohorts, respectively. The sensitivity analysis revealed that most of heterogeneity related to a single outlier study; its exclusion reduced I² without affecting the robustness of the results


Conclusions:

Radiomics features extracted from ^18F-FDG PET show promising diagnostic performance in predicting the conversion from MCI to AD. The pooled estimates indicate high specificity and acceptable sensitivity, suggesting that radiomics can complement conventional imaging assessment in identifying individuals at higher risk of progression

10.1212/WNL.0000000000216271
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.