Artificial Intelligence Applied to Neuroimaging in Alzheimer's Disease: A Scoping Review of Diagnostic, Prognostic, and Biomarker‐Focused Models
Clinical Neuroimaging · 2026-01 · 3 (1)
DOI: 10.1002/neo2.70074 ↗Kurz gefasst
Scoping Review in der Zeitschrift Clinical Neuroimaging, online erschienen am 12.05.2026. Verfasst von Claire Kung und 5 weiteren Autor:innen. Die Arbeit nennt 63 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Background
Artificial intelligence (AI) applied to neuroimaging has expanded in Alzheimer's disease (AD) research, yet recent model types, data sources, and clinical aims remain heterogeneous. This scoping review maps AI‐based neuroimaging studies in AD by clinical task, data modality, study tag (specific vs multimodal), and data collection design.
Methods
We compiled recent AD studies in which AI models were applied to neuroimaging data from adult participants. We extracted neuroimaging modality, AI approach, clinical purpose, high‐level classification (diagnosis/detection, prognosis/outcome prediction, subtype discovery, biomarker prediction, image enhancement, workflow integration), study tag (Specific vs Multimodal), and data collection design (Retrospective vs Prospective). We focused on studies using MRI, PET, or combinations with clinical, cognitive, genetic, or plasma biomarkers.
Results
Most studies were retrospective and used structural T1‐weighted MRI. “Specific” studies applied deep learning models to MRI for multi‐class diagnosis and staging, distinguishing cognitively normal individuals, mild cognitive impairment (MCI) subtypes, and AD. PET‐based models were used for diagnosis and biomarker‐related tasks, including prediction of metabolic or tau changes and classification of amyloid or tau positivity. “Multimodal” studies integrated imaging with clinical or biomarker data and were prominent in prognostic and biomarker‐focused applications. Several studies targeted subtype discovery using generative or normative modeling, while explainability and workflow‐oriented tools were less common.
Discussion
AI‐based neuroimaging in AD is dominated by retrospective MRI‐ and PET‐driven deep learning models focused on diagnosis and prognostication. Multimodal approaches are increasing, but real‐world implementation studies remain limited. This scoping review provides a map of current AI–neuroimaging applications to inform future development.
Zitieren
APA
Kung, C., Shaik, R., Javier, J., Dudhia, K., Bansal, V., & Kashyap, R. (2026). Artificial Intelligence Applied to Neuroimaging in Alzheimer's Disease: A Scoping Review of Diagnostic, Prognostic, and Biomarker‐Focused Models. Clinical Neuroimaging, 3(1). https://doi.org/10.1002/neo2.70074
Vancouver
Kung C, Shaik R, Javier J, Dudhia K, Bansal V, Kashyap R. Artificial Intelligence Applied to Neuroimaging in Alzheimer's Disease: A Scoping Review of Diagnostic, Prognostic, and Biomarker‐Focused Models. Clinical Neuroimaging. 2026;3(1). doi:10.1002/neo2.70074
BibTeX
@article{Kung2026,
author = {Kung, Claire and Shaik, Roshan and Javier, James and Dudhia, Krishi and Bansal, Vikas and Kashyap, Rahul},
title = {Artificial Intelligence Applied to Neuroimaging in Alzheimer's Disease: A Scoping Review of Diagnostic, Prognostic, and Biomarker‐Focused Models},
journal = {Clinical Neuroimaging},
year = {2026},
volume = {3},
number = {1},
doi = {10.1002/neo2.70074}
}
EUVIMED