Artificial Intelligence for Brain MRI Diagnosis: A Scoping Review and Evidence-Gap Map Across Neurological Diseases.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
INTRODUCTION/OBJECTIVE: Artificial Intelligence (AI), including machine learning and deep learning, is increasingly applied to brain Magnetic Resonance Imaging (MRI) for disease detection, classification, segmentation, and clinically relevant prediction. This review aimed to map the evidence on AI-assisted brain MRI diagnosis, identify methodological and clinical translation gaps, and propose priorities for future research. METHODS: This scoping review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews. PubMed/MEDLINE, Scopus, IEEE Xplore, and Web of Science were searched for peer-reviewed English-language publications from January 2018 to April 2026. Data were charted according to disease domain, study design, AI method, dataset, validation strategy, performance metrics, explainability, and key limitations. RESULTS: The search identified 1,020 records. After duplicate removal and screening, 61 publications were included, comprising 24 original studies used for the primary evidence map and 37 reviews or commentaries used for contextual synthesis. Evidence was strongest for brain tumor imaging and Alzheimer's disease, whereas stroke, multiple sclerosis, pediatric neuroimaging, and rare neurological conditions were underrepresented. Most studies relied on public benchmark datasets and retrospective validation, whereas external validation, prospective clinical evaluation, subgroup analysis, explainability, and workflow integration were inconsistently reported. DISCUSSION: The findings indicate that AI-assisted brain MRI research remains unevenly developed and is constrained by dataset dependence, limited external validation, limited prospective clinical evaluation, and inconsistent reporting of explainability, subgroup performance, and workflow integration. CONCLUSION: AI shows promising research-stage potential for brain MRI-based diagnosis, particularly in brain tumor analysis and Alzheimer's disease classification. Future research should prioritize multicenter validation, diverse and disease-specific datasets, explainable and equitable AI design, transparent reporting, and prospective evaluation within real clinical workflows.