Abstract
Background & Aims
Adult diffuse gliomas present a global healthcare challenge, with LMICs facing diagnostic and treatment barriers due to limited neuroimaging, molecular testing, and expertise. These constraints delay management and worsen outcomes. AI offers cost-effective, scalable solutions for neuro-oncology in resource-limited settings. This review examines AI’s potential to enhance glioma diagnosis, prognostication, and treatment monitoring in LMICs.
Methods
We reviewed AI applications in neuro-oncology, focusing on tumor segmentation, deep learning-based glioma classification, and MRI-based molecular prediction. Feasibility in LMICs was evaluated, particularly radiogenomics for IDH mutation and 1p/19q co-deletion prediction. We also analyzed AI-driven prognostic models and radiomic differentiation of tumor progression from pseudoprogression, emphasizing the need for LMIC-representative datasets.
Results
AI tools help bridge diagnostic and management gaps in LMICs. Radiogenomic models predict molecular markers non-invasively, reducing the need for costly genomic tests. AI-driven prognostic models integrate imaging and clinical data for personalized risk stratification. Machine learning aids in distinguishing tumor progression from pseudoprogression, reducing reliance on expert interpretation. However, challenges like data scarcity, limited generalizability, and insufficient clinical validation in LMICs hinder implementation.
Conclusions
AI has the potential to revolutionize glioma management in LMICs by improving early detection, prognostication, and treatment monitoring while reducing reliance on costly infrastructure. However, successful implementation requires overcoming data limitations, ensuring model generalizability, and fostering multidisciplinary collaborations. Strengthening AI research, capacity building, and policy support is essential to achieving equitable access to AI-driven neuro-oncology innovations in resource-limited settings.