Decoding the sella: a review of MRI in pituitary lesions—from dynamic imaging to artificial intelligence

Pituitary lesions are increasingly detected due to widespread use of magnetic resonance imaging (MRI), with incidental findings (“incidentalomas”) now identified in 10–20% of brain imaging studies. MRI remains the gold standard for evaluating sellar and parasellar pathology, offering superior soft-tissue contrast without ionizing radiation. This comprehensive narrative review synthesizes current evidence on the role of MRI in evaluating pituitary lesions, with a primary focus on intrasellar pathologies and post-operative imaging. We highlight technical advances from dynamic contrast-enhanced (DCE) imaging to emerging artificial intelligence (AI) applications. Dynamic contrast-enhanced MRI significantly improves microadenoma detection, with sensitivity exceeding 90% compared to approximately 50% with static imaging. High-field 3 T imaging offers 10–15% improved detection compared to 1.5 T systems. Post-operative baseline imaging optimally performed at 3–4 months minimizes false-positive residual tumor interpretation. AI-based radiomics models demonstrate promising accuracy (AUC 0.85–0.95) for predicting tumor consistency, invasiveness, and recurrence, though prospective validation remains limited. A systematic approach to pituitary MRI interpretation, integrating optimized protocols with emerging AI tools, enhances diagnostic accuracy and guides clinical decision-making.