Language Models are being leveraged to automatically classify brain MRI reports as normal or abnormal, addressing the challenge of analyzing large volumes of clinical imaging data.

• Fine-tuned models (BERT, BioBERT, ClinicalBERT, RadBERT) successfully classify reports from 44,661 cases
• Gemini 1.5-Pro's reasoning capabilities explored for normal report categorization
• Enables automated brain growth chart generation from medical imaging data
• Tackles data heterogeneity challenges in clinical settings

This research significantly enhances clinical workflow efficiency by reducing manual analysis burden while facilitating population-level insights from previously underutilized medical imaging data.

Language Models for Automated Classification of Brain MRI Reports and Growth Chart Generation