EEG-to-Report Framework Automates Clinical EEG Reporting

Clinical electroencephalography (EEG) reporting is a labor-intensive manual process, and existing software often lacks the structured data generation needed for training modern AI models. To bridge this gap, researchers developed EEG-to-Report, a browser-based framework designed to integrate routine EEG review with the creation of high-quality, AI-ready datasets. The framework supports multi-format EEG ingestion, channel standardization, and features an interactive viewer with a multimodal annotation layer, allowing for both typed text and transcribed voice notes. For each annotated segment, a feature extraction engine automatically computes a standardized set of spectral, temporal, and other physiological descriptors, storing them alongside clinical descriptions in a portable JSON schema. This process yields aligned feature-text pairs ideal for supervising multimodal EEG-language models. EEG-to-Report also includes an auto-report module that combines convolutional networks with a large language model to draft preliminary clinical narratives for neurologist review. This system streamlines annotation workflows and produces editable draft reports, laying a foundation for more automated and efficient EEG reporting systems.