Researchers have developed a novel two-stage multimodal framework for interpreting chest X-rays, integrating radiologist eye-tracking data to improve diagnostic accuracy and report generation. The first stage employs a gaze-token classifier that fuses image patches, transcriptions, and radiologist fixations, with added fixation supervision boosting AUC by 4.4% and F1 by 13.3%. The second stage translates these predictions into region-specific diagnostic sentences, extracting confidence-weighted keywords and using a prompted large language model to enhance clinical-term scores. This approach offers a new benchmark for interpretable, gaze-aware chest X-ray analysis. AI
IMPACT Enhances diagnostic accuracy and transparency in medical imaging analysis, potentially improving patient care.
RANK_REASON Research paper detailing a novel multimodal learning framework for medical imaging analysis. [lever_c_demoted from research: ic=1 ai=1.0]
- alphaXiv
- arXiv
- BERTScore
- CatalyzeX
- DagsHub
- Gotit.pub
- Hugging Face
- MIMIC-Eye
- Rouge
- ScienceCast
- Shuchismita Anwar
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
