Diabetic macular edema (DME) and age-related macular degeneration (AMD) are the leading causes of blindness among working-age and elderly populations. Current management requires frequent imaging and manual interpretation of hundreds of optical coherence tomography (OCT) scans per patient each year, consuming up to half of each clinical visit. This technology introduces an AI-based decision-support system that integrates OCT imaging with electronic medical records to enable personalized and data-driven treatment planning for retinal diseases.
- Clinical decision support - automated analysis of OCT volumes and EMRs to suggest next treatment steps.
- Precision ophthalmology – comparison of individual disease progression and treatment response to large patient datasets.
- Diagnostic workflow optimization – automated segmentation of retinal layers and fluid accumulation for efficient review.
- Medical data integration – use of large foundation models to create a combined knowledge base, that integrates information from medical records and imaging data.
- Integrates AI computer vision and LLMs for combined image + text analysis
- Quantitative OCT interpretation with validated deep-learning segmentation (>70% Dice)
- Patient-specific treatment recommendations based on imaging and EMR data
- Clinically validated using real-world datasets from Hadassah Medical Center
AI-based segmentation and fluid-quantification models have been developed and validated on clinical OCT data from Hadassah Medical Center. Proof-of-concept achieved for LLM-based interpretation of medical records and imaging (integration of RETFound and Llava-Med frameworks).
