Researchers developed the MONET foundation model to connect medical images to text based on medically relevant ideas, aiding in medical AI development. They collected over 100,000 dermatology image-text pairs to train MONET to identify various concepts in dermatological images. MONET’s concept annotation capabilities were evaluated by comparing its performance to other models and used for automated data evaluation. The study found that MONET outperformed other models in annotating dermatological images. The researchers concluded that MONET can improve transparency and trustworthiness in dermatological AI systems, influencing data collection and optimization for more reliable models and enabling auditing and fairness analysis in clinical settings.
Source link
New AI tool enhances medical imaging with deep learning and text analysis
