Development of a Deep Learning Model to Estimate Anemia from Palpebral Conjunctiva Taken with a Portable Slit-Lamp Microscope.

Deep learning analysis of palpebral conjunctiva images acquired with a portable slit-lamp microscope demonstrated the feasibility of non-invasive hemoglobin estimation and anemia screening.

Although the proposed approach achieved moderate performance, further improvements in model accuracy and prospective multi-center validation are required before clinical implementation.