Artificial Intelligence Applications in Sickle Cell Retinopathy Imaging: Current Progress, Challenges, and Future Directions

Résumé fourni par la source

Introduction: Sickle cell retinopathy (SCR) is a leading cause of vision loss in patients with sickle cell disease, but its detection and monitoring remain difficult due to heterogeneous retinal microvascular changes and reliance on expert interpretation. Artificial intelligence has shown success in retinal disease detection, classification, and staging on retinal images, achieving expert-level performance. This review summarizes recent artificial intelligence progress in SCR imaging and highlights future opportunities for clinical translation. Materials and Methods: This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A comprehensive literature search was performed in PubMed/MEDLINE, Embase, and Web of Science. Results: A comprehensive PubMed search returned 15 records. After removal of duplicates and screening of titles and abstracts, 7 full-text articles were assessed for eligibility. Of these, 4 studies met inclusion criteria and were included in the final review. Three full text articles were assessed for eligibility and inclusion from non-PubMed sources. Two studies applied classical machine learning on features extracted from imaging data to classify SCR. Four studies used deep-learning algorithms to detect SCR features on ophthalmological images. One study applied deep learning algorithms to classify SCR from other retinal diseases. Conclusions: Deep learning has the potential to improve detection, stage, and monitor sickle cell retinopathy across multiple ophthalmological imaging methods. Additional research is needed to support clinical adoption. With continued development, AI-based tools could enhance diagnostic precision, enable personalized care, and ultimately improve outcomes for patients with sickle cell retinopathy.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.