Niccolo' Fineschi
Through the eyes of a Vision-Language Model: Age-related Macular Degeneration analysis using fundus images.
Rel. Samanta Rosati, Gabriella Balestra, Jovana Panic. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Biomedica, 2026
Abstract
Age-related Macular Degeneration (AMD) is a progressive retinal disease and leading cause of irreversible vision loss worldwide. Characterized by heterogeneous phenotypes, AMD severity spectrum evolves from early pathological changes to advanced stages. As clinical intervention is mainly focus on slowing down its unfolding and most effective when administered early, prompt evaluation of disease onset and progression is critical. Retinal fundus images offer a non-invasive, accessible and cost-effective imaging modality for widespread screening across diverse clinical settings. In this work, technical translation of this clinical question is addressed through three interconnected tasks: classification of current disease status, discrete prediction of future progression, longitudinal trajectories forecasting.
Probing the latent space of several Vision-Language Models, RET-CLIP is demonstrated to achieve superior representation quality in dichotomizing healthy and AMD pathological eyes, as confirmed by statistical evaluations and multiple Machine Learning models implemented
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Informazioni aggiuntive
Corso di laurea
Classe di laurea
Aziende collaboratrici
URI
![]() |
Modifica (riservato agli operatori) |
