Toward Reliable Diabetic Retinopathy Screening
Sensors · 16.07.2026 · 26 (14) · 4515
DOI: 10.3390/s26144515 ↗Kurz gefasst
Fachartikel in der Zeitschrift Sensors, online erschienen am 16.07.2026. Verfasst von Hendrio Bragança und 5 weiteren Autor:innen. Beteiligte Einrichtung: Núcleo de Capacitação em Inteligência Artificial (NCIA), Fundação Paulo Feitoza-FPFtech, Av. Danilo de Matos Areosa, 1170, Distrito Industrial, Manau…. Hauptthemen laut MEDLINE: Diabetic Retinopathy. Die Arbeit nennt 72 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Fundação Desembargador Paulo Feitosa (FPFtech), Innovation and Training Center in Artificial Intelligence (NCIA) at the Paulo Feitosa Foundation, Foxconn Brasil.
Abstract (Originalsprache)
Diabetic retinopathy (DR) grading requires reliable five-grade severity assessment under substantial acquisition variability and cross-dataset distribution shift. We propose PRISM-DR, a multi-objective five-grade DR grading framework trained under a gradient-partitioned strategy. The architecture is organized as a feedforward pipeline: a data-driven preprocessing stage followed by a ConvNeXtV2-Base backbone, a Recurrent BiFPN neck for multi-scale feature fusion, a Frequency-Aware Fusion module, a lightweight multi-scale reasoning transformer, dual classification heads with gradient-isolated pathways (categorical and ordinal), and a prototype memory module for embedding regularization. The CORAL ordinal head operates through a dedicated projection layer and is gradient-isolated from the backbone; the backbone is shaped by the cross-entropy, prototype contrastive, and view-consistency objectives, which carry indirect ordinal signal through severity-weighted class penalties and grade-indexed cluster regularization. The model is trained in a multi-crop setting with a phased loss curriculum designed for severely imbalanced DR datasets. Evaluated across six datasets under Fixed-Source, Multi-Target (FSMT) protocols, PRISM-DR trained on EyePACS + DDR achieves QWK of 0.835 on IDRiD, 0.865 on APTOS2019, and 0.720 on Messidor-2, with in-domain QWK = 0.920 and AUC-PR = 0.941 on EyePACS, outperforming RETFound, RETFound-Green, and MedGemma-4B in AUC-PR across all evaluated datasets. Quantitative interpretability evaluation against 755 expert-annotated lesion images yields 8.0× Energy Ratio Enrichment and a FAF gate retention ratio of 4.4× inside lesion regions, confirming that anatomically plausible spatial priors emerge from grade-level supervision alone, without pixel-level annotation. PRISM-DR establishes a superior accuracy–robustness–capacity trade-off for scalable, automated DR screening.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Núcleo de Capacitação em Inteligência Artificial (NCIA), Fundação Paulo Feitoza-FPFtech, Av. Danilo de Matos Areosa, 1170, Distrito Industrial, Manaus 69075-351, Amazonas, Brazil.
ORCID-Profile: Wington L Vital, Antonio Fontenele
Förderung
- Fundação Desembargador Paulo Feitosa (FPFtech)
- Innovation and Training Center in Artificial Intelligence (NCIA) at the Paulo Feitosa Foundation
- Foxconn Brasil
Zitieren
APA
Bragança, H., Caliari, Í. P., Vital, W. L., Fontenele, A., Cavalcante, S., & Messias, G. (2026). Toward Reliable Diabetic Retinopathy Screening. Sensors, 26(14), 4515. https://doi.org/10.3390/s26144515
Vancouver
Bragança H, Caliari ÍP, Vital WL, Fontenele A, Cavalcante S, Messias G. Toward Reliable Diabetic Retinopathy Screening. Sensors. 2026;26(14):4515. doi:10.3390/s26144515
BibTeX
@article{Bragana2026,
author = {Bragança, Hendrio and Caliari, Ítalo P and Vital, Wington L and Fontenele, Antonio and Cavalcante, Sergio and Messias, Glaucio},
title = {Toward Reliable Diabetic Retinopathy Screening},
journal = {Sensors},
year = {2026},
volume = {26},
number = {14},
pages = {4515},
doi = {10.3390/s26144515}
}
EUVIMED