Enfermedades cardiovascularesscreening de la población general
- X. Múgica 2
- S. Gómez Sánchez 3
- R. Cilla 2
- S. Banderas García 3
- R. Muñoz 4
- X. Gràcia Aloy 4
- MI. Martínez Segura 1
- A. Sánchez-Fortún Sánchez 3
- MÀ. Pouplana Sardà 3
- P. Campos Figueroa 3
- R. Bouchikh El Jarroudi 3
- A. Garcés 3
- A. Sabala Llopart 3
- F. Ortuño 4
- I. Besada 2
- JA. de Frutos 2
- O. Estrada Cuxart 5
- E. Isusquiza 2
- S. Ruiz Bilbao 3
-
1
Institut d'Investigació en Ciències de la Salut Germans Trias i Pujol
info
Institut d'Investigació en Ciències de la Salut Germans Trias i Pujol
Barcelona, España
- 2 ULMA Medical Technologies. Departamento de Desarrollo. Oñati, Gipuzkoa.
- 3 Servicio de Oftalmología. Hospital Universitari Germans Trias i Pujol (HUGTiP). Badalona. Barcelona.
- 4 Dirección de Sistemas de Información. HUGTiP . Badalona. Barcelona.
- 5 Direcció d’Estratègia Assistencial i d’Innovació. HUGTiP. Badalona. Barcelona
- Romero Aroca, Pere
- Zapata Victori, Miguel Ángel
- Zarranz Ventura, Javier
ISSN: 1133-7737
Datum der Publikation: 2023
Titel der Ausgabe: INTELIGENCIA ARTIFICIAL Y OFTALMOLOGÍA: ESTADO ACTUAL EN CATALUÑA
Ausgabe: 31
Nummer: 4
Art: Artikel
Andere Publikationen in: Annals d'oftalmologia: òrgan de les Societats d'Oftalmologia de Catalunya, Valencia i Balears
Zusammenfassung
Macular pathologies affect the fovea, the most important area of the retina for visual function. New retinal imaging techniques, such as optical coherence tomography (OCT) and retinography, allow direct, non-invasive, high-resolution visualization of the retinal layers. The aim of the artificial intelligence and macular optical coherence tomography (OCT) project AI4Ret (PI-20-113) is the creation of an artificial intelligence (AI) software based on deep learning (DL) to identify biomarkers for cardiovascular risk prediction from macular OCT images. AI4Ret has curated a dataset of macula-centered OCT images from the overall population attended by the ophthalmology department of the Hospital Universitario Germans Trias i Pujol (HUGTiP). Then, a deep learning model has been trained to identify specific biomarkers of arterial hypertension (AHT), type 2 diabetes mellitus (DM2), and dyslipidemia in OCT images of patients without macular pathology. These factors could be a predictor for an additional risk from suffering a cardiovascular event. Currently, the model achieves a sensitivity and specificity of 84% and 69% for AHT, 86% and 67% for DM2, and 74% and 78% for dyslipidemia. Thus, patients without macular pathology present retinal alterations that the AI model detects. Their presence could mean a higher risk of metabolic alterations and a higher probability of suffering a cardiovascular event.