retinar: AI for diabetic retinopathy screening
Diabetic retinopathy is the leading cause of preventable blindness in working age populations. Color fundus photography is currently used for telemedicine campaigns in which diabetic patients are remotely screened once an year through this imaging modality. The images are acquired by trained technicians, who transfer them to a study center in which retina experts analyzed them to determine the diagnostic. When the amount of patients increases, so does the number of images to analyze. As a consequence, these platforms suffer from scalability issues, which seriously affect their accuracy and efficiency.
In this project we aim to develop an AI-assisted platform to allow for efficient screening of diabetic retinopathy. Our system will use deep learning techniques to ensure capturing high quality photographs in the acquisitions centers and to provide early diagnostics to the patients. It will also feature lesion detections modules that will be used by the ophthalmologists to produce more accurate reports.
PI: José Ignacio Orlando, PhD
- Mercedes Leguía, MD - Hospital de Alta Complejidad El Cruce Dr. Néstor Carlos Kirchner, Florencio Varela, PBA, Argentina.
- Alejandro Koch, MD - Hospital de Alta Complejidad El Cruce Dr. Néstor Carlos Kirchner, Florencio Varela, PBA, Argentina.
- Ignacio Larrabide, PhD - CONICET / PLADEMA-UNICEN.
- Ezequiel Rosendi, MD - Hospital de Alta Complejidad El Cruce Dr. Néstor Carlos Kirchner, Florencio Varela, PBA, Argentina.
- Tomás Castilla, Undergraduate student - Facultad de Ciencias Exactas, UNICEN.
- Leandro Rocamora, MSc - PLADEMA-UNICEN.
- JOVIN 2020/2021 Grant (“Towards a smart platform for remote diabetic retinopathy screening: quality control in fundus photographs using autoencoders”). Convocatoria Jóvenes Investigadores JOVIN 2020/2021 (Programa de Fortalecimiento a la Ciencia y la Tecnología en Universidades Nacionales, Secretaría de Ciencia, Arte y Tecnología, UNICEN).
- INI 2020 Scholarship (“Deep learning algorithms for computer assisted diagnostic of diabetic retino-pathy from fundus photographs”). Convocatoria Beca INI de Ingreso a la Investigación (Programa de Fortalecimiento a la Ciencia y la Tecnología en las Universidades Nacionales). Becario: Tomás Castilla.
This project was awarded with the Voted Best by the Audience prize at Concurso Prendete (Tandil, Argentina). As such, it was assigned with U$D 10.000 in AWS credits.
- An ensemble deep learning based approach for red lesion detection in fundus images
- Proliferative diabetic retinopathy characterization based on fractal features: Evaluation on a publicly available dataset
- Aprendizaje automático para asistencia al diagnóstico de enfermedades visuales basado en imágenes de fondo de ojo (Machine learning for ophthalmic screening and diagnostics from fundus images)