Ophthalmic Res2023;66:1278–1285
DOI: 10.1159/000534251

Received: March 22, 2023
Accepted: September 18, 2023
Published online: September 30, 2023



Deep Learning Application to Detect Glaucoma with a Mixed Training Approach: Public Database and Expert-Labeled Glaucoma Population


Florencia Cellini; Deborah Caamaño; Belen Carrasco; José R. Juberías; Carolina Ossa; Ramón Bringas; Francisco de la Fuente; Pablo Franco; David Coronado; Jose Carlos Pastor


Introduction: Artificial intelligence has real potential for early identification of ocular diseases such as glaucoma. An important challenge is the requirement for large databases properly selected, which are not easily obtained. We used a relatively original strategy: a glaucoma recognition algorithm trained with fundus images from public databases and then tested and retrained with a carefully selected patient database.

Methods: The study’s supervised deep learning method was an adapted version of the ResNet-50 architecture previously trained from 10,658 optic head images (glaucomatous or non-glaucomatous) from seven public databases. A total of 1,158 new images labeled by experts from 616 patients were added. The images were categorized after clinical examination including visual fields in 304 (26%) control images or those with ocular hypertension and 347 (30%) images with early, 290 (25%) with moderate, and 217 (19%) with advanced glaucoma. The initial algorithm was tested using 30% of the selected glaucoma database and then re-trained with 70% of this database and tested again.

Results: The results in the initial sample showed an area under the curve (AUC) of 76% for all images, and 66% for early, 82% for moderate, and 84% for advanced glaucoma. After retraining the algorithm, the respective AUC results were 82%, 72%, 89%, and 91%.

Conclusion: Using combined data from public databases and data selected and labeled by experts facilitated improvement of the system’s precision and identified interesting possibilities for obtaining tools for automatic screening of glaucomatous eyes more affordably.


Nuestra última publicación es un primer trabajo hecho con la empresa Transmural Biotech, para intentar mejorar un algoritmo de Inteligencia Artificial (IA) que obtenga buenos valores de sensibilidad y especificidad como para poder ser empleado en programas de cribado automático de glaucoma. Es una de las primeras aplicaciones de la IA dentro del centro de lectura del IOBA, pero estamos seguros que en breve se abrirá una intensa linea de colaboracion con esta empresa y con otros grupos de la Universidad.

Our last publication is a first work done with the company Transmural Biotech, to try to improve an Artificial Intelligence (AI) algorithm that obtains good values of sensitivity and specificity to be used in automatic glaucoma screening programs. This is one of the first applications of AI within the IOBA reading center, but we are sure that soon an intense line of collaboration will be opened with this company and with other groups of the University.


Acceda al documento >