A hybrid visual field classifier to support early glaucoma diagnosis

Autores: Aránzazu Simón Hurtado , Luis Alonso, A. Antón López
Localización: Inteligencia artificial: Revista Iberoamericana de Inteligencia Artificial, ISSN-e 1988-3064, ISSN 1137-3601, Vol. 9, Nº. 26, 2005 (Ejemplar dedicado a: Simpposio argentino en Inteligencia Artificial), págs. 9-17
Idioma: español
Enlaces
- Texto completo

Resumen
- Primary Open Angle Glaucoma is an eye disease that can, eventually, cause irreversible damage to the optic nerve. Because of this an accurate diagnosis at early stages of the disease is necessary to stop or delay its progression. Perimetry, one of the most important tests to detect glaucoma, gives a large amount of numerical data that is di_cult to analyze. A number of approaches are described in the literature to overcome this problem, some of them using arti_cial neural networks, mainly MLP with BP. In this paper, a Hybrid Visual Field Classi_er System is proposed, comprising a Self-Organizing Map (SOM) and a rule based expert system, integrating the knowledge that the SOM discovers with the expertise of the ophthalmologist. With this association, individual results of each component are improved up to a diagnostic precision of 97%.