Integration of Optical Coherence Tomography Images and Real-Life Clinical Data for Deep Learning Modeling: A Unified Approach in Prognostication of Diabetic Macular Edema.

Abstract

The primary ocular effect of diabetes is diabetic retinopathy (DR), which is associated with diabetic microangiopathy. Diabetic macular edema (DME) can cause vision loss for people with DR. For this reason, deciding on the appropriate treatment and follow-up has a critical role in terms of curing the disease. Current artificial intelligence (AI) approaches focus on OCT images and may ignore clinical, laboratory, and demographic information obtained by the specialist. This study presents a novel deep learning (DL) framework for evaluating the visual outcome of the TREX anti-VEGF intravitreal injection regimen. DL models are trained to extract deep features from OCT and ILM topographic images and the obtained deep features are combined with patients' demographic, clinical, and laboratory findings to predict the direction of the treatment process. When the ResNet-18 network is used, the proposed DL framework is able to predict the prognosis status of patients with the highest accuracy.