Aprimorando a segmentação do edema macular diabético: uma abordagem computacional em imagens de OCT

Abstract

Diabetic Macular Edema (DME) is a complication of diabetic retinopathy that affects the macula, the central region of the retina, and is one of the main causes of irreversible blindness in diabetic patients, which according to the World Health Organization has increased from 200 million people with diabetes in 1990 to 830 million in 2022. In this context, this work proposed the development of a computational tool to support the detection of DME in Optical Coherence Tomography (OCT) images, using basic digital image processing (DIP) techniques. 484 images from a public database, obtained between July 2013 and March 2017 in five international institutions, were used to test the algorithm. The proposed strategy achieved an accuracy of 86%, generating overlapping masks that evidence the presence of edema and demonstrating performance comparable to the methods described in the literature. Thus, the developed tool stands out as a low-cost and highly interpretable solution, with the potential to improve diagnostic accuracy and contribute to the prevention of visual loss in diabetic patients.