A review of methods for automatic detection of macular edema

Diabetic Macular Edema (DME) is a sight-threating complication of diabetic retinopathy (DR) and the major cause of vision impairment in people with diabetes. Damage to the retinal vasculature causes them to leak, triggering an inflammatory response and deposition of exudative material on the retina. DME can occur at any stage of DR and diagnosis and severity classification of the disease is done by imaging of the retina. This paper presents a narrative review of the literature to identify the strengths and limitations of the different imaging modalities for automated DME detection and monitoring. Comprehensive literature search was conducted and a total of 143 relevant peer-reviewed articles published from 2000 to 2020 in the English language were selected. The authors observed the large diversity in the journals and conferences where papers on this topic have been published. We also found the obvious rapid uptake of technology by clinicians to investigate Macular edema (ME) and DME. Another observation was inroads made by deep-learning and artificial intelligence in the field. One limitation appears to be that there are not enough publicly available large datasets for different imaging modalities that are annotated and labelled for DME.