Regular eye examinations to detect diabetic retinopathy at an early stage is essential, because it can be cured with early detection. Viewing and photographing the fundus of the human eye is an important diagnosis method in ophthalmology. The cost of fundus cameras is typically about 20,000–50,000 US$ [Michelson, G., Teleophthalmology in Preventive Medicine. 2015: Springer]. Provide a portable device for imaging and automatic detection of diabetic retinopathy that it costs is also low, can greatly help in the prevention of blindness in disadvantaged areas.

The proposed system includes a smartphone, +20 D lens, lens holder and software for automated detection of microaneurysms and exudates lesions. Mobile flash lights up the surface of the retina and the image taken by the camera and lens. To diagnose the lesion exudates, in pre-processing stage to green band, noise reduction and contrast enhancement applied and after removing the optic disc and vessels, the morphological reconstruction is used to obtain the candidate areas and the final stage, the characteristics of the lesion edges for a more accurate diagnosis is used. Also for microaneurysms diagnosis, in the pre-processing, noise reduction, contrast enhancement and correction of non-uniform illumination to the green band is applied and the three images is created. Then with using morphological Black-hat transform, an estimate of the MAs lesion obtained. And optic disc, Vessels and lesion exudates removed and finally using MAs properties, the candidate areas evaluated.

The dataset of ROC and ddb1_v02_01 is used to evaluate the algorithms proposed system. For lesion MAs, average sensitivity of 39%, for 9.12 FPs per image is obtained and for lesion exudate, sensitivity and specificity of 89.58 and 73.17 for the detection of patient images from normal images were obtained.

The proposed system can be used in the early detection of diabetic retinopathy in disadvantaged areas and developing countries that do not have access to an ophthalmologist and fundus imaging devices and it can also be useful in telemedicine.