Diabetic retinopathy (DR) is undoubtedly one of the leading causes of blindness worldwide and is the most important ocular complication of diabetes. This pathological condition can develop in both individuals with type 1 and type 2 diabetes mellitus and its likelihood of occurrence is proportional to the duration of diabetes. One of the most serious problems related to DR is that it can often develop without any symptoms, creating only mild visual problems that are underestimated by the patient, who remains completely unaware until the DR is in an advanced and irreversible stage. Therefore, the diagnosis of this retinal disease at an early stage is crucial in order to preserve the subject's sight, especially since RD is a progressive disease.
Four different phases can be identified, with the severity of the pathology gradually increasing:
-mild non-proliferative retinopathy, characterised by the presence of micro-aneurysms;
-moderate non-proliferative retinopathy, in which the permeability and functionality of the retinal blood vessels are impaired, resulting in the formation of oedematous areas;
-severe non-proliferative retinopathy, in which there is a very clear reduction in blood flow due to the collapse of several blood vessels, in turn leading to a signal that new blood vessels are required;
-proliferative diabetic retinopathy, characterised by uncontrolled production of new retinal blood vessels. In this more advanced stage, retinal detachment and loss of vision may occur.
To date, the diagnosis of DR is essentially based on colour photographs of the ocular fundus that highlight some of the clinical signs of DR, such as haemorrhages, soft and hard exudates and micro-aneurysms. However, the diagnostic techniques available today, despite being very advanced, are not always able to make an accurate distinction between the different stages of DR.
Recently, however, a study published in International Journal of Grid and Distributed Computing proposed an optimal computer model for the detection of DR. Thanks to sophisticated image processing, it is possible to obtain photographs that provide a detailed view of the clinical signs of DR. The results of this study, although preliminary, lay the foundations for new and good diagnostic perspectives that could help the medical specialist in making a faster and more accurate diagnosis and selecting the most timely therapeutic treatment.
Source
Dutta A et al. Classification of Diabetic Retinopathy Images by Using Deep Learning Models. International Journal of Grid and Distributed Computing Vol. 11, No. 1 (201
Dr. Carmelo Chines
Direttore responsabile