Search

Introduction: Diabetic retinopathy is the damaging effect of diabetes on retinal blood vessels that can cause blindness when diagnosed late. Microaneurysms are early signs of the disease that their early diagnosis promotes timely treatment and prevents disease progression. Since this disease is asymptomatic and can only be detected by ophthalmologists, diabetic patients should be tested regularly. On the other hand, given the fact that the increase rate of the number of ophthalmologists is less than the growth of the diabetic population, manual diagnosis of the lesion is time consuming and costly, and thus the design of automatic detection systems is essential. Method: In this descriptive analytic study, the fundus images of the retina were subjected to preprocessing. Then, the candidate regions of microanurysms were determined using the metric and morphological operators Bottom-hat and Hit-or-Miss. In the next step, using principal component analysis, the specificity of main feature of real microanurysms diagnosis was extracted. The DiaRetDB1 database images were used to evaluate the proposed algorithm. Results: The purpose of this research was to develop an automated method for the detection of microanurysms that can help ophthalmologists in the process of diabetic retinopathy screening and diagnosing the symptoms faster, easier and at lower cost. In evaluation, the proposed method achieved a sensitivity of 87.6%, specificity of 98.7% and the precision of 85.7%. Conclusion: According to the results obtained based on evaluation parameters, the proposed method is one of the most accurate algorithms in this field.