Automatic quality assessment and segmentation of diabetic retinopathy images

Diabetes is considered to be one of the most dangerous genetic disorders. It could cause the loss of sight in the case of the diabetic retinopathy. Several medical technologies have been developed and improved to cure the diabetes and overcome the lacks of human experts. This can be done by replacing the manual process using human labors with the automatic diagnosis. In order to achieve this purpose, it requires two main steps which will be focused in this paper. They are the quality assessment and the segmentation of diabetic retinopathy images. In the image quality assessment, four features (namely color, contrast, focus, and illumination) have been investigated. As a result, the contrast histogram in the Principal Component Analysis (PCA) space is used. In the image segmentation, the histogram equalization is used in the pre-processing. Then, the image segmentation based on the iterative selection and the grabcut algorithm is applied. The experimental results demonstrate that the proposed method can achieve very promising performance.