| Title |
CNN based Diabetic Retinopathy Feature Extraction and Grade Classification |
| Authors |
정영훈(Younghoon Jung) ; 조경진(Kyungjin Cho) ; 김대원(Daewon Kim) |
| DOI |
https://doi.org/10.5573/ieie.2019.56.11.61 |
| Keywords |
Faster R-CNN ; Random Forest ; Classification ; |
| Abstract |
Non-proliferative diabetic retinopathy is a representative complication of diabetic patients and is known to be a major cause of impaired vision and blindness. There has been ongoing research on automatic detection of diabetic retinopathy; however, there is also a growing need for research on an automatic severity classification system. This study proposes an automatic detection system for pathological symptoms of diabetic retinopathy such as microaneurysm, retinal hemorrhage, and hard exudate by applying the Faster R-CNN technique. An automatic severity classification system based on the features of pathological symptoms of diabetic retinopathy was devised by training and testing a random forest classifier based on the data obtained through preprocessing, such as histogram smoothing of the detected features. The proposed system enables accurate judgment using objective data and indices while avoiding the subjective interpretation of testers and improving the efficiency of medical image analysis. An experiment of classifying 103 test fundus images with the proposed classification system showed 98% accuracy. The proposed automatic severity classification is expected to show a higher degree of accuracy if a greater amount of meaningful data can be collected in the future. |