A novel diabetic retinopathy classification scheme based on compact bilinear pooling CNN and gradient boosted decision tree

作者：Lian Chun-Yan; Liang Yi-Xiong^*; Kang Rui; Mao Yuan; Xiang Yao; Gao Ling

来源：Journal of Information Hiding and Multimedia Signal Processing, 2018, 9(4): 821-829.

• Boosted decision trees
• Classification scheme
• Compact bilinear pooling
• Convolutional neural network
• Convolutional Neural Networks (CNN)
• Decision tree classifiers
• Diabetic retinopathy
• Tree-based approach

Diabetic retinopathy (DR) is one of the leading causes of blindness, but the classification of DR requires experienced ophthalmologist to distinguish the presence of various small features, which is time-consuming and difficult. Therefore, automated DR classification is essential for medical treatment. In this paper, a novel scheme is proposed for automated DR classification, in which a compact bilinear pooling Convolutional Neural Network (CNN) is applied to extract DR features and a Gradient Boosted Decision Decision Tree classifier is trained based on these extracted features to classify DR. Our results on the EyePACS dataset demonstrate the proposed scheme which combines deep learning and tree based approaches achieves a superior performance for automated DR classification with a Kappa score of 0.73, a average F1-score of 0.79 and a micro-average AUC of 0.95.

• 出版日期2018
• 单位中南大学