Abstract

Diabetic retinopathy is a leading cause of vision impairment globally, necessitating timely and accurate diagnosis to prevent irreversible damage. This paper proposes a novel hybrid deep learning framework that combines local and global feature representations for robust DR classification from retinal fundus images. Local features are extracted using a convolutional neural network branch that captures fine-grained pathological patterns such as microaneurysms and hemorrhages. Simultaneously, global contextual features are learned through a Vision Transformer, which models long-range dependencies across the retinal image. The extracted features from both branches are fused and passed through a series of dense layers for initial classification. To further enhance generalization and interpretability, features from the Global Average Pooling layer are used to train a Random Forest classifier. The proposed methodology is evaluated on a benchmark DR dataset with five severity classes. Extensive experiments and ablation studies demonstrate the effectiveness of our architecture in capturing both fine-grained and holistic features, leading to improved classification performance. Our results suggest that the fusion of local and global features, combined with ensemble post-classification, can provide a robust and scalable solution for automated DR diagnosis.

References

• Anas Bilal, Azhar Imran, Talha Imtiaz Baig, Xiaowen Liu, Haixia Long, Abdulkareem Alzahrani, and Muhammad Shafiq. Improved support vector machine based on cnn-svd for vision-threatening diabetic retinopathy detection and classification. Plos one, 19(1):e0295951, 2024.
• Milena Cioana, Jiawen Deng, Ajantha Nadarajah, Maggie Hou, Yuan Qiu, Sondra Song Jie Chen, Angelica Rivas, Parm Pal Toor, Laura Banfield, Lehana Thabane, et al. Global prevalence of diabetic retinopathy in pediatric type 2 diabetes: a systematic review and meta-analysis. JAMA network open, 6(3):e231887–e231887, 2023.
• Ling Dai, Bin Sheng, Tingli Chen, Qiang Wu, Ruhan Liu, Chun Cai, LiangWu, Dawei Yang, Haslina Hamzah, Yuexing Liu, et al. A deep learning system for predicting time to progression of diabetic retinopathy. Nature Medicine, 30(2):584– 594, 2024.
• Sebastian Dinesen, Lonny Stokholm, Yousif Subhi, Tunde Peto, Thiusius Rajeeth Savarimuthu, Nis Andersen, Jens Andresen, Toke Bek, Javad Hajari, Steffen Heegaard, et al. Fiveyear incidence of proliferative diabetic retinopathy and associated risk factors in a nationwide cohort of 201 945 danish patients with diabetes. Ophthalmology science, 3(3):100291, 2023.
• Xuhong Hou, Limin Wang, Dalong Zhu, Lixin Guo, Jianping Weng, Mei Zhang, Zhiguang Zhou, Dajin Zou, Qiuhe Ji, Xiaohui Guo, et al. Prevalence of diabetic retinopathy and vision-threatening diabetic retinopathy in adults with diabetes in china. Nature Communications, 14(1):4296, 2023.
• Cris Martin P Jacoba, Duy Doan, Recivall P Salongcay, Lizzie Anne C Aquino, Joseph Paolo Y Silva, Claude Michael G Salva, Dean Zhang, Glenn P Alog, Kexin Zhang, Kaye Lani Rea B Locaylocay, et al. Performance of automated machine learning for diabetic retinopathy image classification from multi-field handheld retinal images. Ophthalmology Retina, 7(8):703–712, 2023.
• Edward Korot, Mariana Batista Gonc¸alves, Josef Huemer, Sara Beqiri, Hagar Khalid, Madeline Kelly, Mark Chia, Emily Mathijs, Robbert Struyven, Magdy Moussa, et al. Cliniciandriven ai: code-free self-training on public data for diabetic retinopathy referral. JAMA ophthalmology, 141(11):1029– 1036, 2023.
• Martina Kropp, Olga Golubnitschaja, Alena Mazurakova, Lenka Koklesova, Nafiseh Sargheini, Trong-Tin Kevin Steve Vo, Eline de Clerck, Jiri Polivka Jr, Pavel Potuznik, Jiri Polivka, et al. Diabetic retinopathy as the leading cause of blindness and early predictor of cascading complications—risks and mitigation. Epma Journal, 14(1):21–42, 2023.
• Chun-Ling Lin and Kun-Chi Wu. Development of revised resnet-50 for diabetic retinopathy detection. BMC bioinformatics, 24(1):157, 2023.
• Cheena Mohanty, Sakuntala Mahapatra, Biswaranjan Acharya, Fotis Kokkoras, Vassilis C Gerogiannis, Ioannis Karamitsos, and Andreas Kanavos. Using deep learning architectures for detection and classification of diabetic retinopathy. Sensors, 23(12):5726, 2023.
• AM Mutawa, Shahad Alnajdi, and Sai Sruthi. Transfer learning for diabetic retinopathy detection: A study of dataset combination and model performance. Applied Sciences, 13(9):5685, 2023.
• Dharmalingam Muthusamy and Parimala Palani. Deep learning model using classification for diabetic retinopathy detection: an overview. Artificial Intelligence Review, 57(7):185, 2024.
• Md Nahiduzzaman, Md Robiul Islam, Md Omaer Faruq Goni, Md Shamim Anower, Mominul Ahsan, Julfikar Haider, and Marcin Kowalski. Diabetic retinopathy identification using parallel convolutional neural network based feature extractor and elm classifier. Expert Systems with Applications, 217:119557, 2023.
• Sovit Ranjan Rath. Diabetic retinopathy 224x224 gaussian filtered dataset. https: //www.kaggle.com/datasets/sovitrath/ diabetic-retinopathy-224x224-gaussian-filtered, 2022. Accessed: July 21, 2025.
• Hossein Shakibania, Sina Raoufi, Behnam Pourafkham, Hassan Khotanlou, and Muharram Mansoorizadeh. Dual branch deep learning network for detection and stage grading of diabetic retinopathy. Biomedical Signal Processing and Control, 93:106168, 2024.
• Tien-En Tan and Tien YinWong. Diabetic retinopathy: Looking forward to 2030. Frontiers in Endocrinology, 13:1077669, 2023.
• Alebachew Ferede Zegeye, Yemataw Zewdu Temachu, and Chilot Kassa Mekonnen. Prevalence and factors associated with diabetes retinopathy among type 2 diabetic patients at northwest amhara comprehensive specialized hospitals, northwest ethiopia 2021. BMC ophthalmology, 23(1):9, 2023.
• Jing Zhou and Bo Chen. Retinal cell damage in diabetic retinopathy. Cells, 12(9):1342, 2023.