IRO Journals
Journal of Innovative Image Processing

Deep Denoising with Dual-Branch Confidence-Aware Network for Diabetic Retinopathy Classification

Open Access
Published: 30 May, 2026
Pages: 724-746
  • Kokila A. , Kokila A.
    Department of Computer Science, Chikkanna Government Arts College, Vivekananda Nagar, Konganagiri, Tiruppur, India
    Department of Computer Science, Chikkanna Government Arts College, Vivekananda Nagar, Konganagiri, Tiruppur, India
  • Shankar R. , Shankar R.
    Department of Computer Science, Chikkanna Government Arts College, Vivekananda Nagar, Konganagiri, Tiruppur, India
    Department of Computer Science, Chikkanna Government Arts College, Vivekananda Nagar, Konganagiri, Tiruppur, India
  • Duraisamy S. Duraisamy S.
    Department of Computer Science, Chikkanna Government Arts College, Vivekananda Nagar, Konganagiri, Tiruppur, India
    Department of Computer Science, Chikkanna Government Arts College, Vivekananda Nagar, Konganagiri, Tiruppur, India
view PDF
view PDF

How to Cite

A., Kokila, Shankar R., and Duraisamy S. 2026. “Deep Denoising With Dual-Branch Confidence-Aware Network for Diabetic Retinopathy Classification”. Journal of Innovative Image Processing 8 (2): 724-46. https://doi.org/10.36548/jiip.2026.2.015.
ACM ACS APA ABNT Chicago Harvard IEEE MLA Turabian Vancouver AMA

Download Citation

Endnote/Zotero/Mendeley (RIS) BibTeX

Keywords

Image Denoising
Retinal Fundus Images
Diabetic Retinopathy
Deep Learning
Transformer
Dual Attention Model

Abstract

Diabetic retinopathy (DR) is a condition resulting from prolonged increased blood sugar that destroys the retinal blood vessels. Early diagnosis could prevent permanent and complete blindness. Mostly, retinal fundus images are broadly utilized for the detection and localization of intra-retinal limitations. This article offers a Hybrid Deep Denoising and Dual-Branch Confidence-Aware Classification for Accurate Diabetic Retinopathy (HDD-DBCADR) from retinal images. The main purpose of the HDD-DBCADR framework is the combination of a hybrid denoising network and confidence-aware dual-branch classification, which offers a promising direction for developing reliable and clinically applicable automated DR screening systems. The proposed model involves the design of a hybrid deep denoising network, which integrates a residual convolutional neural network with an autoencoder-based refinement module to enhance image quality. For robust feature representation, a dual-branch architecture is applied, where the local branch exploits a multi-scale convolutional neural network for deriving fine-grained spatial features, and the global branch leverages a dual attention vision transformer to model long-range dependencies and contextual data. At last, a dual-branch confidence-aware classification mechanism adaptively combines predictions from both branches, depending on their confidence scores, thereby improving classification reliability.  The simulation investigation of the HDD-DBCADR approach is performed using a benchmark DR detection dataset from the Kaggle repository. Extensive comparative results report the encouraging performance of the proposed HDD-DBCADR model against recent state-of-the-art models with models. achieving a maximum accuracy of 98.96% and SSIM of 0.9760. The inclusion of the denoising network with the deep learning-based DR classification model accomplishes enhanced diagnostic performance, making it useful for real-time clinical settings.

References

  1. Liang, Weizhe, Chee-Onn Chow, Raymond Wong Jee Keen, and Jeevan Kanesan. "Diabetic Retinopathy Classification Network with Multi-Frequency Contextual Attention Module." Medical Engineering & Physics 147, no. 3 (2026): 035008.
  2. Aftab, Shabib, and Samia Akhtar. "Diabetic Retinopathy Severity Classification Using Data Fusion and Ensemble Transfer Learning." Journal of Software Engineering and Applications 18, no. 1 (2025): 1-23.
  3. Radha, K., and Yepuganti Karuna. "Retinal Vessel Segmentation to Diagnose Diabetic Retinopathy Using Fundus Images: A Survey." International Journal of Imaging Systems and Technology 34, no. 1 (2024): e22945.
  4. Krishnadhas, Anugirba, Lal Raja Singh Ravi Singh, and RS Rimal Isaac. "D-RetinoNet: Diabetic retinopathy Stage Classification via Deep Duo-Branch S2 Feature Based Neural Network." Biomedical Signal Processing and Control 119 (2026): 109901.
  5. Devi, T. M., P. Karthikeyan, B. Muthu Kumar, and M. Manikandakumar. "Diabetic Retinopathy Detection via Deep Learning Based Dual Features Integrated Classification Model." Technology and Health Care 33, no. 2 (2025): 1066-1080.
  6. Alanazi, Sultan, Sajid Ullah Khan, Faisal M. Alotaibi, and Mohammed Alonazi. "A Novel Noise Removal and Interpretable Deep Learning Model for Diabetic Retinopathy Detection." International Journal of Imaging Systems and Technology 35, no. 6 (2025): e70245.
  7. Singh, Atulesh Pratap, and Ajay Singh. "Denoising and Diabetic Retinopathy Detection using Machine Learning and Neural Network." health 2025. 1: 2.
  8. Chilukuri, Rajitha, Praveen P, Ranjith Kumar Gatla, and Reem A. Almenweer. "Quantum Denoising Autoencoder Improves Retinal Fundus Image Quality for Early Diabetic Retinopathy Screening." Scientific Reports 16, 5970 (2026).
  9. Rajitha, Chilukuri, P. Praveen, and K. Rajchandar. "Enhanced Detection of Diabetic Retinopathy Using GAI-Net: Deep Learning Model based on GAN with Autoencoder." Ain Shams Engineering Journal 17, no. 3 (2026): 104050.
  10. Li, Yixiao, Boyu Yu, Mingwei Si, Mengyao Yang, Wenxuan Cui, Yi Zhou, Shujun Fu, Hong Wang, Xuya Liu, and Han Zhang. "Enhancing Diabetic Retinopathy Diagnosis: Automatic Segmentation of Hyperreflective Foci in OCT via Deep Learning." International Ophthalmology 45, no. 1 (2025): 79.
  11. Chen, Qiyuan. "Denoising Diffusion with Enhanced Dual-granularity Prior for Diabetic Retinopathy Grading." In 2025 19th International Conference on Complex Medical Engineering (CME), IEEE, 2025, 14-18.
  12. Shamrat, FM Javed Mehedi, Rashiduzzaman Shakil, Bonna Akter, Md Zunayed Ahmed, Kawsar Ahmed, Francis M. Bui, and Mohammad Ali Moni. "An Advanced Deep Neural Network for Fundus Image Analysis and Enhancing Diabetic Retinopathy Detection." Healthcare Analytics 5 (2024): 100303.
  13. Kirubakaran, M., and V. Vijayarajan. "WaveMem-SHAPNet: A Transparent Deep Learning Approach to Early Diagnosis of Diabetic Retinopathy." SN Computer Science 7, no. 1 (2026): 73.
  14. Yu, Jingning. "Based on Gaussian Filter to Improve the Effect of the Images in Gaussian Noise and Pepper Noise." In Journal of Physics: Conference Series, vol. 2580, no. 1, IOP Publishing, 2023, 012062.
  15. Cheng, Jilan, Guoli Long, Zeyu Zhang, Zhenjia Qi, Hanyu Wang, Libin Lu, Shuihua Wang, Yudong Zhang, and Jin Hong. "WaveNet-SF: A Hybrid Network for Retinal Disease Detection Based on Wavelet Transform in Spatial-Frequency Domain." arXiv preprint arXiv:2501.11854 (2025).
  16. Yao, Qihai, Yong Wang, and Yixin Yang. "Underwater Acoustic Target Recognition Based on Data Augmentation and Residual CNN." Electronics 12, no. 5 (2023): 1206.
  17. Li, Fangfang, Qizhou Wu, Bei Jia, and Zhicheng Yang. "Speckle Noise Removal in OCT Images via Wavelet Transform and DnCNN." Applied Sciences 15, no. 12 (2025): 6557.
  18. Rassam, Murad A. "Autoencoder-Based Neural Network Model for Anomaly Detection in Wireless Body Area Networks." IoT 5, no. 4 (2024): 852-870.
  19. Huang, Tengda, Sheng Fu, Haonan Feng, and Jiafeng Kuang. "Bearing Fault Diagnosis Based on Shallow Multi-Scale Convolutional Neural Network with Attention." Energies 12, no. 20 (2019): 3937.
  20. Ding, Mingyu, Bin Xiao, Noel Codella, Ping Luo, Jingdong Wang, and Lu Yuan. "Davit: Dual Attention Vision Transformers." In European conference on computer vision, Cham: Springer Nature Switzerland, 2022, 74-92.
  21. https://www.kaggle.com/competitions/diabetic-retinopathy-detection/data
  22. Jebur, Rusul Sabah, Mohd Hazli Bin Mohamed Zabil, Dalal Abdulmohsin Hammood, Lim Kok Cheng, and Ali Al-Naji. "Image Denoising Using Hybrid Deep Learning Approach and Self-Improved Orca Predation Algorithm." Technologies 11, no. 4 (2023): 111.
  23. Tiantian, Wang, Zhihua Hu, and Yurong Guan. "An Efficient Lightweight Network for Image Denoising Using Progressive Residual and Convolutional Attention Feature Fusion." Scientific Reports 14, no. 1 (2024): 9554.
  24. Karthik, S. A., M. N. Geetha, K. Prabhavathi, Dhananjaya Shashank, K. P. Suhaas, and M. Narender. "Early Detection and Severity Classification of Diabetic Retinopathy Using Convolutional Neural Networks." SN Computer Science 6, no. 7 (2025): 819.
  25. Muthusamy, Dharmalingam, and Parimala Palani. "Deep Learning Model Using Classification for Diabetic Retinopathy Detection: An Overview." Artificial Intelligence Review 57, no. 7 (2024): 185.