Automatic Detection of Diabetes Retinopathy based on Digital Retinal Images

Automatic Detection of Diabetes Retinopathy based on Digital Retinal Images

Faculty of Information and Communication Technology, Mahidol University

Project Title:

Detection of Diabetes Retinopathy using Image Processing

Research Title:

Automatic Detection of Diabetes Retinopathy based on Digital Retinal Images

Researcher:

Asst.Prof. Dr.Worapan Kusakunniran

This research project mainly aims to develop novel approaches and computer programs to detect and classify stages of the diabetic retinopathy in retinal images, in an automatic way. This could help filtering patients who may be in the early stage who have a higher chance to be completely treated, or in the urgent needs of the treatments. It could also help giving the service of diabetic retinopathy check-up for people who stay a distance from the medical service. This research project developed the new image processing based methods to segment key visual signals of the diabetic retinopathy such as microaneurysms, hard exudates, and abnormal blood vessels. Such signals were further used in the classification learning processes for classifying the stage of the disease. The deep learning was also attempted to apply on raw retinal images directly for the stage classification. The developed approaches and computer programs were validated with the retinal images in both perspectives of segmentations and classifications. The experimental results illustrated the promising performances of the proposed method with over 80% accuracy of the stage classification in average.

Publishing:
• W. Kusakunniran, Q. Wu, P. Ritthipravat, J. Zhang, Hard Exudates Segmentation based on Learned Initial Seeds and Iterative Graph Cut, Computer Methods and Programs in Biomedicine (CMPB), 158: 173-183, May 2018, DOI: 10.1016/j.cmpb.2018.02.011
• W. Kusakunniran, S. Kanchanapreechakorn, K. Thongkanchorn, Instance-based Learning for Blood Vessel Segmentation in Retinal Image, pages 111 – 118, Thailand, July 2019, International Conference on Computing and Information Technology (IC2IT), Advances in Intelligent Systems and Computing, Springer Nature Switzerland (IC2IT 2019, AISC 936), DOI: 10.1007/978-3-030-19861-9_11
• R. Kasantikul, W. Kusakunniran, Improving Supervised Microaneurysm Segmentation using Autoencoder-Regularized Neural Network, pages 553 – 559, Australia, December 2018, Digital Image Computing: Techniques and Applications (DICTA)
• W. Kusakunniran, Q. Wu, P. Ritthipravat, J. Zhang, Three-Stages Hard Exudates Segmentation in Retinal Images, pages 1 – 6, Thailand, October 2017, International Conference on Information Technology and Electrical Engineering (ICITEE)

Award Grant related to the Project:
• Research Grant for New Scholar, supported by Thailand Research Fund

Key Contact Person:
Asst.Prof. Dr.Worapan Kusakunniran
Faculty of Information and Communication Technology, Mahidol University