News

A research paper titled “Detection of Diabetic Retinopathy (DR) severity from fundus photographs: An Ensemble Approach using weighted average”, has been published by Dr Mahesh Kumar Morampudi, Assistant Professor of Computer Science and Engineering, in the Arabian Journal for Science and Engineering.

Diabetic Retinopathy is a common diabetic disease that affects the retina and can result in blindness if not treated initially. Deep learning (DL) based models are proposed to detect the blood abnormalities in the retinal tissue due to diabetes mellitus obtained from fundus camera. The drawback of these models is the lack of performance. To address this, we propose to automate the process of detection of the severity of Diabetic Retinopathy (DR) using ensembles of pre-trained models, thus exploring the power of transfer learning in the field of automated diagnosis. Deep learning models perform well when the model is trained on a large amount of data. In this regard, we also put forth data augmentation and preprocessing techniques to generate synthetic images and to improve image quality. Extensive experimental results on publicly available databases illustrate that the proposed ensemble model achieves fair accuracy when compared to existing models. Thus, the proposed model shows good scope for deployment in real-time diagnosis.

Every year multiple people are diagnosed with diabetes. Diabetes is a chronic disease that affects several organs of the human body namely the eyes, kidneys, heart etc. Diabetic Retinopathy (DR) is a situation induced by diabetes in which severe loss happens to the retinal blood vessels that can ultimately lead to blindness. Regular diabetic retinopathy screening is hence needed for detecting it in advance. In the present situation, a trained clinician or an ophthalmologist is required to identify diabetic retinopathy (DR) by the existence of lesions related to the vascular abnormalities induced by the disease. The ophthalmologist needs to evaluate and examine digital colour fundus images of the retina. So, it is a very tedious and sluggish process. He/She needs more time to diagnose DR. The diagnosis of the disease by any manual means seems to be tiresome and usually results in errors. To overcome this limitation, we propose a model to automate the process of detection of the severity of DR using ensembles of pre-trained models, thus exploring the power of transfer learning in the field of automated diagnosis.

The research group believes that the study helps ophthalmologists to identify diabetic retinopathy at its early stage accurately, as a result, the chance of losing the vision due to diabetic retinopathy can be reduced. The work is done in collaboration with Dr Mulagala Sandhya, Assistant professor, NIT-Warangal. In the future, Dr Mahesh Kumar plans to work on a project related to Privacy-preserving Biometric Authentication.