AUTOMATED DIAGNOSIS OF DIABETIC RETINOPATHY USING FUNDUS IMAGES

The objective of the proposed project is to Automate diagnosis of Diabetic Retinopathy using fundus Images. DR causes damage to blood vessels of retina; it is the layer which is light sensitive present at the back of eye. The worldwide rise of diabetes, and its complications, means there is an incre

Project Title

AUTOMATED DIAGNOSIS OF DIABETIC RETINOPATHY USING FUNDUS IMAGES

Project Area of Specialization

Artificial Intelligence

Project Summary

The objective of the proposed project is to Automate diagnosis of Diabetic Retinopathy using fundus Images. DR causes damage to blood vessels of retina; it is the layer which is light sensitive present at the back of eye. The worldwide rise of diabetes, and its complications, means there is an increasing need for health professionals to consider the possibility of diabetic eye disease even before the symptoms begin to show andthe symptoms are; Blurriness, Distorted vision, Difficulty in recognizing different colours, Dark or empty areas in vision, Vision loss, Fluctuating vision.

Globally, the number of people with DR will grow from 126.6 million in 2010 to 191.0 million by 2030. Diabetic retinopathy is the most frequent cause of new cases of blindness among adults aged 20–74 years. During the first two decades of disease, nearly all patients with type 1 diabetes and 60% of patients with type 2 diabetes have retinopathy. Up to 21% of patients with type 2 diabetes have retinopathy at the time of first diagnosis of diabetes, and most develop some degree of retinopathy over time. Increased glucose levels in the body effect the blood vessels mainly. Diagnosing it in early stages is very essential to avoid a diabetic patient from going completely blind. A machine learning (ML) is able to identify, localize and quantify pathological features in almost every macular and retinal disease. So, the primary aim of this study which is presented in this proposal is to train and validate a Machine learning System uses artificial intelligence and representation learning methods to process large data containing of many retinal images and extract meaningful pattern to detect referable diabetic retinopathy. Fundus of the eye photography involves photographing the rear of an eye. The images of fundus camera were processed for the diagnosis of DR manually and it used to take an extended period of time and it is prone to human error, it calls for the need of developing a system that is automated, highly accurate and fast. For achieving highly accurate results, sophisticated image processing algorithms are incorporated. To reduce the time and power consumption in computation of these algorithms, we will implement the system using Machine Learning. The benefit of machine learning in retina which will empower the ophthalmologist to provide high quality diagnosis of diabetic eye disease, in particular diabetic retinopathy so that we can reduce human labour, cost and time, and increase the accuracy.

Project Objectives

The objective of our task is to detect DR in fundus image and if DR is found then classify stage(Mild, Moderate, Severe,
Proliferative) of DR, also UI will able to display patterns of veins and symbtoms of DR in fundus image. All work will done with high accuracy and less processing time using real world dataSet.

Project Implementation Method

The automated DR system comprises of following steps: pre-processing, data labelling, training and validation and then testing on real world dataset .

• To gather dataset of fundus images available online as well as collected it from local private eye hospital.

• To label, prepare and analyse the dataset of fundus images for ML model.

• To train, validate dataset by extracting features and patterns in it then classify the stages of DR if DR found or label it as no DR.

• To increase the efficiency of system, model will be tested on real world dataset

• If accuracy of system is desirable then go further or re-train and re-validate the model.

• To design User Interface (UI) for displaying results on screen.

Benefits of the Project

• It will give faster results and can get the work done in minutes as compared to ophthalmologists.
• It will preserve vision in patients with diabetes.
• It will give more accurate results compared to manual results.

Technical Details of Final Deliverable

a user friendly interface will get fundus image of retina as input and will able to classify it as no DR image or stage(Mild, Moderate, Severe,Proliferative) of DR. Also UI will able to display patterns of veins and symbtoms of DR in fundus image. All work will done with high accuracy and less processing time using real world dataSet.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Medical

Other Industries

Core Technology

Artificial Intelligence(AI)

Other Technologies

Others

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com