Automated Diagnosis of Normal/abnormal Fundus Image using Convolutional Neural Network


Healthcare - Medical Devices


Fundus photography is one of the imaging techniques commonly adopted in eye clinics for the diagnosis of main eye diseases that ultimately leads to blindness, Age-Related Macular Degeneration (AMD), Diabetic Retinopathy (DR) and Glaucoma. However, manual assessment of fundus images is time-consuming and subjective. Frequent eye screening of the elderly put pressures on an already resource-tight medical system. The cost is increasing and there seems to be no end in sight.

We have trained and tested a convolutional neural network that automatically classifies images of age-related macular degeneration (AMD), diabetic retinopathy (DR) and glaucoma as abnormal and images of normal subjects as normal.


This is a Convolutional Neural Network (CNN) model that does not require feature extraction and selection technique during analysis of fundus images. In this technology, a ten layer deep CNN model that uses lesser number of parameters for analysis is used and trained using many image sources with wide variations in image size, image resolution and image quality. We use deep learning to perform the diagnosis and have tested our solutions on a set of 1492 images. Our solution can classify images of DR, AMD and glaucoma as abnormal.


Automated fundus screening in eye clinics or polyclinic and in community centres.

Market Trends & Opportunities

The deep learning market for healthcare is categorized into various applications, such as patient data & risk analysis, medical imaging & diagnostics, precision medicine, lifestyle management & monitoring, drug discovery, inpatient care & hospital management, virtual assistant, wearables, and research. The market for the medical imaging & diagnostics segment is expected to grow at 62.4% from 2015 to 2013. The growth of the deep learning market is attributed to the rise in the use of image recognition-based techniques for the analysis of medical images, such as X-rays, MRI, and tomography, to diagnose diseases. Data mining application is another key factor in the deployment of deep learning solutions. The healthcare industry today generates a large volume of complex data about patients, hospital resources, disease diagnosis, electronic patient records, and medical devices, among others, which is difficult to analyze using traditional methods. Data mining provides tools and techniques that can be applied to these large datasets to extract useful information, which enables decision-making and cost saving.


  • Reduce clinician’s workload
  • Reduce waiting time for people who want a quick diagnose of their eye condition
Potent Anti-bacterial Plant Oil for Skin Health and Ailment
Growth Factor-free Proliferation & Differentiation of Stem Cell
Plant-Harvested Biopolymer for Postsurgical Adhesion Barrier Gel and Film Application
Supercritical CO2 Platform to Sterilize Human and Mammalian Tissues and Medical Equipment
Mobile AI Elderly Fall Risk Assessment
Robotic Capsule Endoscope for GI Tract Inspection
Non-invasive First-void Urine Device for Rapid Detection of Infectious Diseases and Cancer
Printable Low-Powered and Cost Efficient Cold Plasma Technology
Magnetic Navigation Micro-Robots for Vascular Disease
Enabling Community Ophthalmology with the Smartphone Slit Lamp