Net Enabled Retinal Image Analysis and i-telemedicine system

About Project and Funding:

National ICT R&D Fund has provided Rs 13.7 Millions to convert this project into development stage aiming to make it a commercial product. This product when development will increase the software based diagnostic capability of the hospitals in the area of Eye related diseases. This product will enable the ophthalmologists to inspect more patient in less time as compared to existing capability of handling such patients. This software based tool also includes the telemedicine capability to connect the patients at rural areas with the doctors in urban locations. This project is developed with joint collaboration with Armed Forces Institute of ophthalmology (AFIO) at Rawalpindi and Shifa International Hospital (

Project Summary & Scope:

Diabetic retinopathy (DR) is an eye disease caused by the increase of insulin in blood and eventually leads to complete vision loss. In our country, the number of patients with diabetes is very high and they also have lack of knowledge about the effect of diabetes on their retina and vision. There is a national need of awareness about DR and regular monitoring of diabetes patients to save their vision. DR is a progressive disease and needs regular monitoring and screening which is costly and also requires much time of ophthalmologists. The World Health Organization (WHO) has ranked Pakistan as the country having seventh highest number of diabetics in the world. Pakistan has around 7.1 million diabetic patients and almost 12000 patients suffer from severe DR every year.
The ratio of ophthalmologists to diabetes patients is very low in our country which prevents the mass screening of DR and eventually leads to a number of sudden vision loss cases especially in rural areas. There is a need of a telemedicine based self-diagnosis system for detection and grading of human retina for possible DR and referring the severe cases to ophthalmologists. High resolution digital retinal images are used and analyzed in such systems for reliable screening of DR. The self diagnosis system along with high resolution fundus cameras and a facility of telecommunication can be placed in local community and medical centers with trained medical staff for regular screening of DR. They will refer the cases that are graded as possible DR to ophthalmologists sitting in eye care center at main city hospitals using different communication protocols.
The proposed system is a telemedicine system for telescreening of DR with capabilities of self diagnosis, real time access to data from rural areas, assignment of ophthalmologist for expert opinion and management of a national database for regular monitoring. The main aim of project is to provide a low cost and easy access solution to patients with diabetes for regular monitoring of the retina to avoid the risk of sudden vision loss. In this project, we propose a self diagnosis system for DR with capabilities of automated screening and diagnosis of abnormalities present in human retina, a universal communication node to provide real time availability of data and intelligent server to assign expert and schedule patient’s data and help in early detection of DR to save patient’s vision. The system will take real time high quality retinal images from fundus camera and will grade it according to type and number of abnormalities present in it using state of the art image processing and pattern recognition techniques. The images with possible DR will be sent to ophthalmologists for further diagnosis and the data for each patient will be saved for regular monitoring and to check the progress of disease.
In order to provide real time diagnosis, the system will be linked with a universal communication device to transfer the patient’s data to ophthalmologists in main central hospital. The system would provide a real time retinal acquisition and analysis facility from remote areas with universal connectivity through dialup, Ethernet, GPRS, Radio and Satellite and connect it to the central server located in the eye care center of a main city hospital. At this stage, Govt. hospitals don’t have required infra structure for telemedicine but some of private hospitals have such kind of infra structure such as Shifa International hospital in form of blood collection units. So we can deploy this system there. The proposed system will pay off in terms of providing a low cost and easy to access system to remote area patients and reducing the load of outdoor patients in hospitals and cost of travelling.
It is pertinent to mention here that we have only developed algorithms in Matlab for a subset of DR related detection. These algorithms are only tested on test images from the internet. The algorithm extension to complete DR and its working in real environment with a real fundus camera is part of this project. This requires a major research work of algorithm development and its interfacing with a real camera. We also need to develop a telemedicine system for mass screening of patients in rural and far areas in Pakistan. We have won national and international awards such as P@SHA 2012 and APICTA 2012 for the idea and the work on algorithms we have already completed. The acceptance and appreciation of our idea from national and international ICT related community have encouraged us to extend the work to a fully featured system for DR and dedicated Telescreening. We are closely working with Shifa International Hospital and AFIO (Armed Forces Institute of Ophthalmology) for guidance and user feedback.

Project Deliverable & Status:

Quarter No. Time Deliverable Status
1 July 13 – Sept13
  1.   Algorithms developed in MATLAB and tested bench-mark databases* for preprocessing and main component extraction
  2.   Basic layout of application software
  3.   Basic forms for database
  4.   A software containing basic layout for communication interfaces
2 Oct 13 – Dec 13
  1.   Algorithms developed in high level language for preprocessing and main component extraction
  2.   Algorithms developed in MATLAB for candidate lesion extraction and feature set formulation
  3.   An interface for data and retinal image transmission via communication device
3 Jan 14 – Mar 14
  1.   Algorithms developed in high level language for candidate lesion extraction and feature set formulation
  2.   Algorithms developed in MATLAB for classification and their results for DR classification tested on Benchmark databases
  3.   Communication software along with its interface to database and SAN
4 Apr 14 – Jun 14
  1.   Algorithms developed in high level language for DR classification
  2.   Complete software for DR grading
  3.   SQL based database
  4.   Complete communication software
5 July 14 – Sept 14
  1.   Evaluation results for actual patients
  2.   Database and patient management system
  3.   A complete DR detection, grading and screening software as proposed
  4.   International publications
6 Oct 14
  1.   Prototype System comprised of three clients and one server working at Shifa blood collection centers

Team Members:

For details of team members, CLICK HERE