A Machine Learning Based Platform For Retinal Image Analysis

? We are introducing a smart way to find out the disease (Glaucoma) occurs in the eye vessels and the optic disc by writing a machine learning algorithm. ? We require the system which can provide exact differentiation in the variation of normal eye and the affected eye. This work is to desi

Project Title

A Machine Learning Based Platform For Retinal Image Analysis

Project Area of Specialization

Artificial Intelligence

Project Summary

• We are introducing a smart way to find out the disease (Glaucoma) occurs in the eye vessels and the optic disc by writing a machine learning algorithm.
• We require the system which can provide exact differentiation in the variation of normal eye and the affected eye. This work is to design a simple, easy to detect, using microcontroller-based circuit.
• We need publically available retinal fundus image datasets (DRIVE, STARE, DRONSDB, and MESSIDOR) to learn machine and for testing. For learning purposes in which DRIVE and STARE would be used for vessel segmentation and other two will be used for optic disc segmentation.
• System will be integrated with Raspberry Pi and show the results on the screen attached with Raspberry Pi.
 

Project Objectives

To help the doctors get consistent results of medical images in less amount of time with minimum effort.
Time Consumption Manual segmentation is performed by trained experts that is very hard and time-consuming task as it includes different types of manual analysis
techniques. By using this automated detection several precious hours can be saved.
Work Load Before the automation, doctors have to do a lot of work to understand the retinal image as these problems are complex and critical but now automation will reduce their burden completely.
Consistent Results It presents results more coherent to the gold standard than the second human annotator .
Better prognosis Improving the disease detection process, eventually helps an improved quality of life for patients.
 

Project Implementation Method

we need to test a retinal image to identify glaucoma disease.
1. The image will be provided to Raspberry PI, System will use some pre-processing techniques in order to enhance the quality of image.
2. System will use the base network architecture (figure 2) combined with the specialized layers to segment out blood vessels and optic disc.
3. System will match the image with different database images to check the optic disc
and optic cup ratio (0.3) to detect the actual disease
4. System will predict the results by image analysis on the screen attached with Raspberry Pi.
 

Benefits of the Project

Manual extraction of Retinal lesion (Glaucoma) that is associated with eye is often a problem, for that purpose need experts. Computer based are application are highly encourage to solve this problem.
• Due to the small size of vessels it is difficult to identify the real problem by even spending a lot of effort and time, at the end no satisfied accuracy in results.
• The disease can lead to severe visual impairment and blindness if left untreated.
• The detection of glaucoma disease (shown in figure 1) through Optical Coherence Tomography (OCT) and Heidelberg Retinal Tomography (HRT) is very expensive.

 

Technical Details of Final Deliverable

The minicomputer embedded system to monitor and record the variation in ratinopathy and effected parts of eye from retinal fundus images. It is an easy way to continuouslynotified and help in order to diagnose eye lesions.

Raspberry pi is a basically a minicomputer. It is usually used in embedded system. In our project this system to be an economical, portable and a low maintenance solution for
detection of eye disease applications, especially for hospitals.
 

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Medical

Other Industries

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

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