Iris based Noninvasive Glucose Monitoring System

Project Title

Iris based Noninvasive Glucose Monitoring System

Project Area of Specialization Artificial IntelligenceProject Summary

Diabetes is the fastest growing chronic disease affecting over 400 million people worldwide. The Diabetes Prevalence Survey 2017 of Pakistan (see Figure 1) has revealed staggering 16.98 percent prevalence of diabetes as 35.3 million people among the adult population are found diabetic in Pakistan. According to recent studies of WHO, right now, 1 person out of 11 adults have diabetes in the world. Pakistan is ranked 7th in the world for diabetes prevalence. The patients have to constantly monitor and measure their blood sugar levels many times in a day and regulate their diet accordingly. The conventional techniques for sugar testing use invasive finger stick tests which are not only painful but are also inefficient and costly i.e., a new test strip is required every single time.

To overcome these problems, in this project, we aim to develop a non-invasive, painless and cost-effective method for blood glucose monitoring using images of the human eyes. The glucose levels in the human eye are correlated with the glucose levels in blood. Thus, changes in the blood glucose level lead to changes in the glucose concentration in the fluid of the eye. This leads to very subtle differences in how the iris appears. In this work, we develop a complete system with a mobile application interface, which requires the patient to capture a high definition image of their eye with their smart phone using a macro lens. The captured image will be the input of our Iris based non-invasive glucose monitoring mobile App. The mobile app will communicate with the server to process images and retrieve results.The server will detect the subtle differences in the captured iris image using a pre-trained deep learning based model by detecting the changes in the fluid of the eye. The mobile app will report the sugar level of a person by processing the changes in Iris shape.

Project Objectives

The broader objectives of our project are the following:

1. The system (mobile application) will capture the digital images using mobile integrated Macro lens.
2. The App will communicate with the server to process images and retrieve results.
3. The server will detect the subtle differences in the captured iris image using a deep learning technique by detecting the changes in the fluid of the eye.
4. The App will display the glucose level of the captured image.

Project Implementation Method

The detailed steps involved in the proposed Glucose Monitoring System are shown in the following diagram:

• Iris Acquistion (Diabetic and Non diabetic images)
• Pre-processing (Read image, Resize Image)
• Sent to server
• Image correction
• Iris segmentation
• Feature Extraction
• Classification using deep learning model
• Show glucose level
• Result set back to application
• Display results on application 

Benefits of the Project

1. A non-invasive, smartphone-based method.
2. The proposed solution is cost-effective and painless.
3. For sugar monitoring, unlike other conventional methods the patient’s blood will not be wasted.
4. Glucose level results will be displayed after just one click.
5. Maintain patient’s glucose history and insights.

Technical Details of Final Deliverable

1. We are going to develop a mobile application that will sends an image of human eye (taken by the user through macro lens using smartphone) on the server.
2. The server will start processing the image using a constrained deep learning model specifically designed to detect changes in the Iris patterns.
3. It concludes information about glucose level and send back to application.
4.  Users can check their glucose level on the dashboard of an application.

Final Deliverable of the Project Software SystemCore Industry ITOther Industries Medical Core Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals No Poverty, Good Health and Well-Being for People, Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	70000
Macro Lens	Equipment	1	15000	15000
GPU (NVIDIA K40)	Equipment	1	55000	55000