At our March event we had the pleasure of welcoming Team Eye Works to NEBytes to present their entry in the 2012 Imagine Cup. We asked the team to provide us with a blog post about their project and so I’ll hand the content over to them:
Mobile Intelligent Retinal Analysis (MIRA)
Imagine Cup entry by Team Eye Works
Four students from the North East of England have been bought together to make up Team Eye Works. The team is competing in the Microsoft Imagine Cup. Consisting of 3 students from Northumberland University, and one student from Newcastle College, the team came together with a wide range of skills, with the intention of tackling one of the world’s great problems.
A lot of people take the gift of sight for granted, and don’t realise how easily this gift can be taken away from them, due to everyday conditions, which, if detected early, can be prevented.
In the UK, recent studies have shown that 2 million people were living with various degrees of sight loss. It has been shown that 50% of the 2 million cases could have been prevented, and the World Health Organisation has discovered that 90% of people suffering from sight loss are living within the developing world.
Out of the various causes of sight loss, the main contributors are cataracts, glaucoma, and age-related macular degeneration.
Team Eye Works have proposed to tackle this problem by making a mobile platform app capable of detecting the symptoms of the previously mentioned causes of sight loss.
This shall be accomplished by creating a piece of software which will be capable of analysing an image of the human retina to detect different abnormalities. This will be done by comparing the taken image against a library of images. The library of images shall contain pictures of retinal abnormalities. By using Microsoft Azure, the software will be able to upload and download images for analysis. It is important to use a system such as Microsoft Azure, as it can contain, and store a large amount of data, which will be available almost anywhere in the world, without the need for a physical backup. This will allow Team Eye Works’ solution to be used globally. Apart from the cloud, the mobile app will also be able to do a local comparison of the retina image using neural networks. A different neural network will be trained beforehand for each disease and used on the mobile app to quickly analyse the retinal image and give a probability of infection. As internet connectivity is not as reliable in developing countries this feature is required. It was decided to go with the neural network approach to effectively use the limited memory and computational power of mobile phones, as the approach of comparing retinal image with the image database (used by the cloud) is highly expensive, demanding a lot of computational and memory requirements.
Whilst the software is a very important element of the proposed solution from Team Eye Works, hardware is just as important. The solution will be developed for Windows Mobile technology, and will require other physical elements. Currently there is no mobile device on the consumer market that has a suitable camera and lens package. This has meant that Team Eye Works have researched, and developed a lens attachment and an apparatus used for chin rest and the camera stand, that will be capable of allowing mobile devices to capture best quality image of the retina, which will be suitable for the image analysis. The image displayed below shows the flow of the proposed system.