Image processing to automate early glaucoma diagnosis
By Nicolas Huguet, CBB Biographer
July 29, 2015
CBB member Prof. Vasudevan Lakshminarayanan is a theorist who works in the field of biomedical image analysis amongst other areas which include optical physics and mathematical modeling. He is a recognized expert in this area and is involved in U.S. national science policy (e.g. National Photonics Initiative).
One of the projects he is currently interested in is the automation of glaucoma diagnosis. Glaucoma is one of the major blinding diseases of the eye. One of the ways people currently detect glaucoma is by measuring the intra-ocular pressure. However, by the time a change of pressure in the eyeball is detected, the neural damage is already done. There is a non-invasive way to detect glaucoma long before it is clinically manifested. It involves measuring the cup to disk ratio. The cup and the disk are two components of the eye that can be identified in a fundus photo (a photo of the interior surface of the eye, on the side opposite the lens). Ophthalmologists can estimate the cup to disk ratio by looking at a fundus photo and can thereby diagnose glaucoma, follow the progression of the disease and the effects of treatment. But there are a limited number of glaucoma specialists who can do this especially in developing countries which is why automating this process is so important.
To automate this process, Prof. Lakshminarayanan and his team first take an image of the fundus, segment the image and identify the cup and the disk. They apply tools such as fuzzy set theory and mathematical Hough transforms to get an outline of the disk and the cup. It is then possible to calculate the surface area of each and then obtain the cup to disk ratio. The next step is to create an artificial neural network that will automatically classify any fundus photo as being glaucoma or not.
In order to validate this measurement, Prof. Lakshminarayanan and his team have developed a large database of normal and sick fundus photos and have asked 6 glaucoma specialists to independently mark the outline of the cups and disks and then to estimate the ratio. Prof Lakshminarayanan was happy to report that they have been getting excellent results.
The next step he has in mind for this project is to have one of his students write an app incorporating this process. It will enable local general practitioners in developing countries to take a picture of the fundus of their patient with their phone and run it through the app to diagnose glaucoma. Thanks to telemedicine, if the disease is diagnosed, this information will be sent to a glaucoma specialist at the closest hospital and the disease can be effectively treated in its earliest stages.
Outside of his research activities, Prof. Lakshminarayanan is also involved in education in developing countries. He works with the International Center for Theoretical Physics (Trieste, Italy) on their optics programs. He is a founding member of a UNESCO project called Active Learning in Optics and Photonics. The program consists in teaching teachers all over the world using low-cost, home-made materials to get the concepts across without using pricey equipment. Prof. Lakshminarayanan and the other founders chose Optics because they find it is an enabling technology and omnipresent in our modern life. UNESCO considers this program to be its most successful education program.