Ophthalmoscope: Saving eyes with Raspberry Pis
The Raspberry Pi is being used to save the eyesight of people in India thanks to the Open Indirect Ophthalmoscope (OIO) project.
Inside the OIO, machine learning technology is used to spot eye problems. Subsequently, the OIO becomes better at checking for problems over long-term use.
“The Open Indirect Ophthalmoscope is a portable retinal camera that uses machine learning to make diagnosis not only affordable but also accurate and reliable,” says Sandeep Vempati, a mechanical engineer at the Srujana Center for Innovation, a part of the L V Prasad Eye Institute (LVPEI).
The heart of the OIO is a Raspberry Pi. Our low-cost computer drives down the cost of taking high-quality photos of the retina.
“Currently, visual impairment affects 285 million people worldwide,” said Sandeep. “What’s more surprising is the fact that 80 percent of all visual impairment can be prevented or cured, if diagnosed correctly.”
“India is the diabetes capital of the world,” explained Dr Jay Chhablani, a specialist in retinal disease at the LVPEI. “Diabetes leads to something called diabetic retinopathy”.
For that reason, it’s important to remove barriers to treatment. “If we see the patient at an early stage,” says Dr Chhablani, “we can treat them by controlling diabetes and applying laser treatment”.
“Although eye care services have become increasingly available,” said Sandeep, “diagnosing diseases like diabetic retinopathy is still a problem in many parts of the world.”
Sandeep’s team strove to build an open device. As a result, OIO can be 3D printed and assembled anywhere in the world.
“3D printing creates the OIO for a fraction of the cost of conventional devices, and yet maintains the same quality,” explains Sandeep.
Compared to professional devices, the OIO costs just $800 to build. In contrast, professional retinal cameras can cost around ten times as much.
Over on OIO’s Hackaday page you will find the components. Inside is a Raspberry Pi 3, a Camera Module, a 20 dioptre lens, front-end mirrors, and a 5-inch touchscreen.
“Engineering feels great when you see a product being useful in the real world,” says Sandeep.
So a 10th of the price and better quality (in my opinion)… why do i get the feeling there’ll be restrictions of use over here.
There will be restrictions in use in the UK and similar countries due to the necessary approvals process that all such equipment must go through. Quality and safety standards must be applied to all medical equipment and obtaining these is extremely expensive. In India this, presumably, does not apply, hence the equipment can be marketed unapproved. Further, as I read it the price is for a kit of parts, not a complete unit. If this were to be assembled in the UK at our current minimum wage rate this would also add to the cost. Having said all of that, it looks like a nice piece of kit and a great idea.
This is really incredible!
Technology used right!!
What a brilliant Idea! (eye-deer, get it?) ;-)
I’m not sure where the “machine learning” aspect comes in though, surely a qualified ophthalmologist still needs to assess the retinal images taken by the camera?
I guess it is a simple binary classification problem, where you will have corpus of eye images taken by the camera , and over time use these images as parameters to determine wether there is a problem or not.
thing is a relatively untrained person could take the photos then email them to a a trained consultant for interpretation
so a consultant could cover many more patients and nurses out in the field do the checks
would be great for getting this sort of checks done in middle of africa etc
a bit like the drone for flying blood sample to a hospital a few months back
My late father, an ophthalmologist, used to tell me that he was able to spot diabetic retinopathy in patients who had not yet been diagnosed with diabetes. He would be pleased to know that a relatively inexpensive machine could help in this process. Good job!
why isnt on the news ?
Yet again Wow….
Where can we get more information on OIO ?
Comments are closed