Tinkering in the Retina “Garage” to Make Inexpensive Retinal Imaging Systems
Shizuo Mukai, MD
This story starts in 2004 when one of the ophthalmology trainees with an engineering background and I got together to think about and to build prototypes of inexpensive photography systems to image the retina using standard digital cameras. The medical devices for this are very expensive (upwards of 200K) and cumbersome, and the need was an inexpensive, portable system. The early trials taught us a great deal about the fundamentals of retinal photography, illumination, and digital camera control, as well as the differences in the way engineers (my student) and design people (me) think and approach problems.
The improvement in smartphone cameras allowed for use of the phone as the retinal camera. It was also useful as image-recording devices attached to existing ophthalmic instruments or simply as an image-transfer device of photographs taken with a retinal camera. We devised a minimalist system that can be used almost anywhere in the world. Indeed, the technique was successfully taught even in remote countries such as Uganda, Madagascar, India, and Haiti as well as being used effectively for telemedicine and documentation at home here at the Harvard hospitals. The technique was modified for use in experimental animals including rabbits and mice, both being very important in eye research.
Another smartphone-based approach, an app for automated diagnosis of eye conditions that cause a white pupil on snapshots, was conceived by a father of one of my patients that had such a condition. Since one of the conditions is a cancer of the retina in young children called retinoblastoma, early detection is critical. Using machine learning by analyzing thousands of images with and without the white pupil, an app was developed that is free and available on both iPhone and Android platforms. The app can either evaluate the image live or carry out monthly reviews of all photos in your album. We are currently in discussion with Facebook to create a system that would screen the photographs on social media.
Most retinal cameras require the pupil of the eye to be dilated with eye drops. To address the need for an inexpensive, portable, retinal camera that does not require pharmacologic dilation, we built a working prototype camera using the Raspberry Pi® single-board computer with total material cost of $180. We are currently building a dongle to attach the device to a smartphone that would decrease the cost to half and allow Internet and phone-platform access for telemedicine.
Finally, the process has been a lot of fun, and it has taught us tremendously about medical device development. To motivate my students to develop new innovations, a prize was created consisting of an engraved retinal lens, publication in the Digital Journal of Ophthalmology, and funds to develop the innovation.