The increasing prevalence of diabetes in the population is associated with several health issues. One of those issues is the development of diabetic retinopathy, where a diabetic patient will experience cumulative changes to his retina, which ultimately may lead to blindness. Diabetic Retinopathy progression is mostly asymptomatic until the later stages of the disease, when vision loss settles in, and at that point it is very difficult to reverse. Around 50% of all the diabetic patients for more than ten years exhibit some stage of diabetic retinopathy.
The asymptomatic profile of the initial progression and the high effectiveness of early treatment have motivated the implementation of extensive screening programs covering the diabetic population, in which images of the patient retinas are acquired and subsequently analysed by an expert. However, this requires the use of relatively expensive and cumbersome equipment to acquire the retinal images as well as a time consuming analysis of those images by ophthalmologists.
AICOS is working towards a solution for screening for diabetic retinopathy, by developing a mobile device capable of acquiring images of the retina, and have those images later be analysed automatically by machine learning algorithms.
The mobile prototype device used for retinal image acquisition comprises an optical setup allowing illumination of the retina, and the imaging responsibilities are currently assumed by a smartphone camera.
In the interest of improving the robustness, usability and consistency of the solution, reducing the variability caused by the use of different smartphone models is crucial. The increasing speed of hardware and software development in new smartphone cameras constantly being released to the market, also contributes to the required continuous adjustments of camera control parameters to minimise compatibility issues.
> Improvements in project EyeFundusScope imaging system and acquisition protocol: with actuators for internal fixation points that increases the robustness and coverage of the retinal tissue, and simplify prototype operation by non-expert;
> Customised low-cost optical and sensing devices;
> Android API or component to support the developed camera control.
Author: Helena Piteira
Type: MSc thesis
Partner: Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa