Context-Aware Image Compression for Artificial Retinas
Artificial retinas are considered an “engineering approach” to restoring sight when the eye/brain neural connectivity is intact. They are considered complementary to approaches involving stem cells or gene therapies, and has already seen some success in preliminary trials. As surgical techniques and the engineering capacity to create ever more compact implantable chips evolve, artificial retinas may become the solution of choice for sight restoration.
Moorfields has a significant grant portfolio on implantation and evaluation of artificial retinas. The project’s Clinical Supervisor has worked on the quantification of the improvement that artificial retina’s of varying spatial resolution bring. However, in all cases, the way with which the image information was relayed from an external camera to the retinal implants was crude. Most approaches relied on a simple subsampling of the observed scene. If such a basic image compression approach restores partial sight and navigation capabilities to a patient, it can be argued that context-aware information compression can be transformative.
This project, thus, can deliver significant benefits to blinding ophthalmic pathology sufferers, can guide the engineering developments of artificial retinas, and can provide insights on the visual cues used during ordinary human activities such as navigation, exercise, or face-to-face communication.