How can virtual patients improve performance of real sight recovery patients?

Embedding a computational model that can predict the perceptual distortions encountered by sight restoration (SR) patients in virtual reality (VR) will enable sighted subjects to act as virtual patients in real-world tasks. This will allow us to test novel stimulation strategies in high-throughput experiments. Strategies that result in good VR performance will then be validated in real SR patients.

For example, rather than aiming to ‘restore natural vision’, there is potential merit in borrowing computer vision algorithms as preprocessing techniques to maximize the usefulness of prosthetic vision. Edge enhancement and contrast maximization are already routinely used in Argus II. In the future, more sophisticated techniques such as low-level image enhancements and visual saliency based transforms could further improve visual performance.

Project Lead

Justin Kasowski
PhD Student

Justin Smith is a DYNS PhD student in the Bionic Vision Lab at UCSB.