Rather than aiming to one day restore natural vision, we might be better off thinking about how to create practical and useful artificial vision now.
Roksana Sadeghi
(she/her)
Postdoctoral Researcher
University of California, Santa Barbara
Roksana Sadeghi is a postdoctoral researcher in the Computer Science department. She has a PhD in Biomedical Engineering from Johns Hopkins University, where she worked with Dr. Gislin Dagnelie to study the functional vision of people with visual impairment. She also led the psychophysical experiments to understand the visual perception evoked by the intracortical visual prosthesis (ICVP) in collaboration with Chicago Lighthouse and Illinois Institute for Technology. Then, her PhD was followed by a two-year postdoctoral training at the University of California, Berkeley, in the Vision Science department, where she worked with Dr. Jorge Otero-Millan to develop an open-source, modular, and cost-effective framework for video-based eye trackers called OpenIris.
She joined Dr. Beyeler’s lab at UCSB in February 2025, hoping to combine her skills in eye tracking and human studies, studying the eye movements of people with visual impairment and improving functional vision by incorporating eye-tracking methods.
Outside of the lab, Roksana enjoys running, hiking, playing piano, and listening to music.
- 3205 BioEngineering
Education
-
PhD in Biomedical Engineering, 2016 - 2023
Johns Hopkins University
-
MS in Biomedical Imaging, 2014 - 2015
University of California, San Francisco
-
BS in Physics, 2009 - 2014
Sharif University of Technology, Theran, Iran
Project Lead
Project Affiliate
Simulating Visual Impairment
How are visual acuity and daily activities affected by visual impairment? Previous studies have shown that vision is altered and impaired in the presence of a scotoma, but the extent to which patient-specific factors affect vision and quality of life is not well understood.
Publications
Efficient spatial estimation of perceptual thresholds for retinal implants via Gaussian process regression
We propose a Gaussian Process Regression (GPR) framework to predict perceptual thresholds at unsampled locations while leveraging uncertainty estimates to guide adaptive sampling.
Roksana Sadeghi, Michael Beyeler arXiv:2502.06672
