Bionic Vision Lab at UC Santa Barbara

We are an interdisciplinary research group focused on the development of novel methods and algorithms to interface sight recovery technologies such as retinal implants (‘bionic eye’) with the human visual system, with the ultimate goal of restoring useful vision to the blind.

Logo © 2019 @drdina, Dina’s SciArt.

PCMag: Restoring vision with bionic eyes - no longer science fiction

(Jul 9, 2019) Prof. Beyeler recently sat down with PCMag to talk about bionic vision and his move to UC Santa Barbara.

Read interview

Previous Next

Research

Fig. 1: Degenerative retinal diseases cause irreversible vision loss in more than 10 million people worldwide. Analogous to cochlear implants, retinal prostheses electrically stimulate surviving retinal cells in order to evoke neuronal responses that are inter-preted by the brain as visual percepts (‘phosphenes’).

Restoring vision to the blind

The ‘bionic eye’—so long a dream of the future—is finally becoming a reality with retinal prostheses available in the US and Europe (Fig. 1; over 300 patients implanted). With cortical implants, optogenetic approaches, and stem cell therapy in development, a wide range of sight restoration (SR) options should be available to patients diagnosed with severe blindness within the next decade.

Despite the increasing clinical and commercial use of these devices, the perceptual experience of SR patients is surprisingly poorly understood. A common misconception in the field is that each electrode in an array can be thought of as a ‘pixel’ in an image; to generate a complex visual experience, one then simply needs to turn on the right combination of pixels. However, almost all SR technologies are likely to suffer from perceptual distortions and subsequent loss of information due to interactions between the technology and the underlying neurophysiology.

The goal of our research is therefore:

  • to understand how interactions between SR technologies and neurophysiological mechanisms shape the visual perception of SR patients, and
  • to use this knowledge to develop advanced stimulation strategies for different SR devices, with the ultimate goal of restoring useful vision to the blind.

Clinical studies have demonstrated that the vision provided by current SR devices differs substantially from normal sight.

Rather than predicting perceptual distortions, one needs to solve the inverse problem: What is the best stimulus to generate a desired visual percept?

Novel stimulation strategies can be tested on sighted subjects viewing a simulation of prosthetic vision in virtual/augmented reality.

Lab members

Principal Investigator

Avatar

Michael Beyeler

Assistant Professor

Graduate Students

Avatar

Zuying Hu

MS Student

Avatar

Justin Kasowski

PhD Student

Avatar

Aiwen Xu

PhD Student

Undergraduate Students

Avatar

Anvitha Akkaraju

Research Assistant

Avatar

Dylan Lin

Research Assistant

Avatar

Ryan Neydavood

Research Assistant

Avatar

Rashi Raghulan

Research Assistant

Avatar

Nathan Wu

Research Assistant

Selected Publications

All publications

We systematically explored the space of possible implant configurations to make recommendations for optimal intraocular positioning of Argus II.
Michael Beyeler, Geoffrey M. Boynton, Ione Fine, Ariel Rokem MICCAI 2019
Preprint PDF DOI

In this review, we provide an accessible primer to modern modeling approaches and highlight recent data-driven discoveries in the domains of neuroimaging, single-neuron and neuronal population responses, and device neuroengineering.
Bingni W. Brunton, Michael Beyeler Curr Op Neurobiol 58:21-29
PDF DOI

We show that the perceptual experience of retinal implant users can be accurately predicted using a computational model that simulates each individual patient’s retinal ganglion axon pathways.
Michael Beyeler, Devyani Nanduri, James D. Weiland, Ariel Rokem, Geoffrey M. Boynton, Ione Fine SciRep 9(1):9199
Preprint PDF Code Dataset DOI

pulse2percept is an open-source Python simulation framework used to predict the perceptual experience of retinal prosthesis patients across a wide range of implant configurations.
Michael Beyeler, Geoffrey M. Boynton, Ione Fine, Ariel Rokem SciPy: 81-88
Preprint PDF Code Slides Video DOI

Teaching

All courses

An overview of theory and research into the human performance and biological processes of visual perception. Typical topics may range from the detection of simple stimuli to the identification of objects and events.

Overview of computer vision problems and techniques for analyzing the content of images and video. Topics include image formation, edge detection, image segmentation, pattern recognition, texture analysis, optical flow, stereo vision, shape …

Join Us

We are looking for curious and talented individuals who share our vision of helping the blind see again.

We are an interdisciplinary group working at the intersection of computer science, neuroscience, and psychology. We value a strong work ethic, clarity of thought, creativity, optimism, and computational/programming skills. We practice open science, and we know that science is at its best when it is diverse. If these values resonate with you, read on.

Postdocs

Please contact Michael with your CV and a brief statement of research accomplishments, interests, and career plans. Check the Research section to see a list of projects we are particularly interested in.

To be considered, you will need at least one publication in a high-quality international conference (computer science) or journal (neuroscience) and you need to have a reasonable chance of getting a fellowship to support your stay at UCSB.

PhD Students

If you are a UCSB graduate student looking for a lab, please contact Michael to set up a time to meet.

If you are looking for a graduate position and you are not already at UCSB, the deadline for Fall 2020 has now passed. The next application deadline is in December 2020.

Master's Students

If you are a UCSB CS Master’s student looking for a project, please contact Michael to set up a time to meet.

We are looking for students interested in applying their methodological skills to research problems in bionic vision. All students should have a solid programming background and strengths in one of the following:

  • Machine learning/data science: Build predictive models applied to real-world datasets collected on retinal prosthesis patients (classification, regression, time-series analysis, interpretable models, heterogeneous data).
  • Computer vision: Build predictive models applied to biomedical image datasets (registration, segmentation, retinal fundus imaging, optical coherence tomography).
  • Software engineering/parallel programming: Develop parallelization back ends for pulse2percept, our open-source Python-based simulation framework (Python, Cython, SciPy, OpenMP, GPGPU, JAX).

Undergraduate Students

If you are a UCSB undergraduate, please contact Michael with your prior experience and your transcript (unofficial is fine) and we can arrange a meeting.

Students should have a GPA of ≥ 3.0 (University requirement) and some programming experience (especially Python/Matlab). In general, students stay for several quarters (one quarter is not enough to get up to speed) and work for academic credit or on a volunteer basis.

Contact