Precise control of neural activity using dynamically optimized electrical stimulation

  1. Nishal Pradeepbhai Shah  Is a corresponding author
  2. AJ Phillips  Is a corresponding author
  3. Sasidhar Madugula
  4. Amrith Lotlikar
  5. Alex R Gogliettino
  6. Madeline Rose Hays
  7. Lauren Grosberg
  8. Jeff Brown
  9. Aditya Dusi
  10. Pulkit Tandon
  11. Pawel Hottowy
  12. Wladyslaw Dabrowski
  13. Alexander Sher
  14. Alan M Litke
  15. Subhasish Mitra
  16. EJ Chichilnisky
  1. Department of Electrical Engineering, United States
  2. Department of Neurosurgery, United States
  3. Hansen Experimental Physics Laboratory, Stanford University, United States
  4. Neurosciences PhD Program, United States
  5. Department of Bioengineering, United States
  6. AGH University of Science and Technology, Faculty of Physics and Applied Computer Science, Poland
  7. Santa Cruz Institute for Particle Physics, University of California, Santa Cruz, CA, United States
  8. Department of Ophthalmology, United States
8 figures and 1 additional file

Figures

Algorithmic components of the proposed framework for electrical stimulation.

(A) In a healthy retina, the visual stimulus is encoded in the neural response pattern of retinal ganglion cells (RGCs; top row). In a retina with an implant, the visual stimulus is encoded into …

Visual stimulus reconstruction achieved using the greedy temporal dithering algorithm.

White noise target image shown on left. First column: cumulative stimulation count across electrodes after 500, 3000, and 10,000 electrical stimuli (A, B, and C, respectively). Second column: …

Quantifying the performance of dynamically optimized stimulation.

(A) A sample target checkerboard image. (B) ON and OFF receptive fields shaded with the expected summed response from greedy temporal dithering. Achieved reconstructions are shown for (C) greedy …

Experimental validation of dynamically optimized stimulation in the rat retina.

(A) Four sample target checkerboard images. Achieved reconstructions for these images are shown (B) assuming perfect control of retinal ganglion cell (RGC) firing with the available reconstruction …

Spatial multiplexing by simultaneous stimulation of distant electrodes.

(A) Visualization of temporally dithered and spatially multiplexed stimulation. At each time step, multiple single-electrode stimuli are chosen greedily (gray circles) across the electrode array …

Subsampling electrodes for hardware efficiency.

(A) Frequency of stimulating different electrodes (size of gray circles), overlaid with axons (lines), and somas (colored circles) inferred from spatiotemporal spike waveform across the electrode …

Figure 7 with 1 supplement
Extension of dynamically optimized stimulation to naturalistic conditions with eye movements.

(A) Conversion of a visual scene into dynamic stimulus. A target visual scene (left), with sample eye movement trajectory (blue). For each eye position, the population of ganglion cells accessible …

Figure 7—video 1
Greedy temporal dithering and spatial multiplexing in natural viewing conditions.

Top row: operation of the proposed approach when only saccadic eye movements are used. Left: target visual scene with overlaid eye movement trajectory (red). Middle: target stimulus for the cells …

Extension of dynamically optimized stimulation using Structural Similarity (SSIM) perceptual error metric.

(A) Two target images. (B) Reconstruction with MSE and SSIM error metrics for greedy temporal dithering, with a high budget. (C) Same as B, with a low budget.

Additional files

Download links