A novel interface for cortical columnar neuromodulation with multipoint infrared neural stimulation

Feiyan Tian, Ying Zhang, Kenneth E. Schriver, Jia Hu, Anna Wang RoeNature Communications 2024 · DOI

TL;DR

Researchers have developed a high-precision optical interface using pulsed infrared neural stimulation (INS) to target individual functional columns in the visual cortex. By using dense fiber-optic arrays, they demonstrated the ability to selectively enhance or suppress specific visual features (like orientation) in a way that mimics natural brain processing, paving the way for high-resolution visual prosthetics that don't require genetic modification.

What Problem Does This Paper Address?

Visual Cortical Prosthetics (VCPs) aim to restore sight by bypassing damaged eyes and stimulating the brain directly. Current "state-of-the-art" devices primarily use electrical microstimulation. While effective at evoking basic shapes and letters, electricity suffers from "current spread"—the signal bleeds into adjacent neurons, limiting spatial resolution and making it difficult to target the brain’s fine-grained functional units, known as cortical columns.

Optogenetics offers better precision but requires viral vectors to modify human DNA, which presents significant regulatory and safety hurdles. This paper explores Infrared Neural Stimulation (INS) as a "middle way." INS uses brief laser pulses to trigger neurons via localized heating of the cell membrane. It is non-genetic, non-damaging at low intensities, and offers sub-millimeter precision, making it an ideal candidate for interfacing with the brain's columnar architecture.

Methods

The study utilized anesthetized cats, focusing on visual areas 17 and 18, where neurons are organized into columns that respond to specific orientations (e.g., vertical vs. horizontal).

  • The Interface: The team engineered two types of fiber-optic interfaces: a 5-fiber linear array and a 100-fiber (10x10) bundle. These allowed for multipoint stimulation with sub-millimeter spacing.
  • Stimulation: They used a 1870 nm pulsed laser. By varying the sequence and timing of pulses across the fibers, they could mimic moving visual cues.
  • Monitoring: To prove they were activating functional circuits rather than just causing local heat artifacts, they used Intrinsic Signal Optical Imaging (OI). Crucially, they stimulated one hemisphere and imaged the contralateral (opposite) hemisphere. This allowed them to observe how INS signals traveled across the "callosal" connections that link the two halves of the visual field.
  • Experimental Paradigm: They combined INS with actual visual stimuli (drifting gratings) to see if the optical stimulation could modulate ongoing brain activity.

Datasets

As a biological study, the "dataset" consists of experimental trials across 13 cats.

  • 3 animals were used for initial procedure development.
  • 2 animals provided intensity-dependence data.
  • 5 animals were used for the linear array INS/vision experiments.
  • 3 animals were used for the 100-fiber bundle experiments. The researchers used established brain mapping benchmarks (orientation maps) to validate their results.

Key Insights and Ideas

  • Sub-millimeter Precision: The researchers confirmed that INS activation is confined to areas <0.5 mm in diameter. This matches the size of cat orientation columns, proving that INS can "speak the language" of the brain's mesoscale organization.
  • Functional Modulation: INS isn't just a blunt "on" switch. When the orientation of the INS array matched the orientation of a visual stimulus, the brain's response was enhanced. When they didn't match, the response was suppressed. This mimics the "higher-order integration" naturally performed by the brain.
  • The Intensity "Sweet Spot": The study found a clear threshold effect. At low intensities (0.3 J/cm²), stimulation was focal and feature-specific. However, at higher intensities (0.7 J/cm²), they observed a general suppression of cortical activity, likely due to over-activating inhibitory circuits.
  • Dynamic Patterns: By firing the fibers in sequence, the researchers could simulate motion, showing that the interface is flexible enough to represent complex, time-varying visual scenes.

Key Results

  • Focal Activation: Mean activation sizes for radiant exposures of 0.3, 0.5, and 0.7 J/cm² were 240 μm, 350 μm, and 420 μm, respectively—all well within the bounds of a single cortical column.
  • Orientation Specificity: "Vertical" INS applied to the left hemisphere enhanced the response to vertical gratings in the right hemisphere by approximately 0.05% (reflectance change) and suppressed horizontal responses by roughly 0.1%.
  • Callosal Proof: The fact that stimulation in one hemisphere produced predictable, orientation-specific maps in the other hemisphere provides strong evidence that INS successfully engages long-range functional neural networks.

Limitations and Future Work

  • Anesthesia: The study was conducted on anesthetized animals. Neural responses and the "perception" of these stimuli may differ in an awake, behaving state.
  • Recording Gap: The current setup uses optical imaging to monitor activity, which is a slow hemodynamic measure. The authors note that future versions should incorporate electrodes to monitor real-time neural spiking.
  • Chronic Implantation: The fiber arrays used here were apposed to the surface temporarily. For a true prosthetic, a biocompatible, chronically implanted wireless interface (perhaps using μ-LEDs) would be required.
  • Scaling to Humans: While the cat model is excellent for orientation columns, human visual cortex is larger and more complex; scaling the 100-fiber bundle to thousands of channels will be a significant engineering challenge.

Summary updated

Abstract

Cutting edge advances in electrical visual cortical prosthetics have evoked perception of shapes, motion, and letters in the blind. Here, we present an alternative optical approach using pulsed infrared neural stimulation. To interface with dense arrays of cortical columns with submillimeter spatial precision, both linear array and 100-fiber bundle array optical fiber interfaces were devised. We deliver infrared stimulation through these arrays in anesthetized cat visual cortex and monitor effects by optical imaging in contralateral visual cortex. Infrared neural stimulation modulation of response to ongoing visual oriented gratings produce enhanced responses in orientation-matched domains and suppressed responses in non-matched domains, consistent with a known higher order integration mediated by callosal inputs. Controls include dynamically applied speeds, directions and patterns of multipoint stimulation. This provides groundwork for a distinct type of prosthetic targeted to maps of visual cortical columns.
BibTeX
@inproceedings{tian2024novel,
  author = {Feiyan Tian and Ying Zhang and Kenneth E. Schriver and Jia Hu and Anna Wang Roe},
  title = {{A novel interface for cortical columnar neuromodulation with multipoint infrared neural stimulation}},
  booktitle = {Nature Communications},
  year = {2024},
  doi = {10.1038/s41467-024-50375-0},
}
Figure