Mesoscopic calcium imaging in head-unrestrained non-human primates using a lensless microscope

Jacob T. Robinson, Jimin Wu, Yuzhi Chen, Ashok Veeraraghavan, Eyal SeidemannResearch Square 2023 · DOI

TL;DR

Researchers have developed the "Bio-FlatScopeNHP," a lightweight (<18g) lensless microscope capable of mesoscopic calcium imaging in head-unrestrained rhesus macaques. By replacing traditional bulky optics with a computational phase mask, the team achieved a 20 mm² field of view, enabling the first-ever mapping of cortical orientation columns in a freely moving primate.

What Problem Does This Paper Address?

To understand how the brain coordinates complex behaviors, neuroscientists need to see neural activity across large cortical areas (mesoscopic imaging). In non-human primates (NHPs), this usually requires massive table-top microscopes. Because these microscopes are stationary, the animal’s head must be fixed in place.

This restraint is a major hurdle: head fixation is unnatural and disrupts the "eye-head movement coupling" and vestibular inputs that characterize normal behavior. While "miniscopes" exist for rodents, their field of view (FOV) is typically less than 1 mm²—far too small to capture the large-scale functional maps (like the retinotopic or orientation maps in the visual cortex) found in primates. The authors sought to bridge this gap by creating a device that is small enough to be head-mounted but powerful enough to see "the big picture."

Methods

The core innovation is the Bio-FlatScopeNHP, which abandons traditional lenses in favor of computational imaging.

  • Lensless Hardware: Instead of a lens, the device uses a "contour" phase mask placed 3.5 mm above a CMOS sensor. This mask is designed using Perlin noise and Canny edge detection to create a high-contrast, spatially localized Point Spread Function (PSF).
  • Computational Reconstruction: Because the sensor captures a "scrambled" multiplexed image rather than a direct projection, the researchers use algorithms to reconstruct the scene. They employ Wiener deconvolution for fast, real-time feedback and the FISTA (Fast Iterative Shrinkage-Thresholding Algorithm) for high-quality, spatially variant reconstructions.
  • Integrated Illumination: The device includes four blue LEDs and a hybrid filter set (interference and absorption filters) to excite and isolate the green fluorescence of GCaMP6f, a genetically encoded calcium indicator.
  • Miniaturization: By removing the glass optics, the device weight was reduced to just 8.6g (17.2g including the holder), allowing it to sit comfortably on a macaque’s head.

Datasets

The study validated the device using several benchmarks:

  • Standard Targets: 1951 USAF Resolution Targets were used to determine the <10 µm spatial resolution.
  • Biological Samples: Fixed slices of Convallaria Rhizome and patterned "spiking" HEK293 cells served as low-contrast ground-truth comparisons against traditional 4x microscope objectives.
  • In Vivo Data: The primary dataset consists of recordings from the V1 (primary visual cortex) of a male rhesus macaque. This included both head-fixed sessions (for ground-truth comparison with a table-top microscope) and head-unrestrained sessions.

Key Insights and Ideas

  • Decoupling Size from FOV: In traditional optics, a larger FOV usually requires a larger lens and a longer path length. By going lensless, the authors proved that a tiny, flat device can maintain a massive 20 mm² FOV, which is essential for primate neurology.
  • The "Aha" of Orientation Columns: The most significant insight was that the device could resolve "orientation columns"—fine-scale functional structures where neurons respond to specific edge angles. Resolving these ~0.8 mm structures in a moving animal is a major technical feat.
  • Digital Refocusing: Because the system is computational, researchers can "refocus" the image after data collection by using PSFs calibrated at different depths, providing flexibility that fixed-lens systems lack.

Key Results

  • Resolution & FOV: The device achieved a resolution of 9.84 µm over a 20 mm² FOV. For context, this FOV is significantly larger than current state-of-the-art NHP miniscopes (e.g., nVista 3.0).
  • Ground Truth Correlation: In head-fixed trials, the Bio-FlatScopeNHP orientation maps showed a 0.87 correlation with those captured by a traditional table-top widefield microscope.
  • First Unrestrained Map: The researchers successfully recorded orientation maps in a head-unrestrained macaque. These maps showed a 0.72 correlation with head-fixed ground truth, proving that the system remains stable and accurate even when the animal moves its head freely.
  • Stability: Motion analysis showed maximum displacements of only ~17-27 µm during unrestrained trials, which were easily corrected via software.

Limitations and Future Work

The authors acknowledge a few key limitations:

  • Signal-to-Noise Ratio (SNR): The SNR is lower than table-top systems, particularly at the edges of the FOV, leading to a decrease in the measured change in fluorescence (ΔF/F).
  • Uneven Illumination: Current LED placement leads to dimmer edges in the FOV.
  • Future Directions: The team plans to optimize the phase mask and integrate miniaturized OLEDs for more uniform on-chip illumination. They also suggest that using lighter 3D-printed materials and custom PCBs could further reduce weight, potentially leading to fully implantable mesoscopic sensors.

Reference: Robinson, J., et al. "Mesoscopic calcium imaging in head-unrestrained non-human primates using a lensless microscope." Nature Communications (2024). DOI: 10.1038/s41467-024-45417-6

Summary updated

Abstract

<title>Abstract</title> Mesoscopic calcium imaging enables studies of cell-type specific neural activity over large sections of the cortex and over multiple cortical areas. A growing body of literature from multiple animal models suggests that this neural activity can be different when animals are free to move their bodies compared to when they are restrained. Unfortunately, existing systems for imaging calcium dynamics over large areas of the brain in non-human primates (NHPs) are large table-top devices that require researchers to restrain the animal’s head so that it remains under the lens of the microscope. Here, we demonstrate the first fluorescent imaging device, which is capable of imaging mesoscale calcium activity in non-human primates that are free to move their heads. We successfully miniaturized our system by replacing lenses with an optical phase mask and computational algorithms that reconstruct images from complex sensor data. The resulting lensless microscope can fit comfortably on top of the heads of the NHPs, allowing the animals to move their heads freely while imaging. We demonstrate over 20 mm<sup>2</sup> field of view in the primary visual cortex of rhesus macaques. With this approach we measured the first orientation columns maps from a head-unrestrained macaque. Our work establishes functional mesoscopic imaging using a head-mounted lensless microscope as a powerful approach for studying neural activity in head-unrestrained NHPs. These experiments can help reveal the relationship between neural activity and behavior under more naturalistic conditions.
BibTeX
@inproceedings{robinson2023mesoscopic,
  author = {Jacob T. Robinson and Jimin Wu and Yuzhi Chen and Ashok Veeraraghavan and Eyal Seidemann},
  title = {{Mesoscopic calcium imaging in head-unrestrained non-human primates using a lensless microscope}},
  booktitle = {Research Square},
  year = {2023},
  doi = {10.21203/rs.3.rs-3089568/v1},
}
Figure