Kilohertz-rate two-photon voltage imaging of population dynamics in vivo

This paper introduces HS2PM, a hybrid scanning two-photon microscope that overcomes traditional speed-efficiency trade-offs to enable stable, kilohertz-rate, single-cell voltage imaging of large neuronal populations in vivo, thereby revealing fast circuit dynamics and multimodal physiological processes with unprecedented precision.

The Gist

Imagine trying to film a bustling city square from a helicopter. You want to see everyone (the whole crowd), you want to see them clearly (high resolution), and you want to see them move in real-time (high speed).

For decades, scientists trying to film the brain faced a "three-way traffic jam." They could film a small crowd very fast, or a huge crowd slowly, but they couldn't do both at once. If they tried to speed up the camera to catch lightning-fast brain signals (voltage), the image would get too dim or blurry. If they tried to film a huge area, the camera would be too slow to catch the split-second sparks of thought.

This paper introduces a new invention called HS2PM (Hybrid Scanning Two-Photon Microscope) that finally breaks this traffic jam. Here is how it works, explained simply:

1. The Problem: The "Flashlight" Dilemma

Think of the brain's electrical signals like fireflies blinking in the dark.

• Old Microscopes: To see a firefly clearly, you need to shine a bright, focused flashlight on just one firefly at a time. This gives a great picture, but you have to move the flashlight very slowly to check every firefly in a field. By the time you get to the other side, the first firefly has already blinked again.
• The Trade-off: If you try to shine the flashlight on 100 fireflies at once to go faster, the light gets spread thin. The fireflies look dim, and you can't see them clearly.

2. The Solution: The "Magic Multiplexer"

The researchers built a system that acts like a super-fast, magical flashlight that never loses its brightness.

• The Trick: Instead of moving a single flashlight slowly, they take one super-bright laser beam and use a high-tech prism (called a "spatiotemporal multiplexer") to slice it into 16 separate beams instantly.
• The Analogy: Imagine a single chef (the laser) who can suddenly clone themselves into 16 chefs. Each clone works on a different part of the kitchen (the brain) at the exact same time, but they are all powered by the same single energy source.
• The Result: They can scan a massive area (the size of a postage stamp, but zoomed in to see individual cells) 916 times per second. That is fast enough to catch a neuron "firing" (blinking) before it even finishes the thought.

3. What They Saw: The "Spark" vs. The "Hum"

Using this new camera, they didn't just see the big, obvious events; they saw the tiny, hidden ones too.

• The Spikes (Action Potentials): These are the loud "BANG!" moments when a neuron sends a message. It's like a drumbeat.
• The Subthreshold (The Hum): These are the quiet, rumbling vibrations between the drumbeats. Previous cameras were too slow to hear the hum; they only heard the bangs.
• The Discovery: The team found that while the loud "drumbeats" (spikes) get tired and stop firing when you keep poking the mouse with air puffs (a sensory test), the quiet "hum" (subthreshold voltage) keeps going strong and steady. This means the brain is still "listening" and processing information even when it stops shouting back.

4. Why It Matters: The "City Traffic" Metaphor

Before this, studying the brain was like watching a city traffic report that only updates once an hour. You'd see the major accidents (spikes) but miss the slow build-up of traffic jams (subthreshold changes) that cause them.

With HS2PM, scientists now have a live, high-definition drone feed of the entire city. They can see:

• How thousands of cars (neurons) move together in a split second.
• How the "traffic flow" changes when a new event happens (like a sensory stimulus).
• That the "background noise" of the city is actually carrying important information.

5. Bonus Features: Seeing the "Plumbing"

The microscope is so versatile that it can also do other things without changing the camera:

• Blood Flow: It can watch red blood cells zooming through tiny veins like cars on a highway, measuring their speed instantly.
• Chemical Identity: It can tell different types of molecules apart just by how long they "glow" after being hit by light (like identifying different colored glow sticks by how long they last).

The Bottom Line

This paper isn't just about building a faster camera; it's about unlocking a new layer of reality in the brain. For the first time, we can watch the brain's electrical "conversation" happen in real-time, across thousands of people at once, without blinding them with too much light. It's a giant leap toward understanding how we think, feel, and react to the world.

Technical Summary

1. The Problem

Understanding neural computation requires capturing millisecond-scale voltage dynamics across large populations of neurons in vivo. While two-photon voltage imaging (TPVI) offers a path to deep-tissue, high-speed recording, it faces a fundamental three-way trade-off between:

• Imaging Speed: The ability to capture fast action potentials (spikes).
• Field of View (FOV): The ability to record from large neuronal populations simultaneously.
• Excitation Efficiency: The ability to collect enough photons to maintain a high signal-to-noise ratio (SNR) without damaging tissue.

Existing solutions fail to balance these factors:

• Parallel scanning (line/light-sheet): Increases speed but sacrifices photon efficiency and SNR, leading to high laser power demands and potential phototoxicity.
• Single-point scanning: Maximizes photon efficiency and SNR but is limited by the mechanical speed of galvanometers, restricting frame rates and FOV size.

Consequently, current methods cannot simultaneously resolve spikes and subthreshold dynamics in large populations at depths >500 µm with millisecond precision.

2. Methodology: The HS2PM System

The authors developed HS2PM (Hybrid Scanning Two-Photon Microscope), a novel system designed to overcome the speed-efficiency bottleneck by combining high-speed scanning with photon-efficient single-point excitation.

Key Technical Components:

• Hybrid Scanning Architecture:

  • Electro-Optic Deflector (EOD): Angularly encodes an 80 MHz femtosecond laser pulse train, cycling every four pulses to generate four spatially offset points.
  • Spatiotemporal Division Multiplexing (STDM): A module using polarization beam splitters and achromatic delay lines splits each pulse into four time-staggered replicas (~3 ns spacing). This creates 16 spatiotemporally indexed points from a single laser pulse, effectively raising the sampling clock to 320 MHz.
  • Polygon Scanner: A 36-facet polygon scanner rotating at ~55,000 rpm scans these 16 points, generating a base line rate of ~33 kHz (effective line rate of 527 kHz).
  • Galvanometer Scanner: A bidirectional galvo scanner fills the gaps on the Y-axis, producing 544 lines per frame.
  • Result: The system achieves 916 Hz frame rates over a 650 × 524 µm² FOV (571 × 544 pixels).

• Enhanced Photon Collection:

  • Replaced conventional epi-fluorescence relays with a custom three-lens condenser coupled to a frustum-of-cone light guide (CFCL).
  • This design captures scattered photons at large angles (±8°), achieving >80% detection efficiency, significantly boosting SNR and imaging depth.

• Multimodal Capabilities:

  • FLIM (Fluorescence Lifetime Imaging): By reconfiguring the STDM to use two temporally offset windows (prompt and delayed), the system calculates fluorescence lifetime pixel-by-pixel based on intensity ratios, enabling quantitative mapping without electro-optic gating.
  • Vascular Flow Imaging: High-speed acquisition allows for the simultaneous measurement of blood flow velocity and vessel diameter across multiple vessels.

• Data Processing Pipeline:

  • Includes pixel-wise order mapping, bidirectional scan correction, STDM crosstalk demixing (using a linear mixing model), and motion correction/denoising via the VolPy algorithm.

3. Key Contributions

1. Breaking the Trade-off: HS2PM is the first system to achieve kilohertz-rate imaging (916 Hz) over a population-scale FOV while maintaining the photon efficiency of single-point excitation.
2. Deep-Tissue Stability: Demonstrated stable voltage recording from >160 neurons simultaneously at depths up to 700 µm in awake mice.
3. Subthreshold Resolution: Successfully resolved both suprathreshold action potentials and subthreshold membrane fluctuations, which are typically obscured in calcium imaging.
4. Multimodal Platform: Integrated high-speed voltage imaging, fluorescence lifetime imaging (FLIM), and vascular flow analysis into a single optical platform.

4. Key Results

• Performance Metrics:

  • Recorded 165 neurons in a single plane at 490 µm depth with an average spike amplitude of ~35% (ΔF/F).
  • Maintained high spike detection reliability (d' > 3) and SNR (~5) even at 700 µm depth.
  • Phototoxicity: Long-duration imaging (1 hour) showed minimal photobleaching (<35%) and no significant increase in heat shock protein (HSP-70/72) expression, confirming negligible photodamage.

• Sensory Processing & Population Coding:

  • Sensory Adaptation: Applied air-puff stimuli to awake mice. While spike rates showed rapid adaptation (attenuation) over repeated stimuli, subthreshold voltage dynamics remained robust and reproducible.
  • Information Content: Classifiers trained on subthreshold signals outperformed those based on spike rates in decoding stimulus identity (64% vs. 55% accuracy).
  • Functional Heterogeneity: K-means clustering based on subthreshold dynamics revealed distinct neuronal response types (e.g., persistent responders vs. adapting neurons) that were not as clearly separable using spike rates alone.

• Vascular & FLIM Imaging:

  • Simultaneously measured flow velocity and vessel diameter across 33 vessels, revealing a positive correlation consistent with physiology.
  • Generated pixel-resolved lifetime maps distinguishing between Texas Red (3.57 ns), Rhodamine-B (1.50 ns), and ASAP5 (2.72 ns) with high precision.

5. Significance

This work represents a transformative advance in systems neuroscience:

• Circuit-Level Insights: It enables the direct dissection of how spikes and subthreshold dynamics jointly shape sensory adaptation and population coding, questions previously inaccessible with calcium imaging.
• Scalability: The square-format, population-scale imaging field allows established calcium imaging paradigms to be adapted for voltage imaging, facilitating large-scale connectomics and functional mapping.
• Clinical & Research Potential: The system's ability to perform deep, fast, and quantitative imaging with minimal phototoxicity makes it a versatile tool for studying neural circuits in both healthy and diseased states, as well as for investigating neuron-glia interactions and vascular dynamics.

In summary, HS2PM resolves the long-standing limitations of TPVI, providing a robust platform for millisecond-resolved, population-scale interrogation of brain activity.