🔬 optics

Asynchronous Event-Based Spectroscopy for Microsecond-Resolved Spectral Reconstruction

This paper presents an event-based spectrometer utilizing a Czerny-Turner configuration and asynchronous sensing to achieve microsecond temporal resolution and kilohertz probing rates, significantly outperforming conventional frame-based systems in tracking rapid spectral dynamics for applications such as microfluidic analysis.

Original authors: Joana M. Teixeira, Tomas Lopes, Tiago D. Ferreira, Catarina S. Monteiro, Pedro A. S. Jorge, Nuno A. Silva

Published 2026-05-11

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Joana M. Teixeira, Tomas Lopes, Tiago D. Ferreira, Catarina S. Monteiro, Pedro A. S. Jorge, Nuno A. Silva

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to watch a very fast movie, like a hummingbird's wings beating or a chemical reaction happening in a split second.

The Problem: The "Snapshot" Camera
Most standard spectrometers (devices that break light into a rainbow to see what it's made of) work like a traditional camera that takes photos. They take a "snapshot" of the light, save the picture, and then take the next one. This is like trying to watch a hummingbird by taking a photo every second. You miss all the movement in between. These devices are limited by how fast they can take pictures and save them, usually only managing a few thousand snapshots per second. If the action happens faster than that, the device just sees a blur or misses it entirely.

The Solution: The "Motion Sensor" Camera
The researchers in this paper built a new kind of spectrometer that doesn't take pictures. Instead, it uses a special "event-based" sensor, which works more like a motion detector in a hallway.

How it works: A normal camera records everything, even if the light isn't changing. This new sensor only "speaks up" when something changes. If a pixel of light gets brighter or dimmer, it instantly sends a tiny signal (an "event") saying, "Hey, something changed here!"
The Analogy: Imagine a room full of people (pixels).
- Old Way: Every second, a teacher asks everyone to raise their hand and shout what they see, whether anything changed or not. This is slow and creates a lot of noise.
- New Way: The teacher says, "Only shout if you see something move." If the room is still, it's silent. If a ball flies across the room, hundreds of people shout instantly. This allows the system to track the ball's movement with incredible speed and precision, without wasting time on the still moments.

What They Built
The team combined this "motion sensor" camera with a classic optical setup (using mirrors and a grating to split light into a rainbow). They created a pipeline that turns these thousands of tiny "change" signals into a readable spectrum (a rainbow graph) in microseconds.

What They Found

Super Speed: They tested their device by flashing a light on and off very quickly (30,000 to 40,000 times a second). Their new device could track these changes perfectly. The old-style "snapshot" device they compared it to was completely blind to these fast flashes; it couldn't even see one full cycle of the light turning on and off.
Accuracy: Even though the new device works so fast, it didn't lose the details. It could still accurately identify the specific colors (wavelengths) and how strong they were, just like a standard device, but at a speed 50 to 100 times faster.
Real-World Test: They put the device in a tiny fluid channel (microfluidics) and watched a red dye flow through it. As the dye moved, it blocked certain colors of light.
- The old device saw the change happen, but it only gave them a few blurry "snapshots" of the transition.
- The new device gave them a smooth, high-speed video of the dye moving, showing exactly how the light faded step-by-step as the dye passed through.

Why It Matters
This new approach is like upgrading from a flip phone to a high-speed smartphone for scientific measurements. It allows scientists to see fast, fleeting events in chemistry and physics that were previously too quick to catch with standard tools. It is also very efficient because it ignores the "boring" parts where nothing is happening, saving energy and computer power.

In Short:
The paper demonstrates a new way to measure light that is incredibly fast and efficient. Instead of taking slow, repetitive photos, it listens for changes, allowing it to capture fast-moving events in the world of light and matter that traditional tools simply miss.

Technical Summary: Asynchronous Event-Based Spectroscopy for Microsecond-Resolved Spectral Reconstruction

Problem Statement
Conventional spectroscopic instrumentation, typically relying on frame-based sensors (CCDs or CMOS), faces fundamental limitations in temporal resolution, sensitivity, and data efficiency. These systems are generally constrained to the kilohertz regime, making them insufficient for tracking physical and chemical processes that evolve on microsecond timescales. While specialized detectors (e.g., streak cameras, intensified CCDs) can achieve picosecond resolution, they often come with high costs, reduced dynamic range, and increased complexity. Conversely, single-pixel photodetectors offer high temporal resolution but sacrifice the ability to monitor full spectral bandwidths simultaneously. There is a need for a system that bridges the gap between conventional spectrometers and high-end ultrafast instrumentation, offering high temporal resolution across a broad spectral range without the redundancy inherent in frame-based acquisition.

Methodology
The authors propose a novel spectrometer design that integrates a Czerny–Turner optical configuration with an event-based vision sensor (EVS). The system operates on the principle that EVS pixels asynchronously report changes in logarithmic intensity only when a predefined threshold is exceeded, eliminating the redundancy of static frames.

Optical Setup: Light enters via a 50 µm slit, is collimated by an off-axis parabolic mirror (25.4 mm focal length), dispersed by a reflective diffraction grating (600 grooves/mm, 500 nm blaze), and focused by an achromatic lens onto the sensor plane. The setup includes a beam splitter to simultaneously direct light to the event-based sensor (Prophesee EVK4 HD) and a conventional camera for alignment.
Signal Processing Pipeline: To convert the asynchronous stream of binary events into calibrated spectra, the authors developed a dedicated pipeline:
1. Region of Interest (ROI) Selection: Acquisition is restricted to the full horizontal range (1280 pixels) and a 150-pixel vertical band to reduce data volume and latency.
2. Temporal Accumulation: Events are accumulated into temporal frames with durations of 20–100 µs.
3. Geometric Correction: A linear fit is applied to correct for optical tilt, realigning the dispersed spectrum.
4. Vertical Integration: A band of ±50 rows around the spectral center is summed column-by-column to generate a 1D spectrum. This coherently sums independent measurements to improve the signal-to-noise ratio (SNR).
5. Smoothing: A Savitzky–Golay filter (50-point window, 3rd-order polynomial) suppresses high-frequency noise.
Calibration: Wavelength calibration is performed using red, green, and blue LEDs. A linear mapping ( $\lambda = a \times \text{pixel} + b$ ) is fitted via least-squares optimization against a reference spectrometer (Avantes) to establish the pixel-to-wavelength relationship.

Key Contributions

Novel Architecture: The first demonstration of an event-based sensor used for broadband spectroscopy, utilizing a Czerny–Turner configuration to cover the visible range (414.47–648.21 nm) with a resolution of approximately 0.18 nm/pixel.
High-Speed Reconstruction: The system achieves spectral reconstruction at probing rates up to tens of kilohertz (demonstrated at 60 kHz and 80 kHz effective rates), far exceeding the ~500 Hz practical limit of a parallel conventional frame-based spectrometer.
Data Efficiency: By leveraging the asynchronous nature of the sensor, the system inherently suppresses constant illumination, generating data only in response to temporal intensity changes. This is particularly advantageous for transmission measurements against strong backgrounds.
Microfluidic Validation: The system was validated in a microfluidic environment integrated with an inverted microscope, successfully tracking spectral changes induced by an absorbing dye with higher temporal fidelity than frame-based approaches.

Results

Spectral Fidelity: Under temporally modulated illumination (30 kHz and 40 kHz square waves), the event-based spectrometer accurately reproduced the positions of dominant spectral peaks and relative spectral features compared to a commercial reference spectrometer. While absolute intensity was not preserved in normalized representations, the spectral shape fidelity was confirmed.
Temporal Resolution: The system successfully detected modulation cycles that were undetectable by the frame-based spectrometer (which could not resolve a single cycle at the test frequencies). The effective probing rate was shown to be twice the modulation frequency due to the sensor's sensitivity to both rising and falling edges.
Event Yield Analysis: Event counts decreased at higher modulation frequencies (40 kHz vs. 30 kHz) due to reduced optical power reaching the sensor caused by the LED driver's finite rise/fall times. However, the system maintained stable signal detection.
Dynamic Tracking: In the microfluidic experiment, the event-based system captured multiple intermediate spectral states during the introduction of a red dye, revealing the progressive attenuation of the blue spectral peak. In contrast, the frame-based system recorded only a limited number of spectra during the same transition.
Real-Time Capability: The system demonstrated the ability to perform real-time spectral reconstruction by accumulating events over short windows, with the potential to conditionally process data only when meaningful activity (event thresholds) is detected, thereby reducing computational load.

Significance and Claims
The paper positions event-based spectroscopy as a promising paradigm for real-time, high-temporal-resolution measurements in dynamic and low-light applications. The authors claim that this approach effectively bridges the gap between conventional spectrometers and high-end ultrafast optical instrumentation. Key advantages highlighted include:

Cost-Effectiveness and Flexibility: The system uses standard dispersive optics and a compact sensor, avoiding the complexity and cost of streak cameras or intensified detectors.
Resource Efficiency: The asynchronous, data-driven nature of the sensor minimizes memory usage and computational overhead by avoiding the acquisition of redundant static data.
Enhanced Contrast: The inherent suppression of constant illumination improves spectral contrast, particularly beneficial for detecting small spectral variations against strong backgrounds in transmission measurements.

The authors conclude that while the system is currently a proof-of-concept, it establishes the feasibility of using event-based sensors for high-temporal-resolution spectral tracking in microfluidic and other dynamic environments, offering a complementary approach to traditional spectroscopic methods.

Technical Summary: Asynchronous Event-Based Spectroscopy for Microsecond-Resolved Spectral Reconstruction

More like this