A faster, gentler way to see deep into the brain: HKU unveils AIMED, a new imaging strategy that slashes acquisition time and light exposure in 3D multiphoton microscopy.
Researchers at the University of Hong Kong (HKU) have developed a novel imaging strategy called AIMED (Arbitrary illumination microscopy with encoded depth) that dramatically increases the speed of 3D multiphoton microscopy while significantly reducing light exposure. The work was led by Professor Kenneth K. Y. Wong at the OMEGA laboratory in the Department of Electrical and Computer Engineering.
How AIMED Works
Traditional multiphoton microscopy builds a 3D image by scanning one thin "slice" of a sample at a time. AIMED fundamentally changes this approach.
- Axial Structured Illumination: Instead of scanning plane by plane, AIMED uses specially structured beams of light to excite multiple depth layers simultaneously.
- Computational Reconstruction: A complex algorithm then separates these overlapping signals to reconstruct a clear 3D image. This process, known as compressive sensing, requires far fewer measurements than traditional methods.
- Smart Beam Shaping: A spatial light modulator (SLM) generates phase masks that split the laser beam into multiple focal spots along the optical axis. Critically, the relative intensities of these spots can be adjusted to compensate for how light naturally fades as it travels deeper into tissue.
- Reduced Crosstalk: The inherent nonlinear nature of multiphoton excitation helps minimize interference, or "crosstalk," between the different depth layers being imaged at once.
"Full 3D reconstruction is achieved from fewer encoded illuminations using sparse optimization algorithms," the team explains.
Performance Benchmarks
The AIMED system was rigorously tested and showed impressive results:
- Resolution: In five-plane configurations, lateral resolution remained around 600 nm, with an axial resolution of 2–4 µm.
- Reduced Phototoxicity: In mouse brain imaging, AIMED used one-half to one-third the per-plane optical power compared with conventional scanning. It achieved similar or better image quality for fine structures like dendrites, axons, and spines—key targets in neuroscience.
- Image Quality: Across compression ratios of 62.5% to 87.5%, the reconstructed 3D images had a structural similarity index of ~0.95 and a peak signal-to-noise ratio of 41–42 dB, indicating high fidelity.
- Speed Potential: Simulations suggest the technology could achieve up to an eightfold increase in acquisition speed for samples with up to 47 axial planes.
Significance and Applications
AIMED is designed as a practical, "plug-in" solution. It provides powerful 3D imaging capabilities without requiring expensive new hardware or major system changes.
This makes it particularly valuable for:
- Imaging sparse structures like neuronal networks.
- Studying phototoxicity-sensitive samples, where minimizing light exposure is critical.
- Potentially being adapted for use in other optical imaging modalities.
The research was published in Advanced Photonics (doi:10.1117/1.AP.7.4.046010).