Deep inside a mouse's brain, researchers at the University of Hong Kong have just proved they can watch neurons light up eight times faster than before — while actually being gentler to living tissue.
The breakthrough comes from a technique called AIMED, short for Arbitrary Illumination Microscopy with Encoded Depth, developed by Professor Kenneth K. Y. Wong's OMEGA laboratory in the Department of Electrical and Computer Engineering at HKU. It solves one of the most persistent frustrations in 3D biological imaging: the old way of looking deep into living brains required shining so much light on the sample that it risked damaging the very structures scientists hoped to study. AIMED changes that equation entirely, making it possible to see fine details like dendritic spines and axons with only half to one-third of the optical power that conventional methods demand.
The innovation hinges on a deceptively simple idea: instead of scanning layer by layer through a 3D sample, AIMED excites multiple depth layers simultaneously using specially designed phase masks loaded onto a spatial light modulator. These masks split a laser beam into multiple controllable focal spots along the direction of propagation, each with independently adjustable intensity. The biological sample itself does the heavy lifting — the nonlinear nature of two-photon and three-photon excitation naturally suppresses unwanted light from other layers, keeping each encoded depth clean and independent. Then, instead of collecting full datasets from every plane, the system gathers compressed measurements and reconstructs the entire 3D volume using sparse optimization algorithms, a computational approach borrowed from information theory.
The results speak for themselves. In imaging experiments on mouse brain samples, AIMED resolved fine neuronal substructures like dendrites and axons using only 60% of the typical data — yet the reconstructed images maintained a structural similarity index of approximately 0.95 and a peak signal-to-noise ratio of 41–42 dB, showing negligible degradation compared with fully sampled scans. Even more delicate structures, like dendritic spines, came through with fidelity comparable to or better than traditional high-power sequential scanning. The lateral resolution held steady at about 600 nanometers, with axial resolution ranging from 2 to 4 micrometers.
What makes this finding particularly significant is its scalability. Simulation studies showed that when handling large-scale imaging tasks involving up to 47 axial planes, AIMED achieved approximately an eightfold increase in acquisition speed. That matters enormously for researchers trying to capture fast biological dynamics — the flickering of neural activity, the growth of connections between neurons, processes that happen in real time and demand that cameras keep pace.
The work, published in Advanced Photonics, represents a paradigm shift away from hardware-heavy acceleration strategies toward a more elegant solution: smarter optics combined with smart mathematics. Because AIMED requires minimal additional system complexity, it can be adopted as a plug-in enhancement to existing multiphoton microscopes rather than demanding completely new equipment. For neuroscientists and cell biologists watching the frontiers of deep-tissue imaging, this is the kind of breakthrough that opens doors — literally seeing further, faster, and more safely into the living brain.