Experimental Demonstration of Genetic Algorithm Based Metalens Design for Generating Side‐Lobe‐Suppressed, Large Depth‐of‐Focus Light Sheet

Abstract

Light-sheet fluorescence microscopy (LSFM), sectioning biological samples by illuminating a thin slice of fluorescently labelled live cells or tissues typically with a Bessel beam, requires dithering the beam to form a two-dimensional (2D) light sheet. It usually suffers from severe phototoxicity and low signal-to-noise ratio (SNR) mainly caused by the side-lobe illumination generating unfavorable bio-fluorescence from the adjacent tissues. Here, the first proof-of-concept experimental implementation of genetic algorithm (GA) generated metalens is provided to address the above challenges. It is shown that a dithering-free 2D light sheet produced by a GaN-based metalens with GA-generated prism-like yet non-analytical phase profile, can significantly suppress the side-lobe intensity of the resultant light sheet down to 7.3% of the main lobe intensity and also extends its depth of focus up to 4 mm, surpassing the latest results reported in the literature. When applied under two-photon excitation, the light sheet exhibits an enhanced axial resolution and SNR. These results demonstrate the feasibility of applying artificial intelligence generated metalens in addressing some special issues encountered by conventional analytical design approaches, and the metalens device produced here could find an important role in fast-LSFM-based large-scale bioimaging applications without mechanical dithering.

Artificial Intelligence in Meta-optics

“…Genetic algorithm (GA, sometimes named evolutionary algorithm) is one of the most frequently used evolutionary computation strategies. Additionally, GA has greatly promoted the inverse design of metasurfaces in recent years, such as a metalens, a terahertz quarter-wave plate, programmable metamaterials, and subwavelength lattice optics …”

“…Genetic algorithm (GA, sometimes named evolutionary algorithm) 162 is one of the most frequently used evolutionary computation strategies. Additionally, GA has greatly promoted the inverse design of metasurfaces in recent years, such as a metalens, 163 a terahertz quarter-wave plate, 164 programmable metamaterials, 165 and subwavelength lattice optics. 166 Figure 17a demonstrates a general working flow of GA. GA usually takes a set of randomly generated solutions or artificially set initial points as the first generation of the population.…”

Large-scale achromatic flat lens by light frequency-domain coherence optimization

“…2a, which includes three main steps. The first is Search, which can yield the optimal height distribution by combining genetic algorithm (GA) and Hook-Jeeves algorithm (HJA) [48][49][50] to avoid local minima and speed up the convergence as well. The second is Smooth, which can remove the high aspectratio structures in the lens.…”

Large-scale achromatic flat lens by light coherence optimization

“…2a, which includes three main steps. The first is Search, which can yield the optimal height distribution by combining genetic algorithm (GA) and Hook-Jeeves algorithm (HJA) [46][47][48] to avoid local minima and speed up the convergence as well. The second is Smooth, which can remove the high aspect-ratio structures in the lens.…”