Breaking the depth-resolution trade-off in nonlinear microscopy : adaptive optics and computational super-resolution with upconversion nanoparticles

Abstract

Nonlinear optical microscopy (NLOM), typified by multiphoton microscopy, provides intrinsic optical sectioning for deep-tissue imaging but typically requires high-intensity pulsed excitation that induces phototoxicity and photobleaching. Lanthanide-doped upconversion nanoparticles (UCNPs) effectively overcome these constraints. Upon continuous-wave (CW) near-infrared (NIR) excitation, they sequentially absorb multiple photons and emit higher-energy NIR or visible fluorescence, thereby enabling high-resolution, high signal-to-noise ratio and multicolor imaging with substantially reduced excitation requirements and minimal photodamage. Furthermore, their exceptional photostability effectively precludes photobleaching, and excitation around 975 nm markedly suppresses both scattering and endogenous autofluorescence, thus permitting robust, stable, high-contrast imaging at depths extending to several hundred micrometers. The overarching objective of this thesis is to exploit the unique photophysical properties of UCNPs in combination with two complementary strategies, namely adaptive optics (AO) and computational super-resolution optical fluctuation imaging (SOFI), in order to surmount the inherent trade-off between imaging depth and spatial resolution in nonlinear microscopy. This thesis is structured around three major innovations. The first is UCNP-assisted direct AO microscopy. We have developed a real-time AO nonlinear laser-scanning microscope that incorporates a Shack-Hartmann wavefront sensor and a deformable mirror. The nonlinear fluorescence from UCNPs sample serves simultaneously as an imaging signal and as intrinsic guide stars for direct wavefront sensing. By applying real-time closed-loop correction, distorted excitation light can be precisely refocused, restoring near diffraction-limited spots even in large-aberrations sample. This correction results in significant improvements in image sharpness and signal strength under CW excitation. The second innovation is homologous dual-emission upconversion AO microscopy (HDU-AOM). This approach employs Tm³⁺/Yb³⁺ co-doped UCNPs' dual 455 nm/800 nm emission: the 800 nm emission for aberration measurement (guide-star) in deep tissues and the 455 nm emission for high-resolution imaging at matching depths. Using a home-built nonlinear laser scanning microscope with a 975 nm CW laser, we achieved near-diffraction-limited imaging (~480 nm laterally) at a 500 μm depth in the mouse brain environment with significant optical aberrations, thereby directly coupling near-infrared sensing with visible-band imaging to effectively overcome the depth-resolution trade-off. The third innovation is speckle-based upconversion nonlinear SOFI (SUN-SOFI). In this approach, we replace molecular blinking with dynamically modulated speckle illumination generated by a spatial light modulator. Speckle-induced intensity variations supply the fluctuation statistics necessary for cumulant analysis while preserving the nonlinear excitation and high-photostability benefits of UCNPs. These fluctuations are subsequently analyzed using high-order SOFI cumulants to recover spatial information beyond the diffraction limit. Experimental results demonstrate approximately a twofold resolution enhancement over traditional wide-field imaging, achieving lateral resolution down to ~160 nm with sixth-order analysis. The method is straightforward to implement and preserves the low phototoxicity and high-resolution of NIR-driven UCNP excitation, making it particularly attractive for live-cell super-resolution applications. Collectively, by integrating UCNPs with AO and computational super-resolution imaging, this thesis demonstrates high-resolution deep-tissue and super-resolution imaging under CW excitation with minimal photodamage. Leveraging the superior optical properties of UCNPs, including high-order nonlinearity, dual-emission capability, extended penetration depth, and exceptional photostability, enables sub-micron spatial resolution at the single-nanoparticle level within scattering tissues, thereby strengthening the capacity of nonlinear microscopy for real-time monitoring of deep cellular dynamics. These advances significantly extend the potential of nonlinear optical microscopy for biomedical research and clinical translation.