Wei Ren

Deep learning has demonstrated remarkable success in augmenting fluorescence imaging under photon-limited conditions. However, existing restoration networks are typically devised for training with augmented patches far smaller than the full-view raw data, an overlooked aspect that compromises fidelity and noise-resistance due to the loss of global statistics. To address this limitation, we propose a large-patch network (LargePNet), which synergizes the large effective receptive field provided by shallow ultra-large-kernel convolutions and the nonlinear representation capabilities of deep networks through scale separation. It effectively and efficiently leverages large-view global information for restoration. Directly trained with large-view images, LargePNet shows contrasting advantages over state-of-the-art small-patch networks, with 0.5-2 dB higher peak signal-to-noise ratio across eight representative restoration tasks, involving implementations for single-image, video, and volumetric fluorescence data. For full-view processing, LargePNet generally holds around 4-fold and 20-fold higher computational efficiency compared to advanced convolution-based and Transformer-based networks, respectively. The assistance of LargePNet helps achieve 30-hour-long fluorescence imaging to monitor cytoskeleton dynamics, and hour-long tri-color super-resolution imaging to investigate organelle interaction, showcasing its advancement in live-cell imaging.

A fundamental challenge in fluorescence microscopy is the defocused background caused by scattering light, optical aberration, or limited axial resolution. Severe defocus backgrounds will submerge the in-focus information and cause artifacts in the following processing. Here, we leverage a priori knowledge about dark channels of biological structures and dual frequency separation to develop a single-frame defocus removal algorithm. It stably improves the signal-to-background ratio and structural similarity index measure of images by approximately 10-fold, and recovers in-focus signal with 85% accuracy, even when the defocus background is 50 times larger than in-focus information. Our Dark-based optical sectioning approach (Dark sectioning) is fully compatible with various microscopy techniques, such as wide-filed microscopy, polarized microscopy, laser-scanning / spinning-disk confocal microscopy, stimulated emission depletion microscopy, lightsheet microscopy, and light-field microscopy. It also complements reconstruction or processing algorithms such as deconvolution, structure illumination microscopy, and super-resolution optical fluctuation imaging.

In recent years, notable progress has been achieved in both the hardware and algorithms of structured illumination microscopy (SIM). Nevertheless, the advancement of 3DSIM has been impeded by challenges arising from the speed and intricacy of polarization modulation. In this study, we introduce a high-speed modulation 3DSIM system, leveraging the polarization maintaining and modulation capabilities of a digital micro-mirror device (DMD) in conjunction with an electro-optic modulator. The DMD-3DSIM system yields a 2-fold enhancement in both lateral (133 nm) and axial (300 nm) resolution compared to wide-field imaging, and can acquire a data set comprising 29 sections of 1024×1024 pixels, with 15 ms exposure time and 6.75 s per volume. The versatility of the DMD-3DSIM approach was exemplified through the imaging of various specimens, including fluorescent beads, nuclear pores, microtubules, actin filaments, and mitochondria within cells, as well as plant and animal tissues. Notably, polarized 3DSIM elucidated the orientation of actin filaments. Furthermore, the implementation of diverse deconvolution algorithms further enhances three-dimensional resolution. The DMD-based 3DSIM system presents a rapid and reliable methodology for investigating biomedical phenomena, boasting capabilities encompassing 3D superresolution, fast temporal resolution, and polarization imaging.

STimulated Emission Depletion (STED) microscopy attains super‐resolution in biological imaging beyond the diffraction limit. Here, we give a concise protocol to construct a dual‐pulse STED setup with one super‐continuum laser. Moreover, a flexible and dismountable Bessel modulation module is introduced for potential 2D‐stack STED imaging. Experiments and notices are introduced in detail, with discussion on some important check‐points for STED, such as detector saturation. Finally, the results validate the system working.