LAUSR

Cells are the fundamental building bricks for all life. Optical technologies have served as the mainstream for single-cell analysis, which may trace back to the 1600s when Robert Hooke reported “cell” in his groundbreaking publication of Micrographia. In the 1900s, novel technologies such as Raman spectroscopy, flow cytometry, and fluorescence microscopy emerge that widely open the fields of cell analysis [1–3]. It is worthy to point out that the invention of laser has a great impact on cell study, which is not only used in novel microscopic technologies such as laser scanning microscopy but also in flow cytometry [3, 4]. The Nobel laureates of chemistry in 2014 for super-resolution (SR) microscopy have pushed single-cell imaging to be far below the optical diffraction limit. Closely watching those quantitative single-cell optical technologies, we are on the track of obtaining better resolution and higher throughput while measuring single cells (Table 1). Flow cytometry measures the physical or chemical properties of single cells in a fluidic stream. It has been widely used in biomedicine as its capability to offer high throughput data of single cells. Dating back to the first generation of flow cytometry, scattered lights are measured for cell analysis [5]. It then comes to the era of multi-fluorescence labeling of cells that can be up to around 20 markers per cell [6]. With the advancement of sensor technology, microfluidic technology, machine learning technology, and so on, the recently developed imaging flow cytometry, label-free cytometry, and in-vivo flow cytometry have shown the great applications and capabilities of quantitative single-cell optical analysis [7–9]. To observe the specific structure and dynamical processes of life activity within living cells more clearly, some SR fluorescence microscopic imaging methods were proposed in the early 21st century, including photoactivated localization microscopy, stochastic optical reconstruction microscopy, stimulated emission depletion microscopy, and structured illumination microscopy (SIM) [10–13] These SR imaging methods extend the resolution limits of microscopy, allowing observation of delicate structures such as the clathrin-coated pits, cytoskeleton of actin and microtubulin, and so forth. The SIM microscopy requires fewer photons among the various SR imaging methods and is more suitable for live-cell imaging. Nevertheless, SIM microscopy is prone to reconstruction artifacts, which affects the identification and quantification of the true structure. Moreover, live-cell SR imaging with SIM still requires strong excitation light that may generate phototoxicity. The Hessian SIM microscopy, invented by optimizing the optical path design in hardware and exploiting the spatiotemporal continuity of biological structures in the algorithm, reduces the phototoxicity of traditional SIM by an order of magnitude [14]. Most of the above mentioned flow cytometry or SR technologies rely on the invention of fluorescence labeling that allows observation of specific organelles and their interactions. Here a review by Wang et al. summarizes the recently developed fluorescent probes or labeling techniques for visualizing biomolecules at the cellular, subcellular or molecular scale by using Caenorhabditis elegans as the major model organism (Wang et al., this issue, p. XXX). Due to the diffraction limit, the resolution of traditional fluorescence microscopes is weaker than 200 nm, hardly to observe the fine structure of organelles, which is usually resolved by electron microscopy (Zhou et al., this issue, p. XXX). However, the SR optical microscopies mentioned above reach the resolution of about 10 nm. The SR microscopies, including SIM, suffer from aberrations in thick sample [15], leading to serious deterioration. Here Zheng et al. report the development of adaptive optics for SIM based on deep learning, attempting to solve this problem (Zheng et al., this issue, p. XXX). Unlike most SR microscopies that can only observe the structure of organelles, fluorescence resonance energy transfer (FRET) microscopy can observe the interactions between molecules or proteins. Yin et al. develop the mTA-G method that could simply and accurately measure the critical factors in FRET (Yin et al., this issue, p. XXX). Label-free analysis of single cells has attracted wide attention from various research fields recently [5, 8, 16, 17]. Compared with conventional fluorescence labeling, label-free analysis is non-invasive, cost-less, and may easily be performed without high demand for operation skills such as fluorescence labeling. In this issue, Su and colleagues have developed a light scattering pattern-specific convolution net cytometer (LSPS-net Cytometer) for label-free analysis of cervical cells by integration of deep learning and 2D light scattering techniques (Liu et al., this issue, p. XXX). In their work, they measure the spatially distributed light scattering from single cells onto a twodimensional (2D) sensor. The information-rich 2D light scattering patterns of label-free cervical cells are analyzed with deep learning for automatic feature extractions, and accuracies of more than 90% are obtained for label-free cervical cells classification in terms of normal and cancerous samples and subtyping of cancerous cell lines. Huang and colleagues have investigated the label-free techniques and the biochip techniques for long-term in-situ imaging of single cells after on-chip capture and isolation (Fu et al., this issue, p. XXX). Motivated by the development of biochip technologies for simultaneous culturing and imaging of single cells, they developed a microfluidic chip with a compact incubator. The optical imaging of single cells was performed with quantitative self-interference spectroscopy (QSS). Their system is capable of long-term culturing of cells on the Received: 16 May 2021 Revised: 17 May 2021 Accepted: 19 May 2021