結構光照明超分辨光學顯微成像技術與展望

陳廷愛，陳龍超，李 慧，余 佳，高玉峰，鄭 煒*

(1.中國科學院 深圳先進技術研究院，生物醫學光學與分子影像研究室，廣東 深圳 518055;2.睿芯生命科技(深圳)有限公司，廣東 深圳 518067)

1 Introduction

引 言

Optical microscopy is one of the greatest inventions in human history. Since its “born” in the 17th century, optical microscopy has always played an important role in the study of modern life sciences with its non-destructive and flexible observation methods. However, traditional optical microscopes are limited by Abbe′s diffraction limit[1](about 200 nm), which hinders scientific research on the finer nanoscale science in cells. In order to observe the intracellular molecular structure, localization, and their interactions to further reveal the nature of life phenomena, scientists have successively proposed a series of optical microscopy imaging systems and methods that break the diffraction limit over the past two decades[2-3]. They can be broadly divided into three categories:(1)Stimulated Emission Depletion(STED) based on the reconstruction of point spread function[4-5], which can usually obtain super resolution images with lateral resolution of 20-60 nm, but the loss of optical power up to GW/cm2restricts its use on living cells. At the same time, limited by the laser output wavelength, only specific fluorescent dyes can be imaged, such as Phototube Fluorescent Proteins; (2)Single Molecule Localization Microscopy(SMLM) based on single molecule localization, including Photo-activated Localization Microscopy(PALM)[6-8]and Stochastic Optical Reconstruction Microscopy(STORM)[9-11]. The image resolution of this technology is usually between 10-30 nm. Although this technology does not require high excitation intensity(kW/cm2), it takes nearly 10 000 exposures to the same sample to acquire a super-resolution image. The lower time resolution also restricts its application in the dynamic observation of living cells. In addition, the requirements for fluorescent dyes are relatively high in this type of technology. The selected fluorescent dye must have ideal “excitation-quenching” efficiency, such as Photoswitchable Fluorescent Proteins; (3)Structured Illumination Microscopy(SIM) based on the expansion of optical transfer function[12-14]. The lateral resolution of super-resolution images is usually between 50-120 nm, and the low excitation intensity(W/cm2), the non-specific requirements for fluorescent dyes, and the advantages of fast wide-field microscopy, make SIM technology the most applicable of these three technologies and SIM is currently the most widely used technique for live-cell super-resolution optical microscopy.

光學顯微鏡是人類歷史上最偉大的發明之一，自17世紀“誕生”以來，光學顯微鏡以其無損、靈活的觀察手段在現代生命科學的研究中起到了重要作用。然而傳統的光學顯微鏡會受到阿貝光學衍射極限(Abbe diffraction limit)的限制[1](約200 nm)，阻礙了科學家對細胞內更細小的納米尺度的科學研究。為了觀察細胞內分子結構、定位以及其相互作用，進一步揭示生命現象的本質，在過去的20年中，科學家們相繼提出了一系列突破衍射極限的光學顯微成像系統與方法[2-3]，大體分為三類：(1)基于點擴散函數改造的受激發射損耗顯微技術(Stimulated Emission Depletion,STED)[4-5],它通常可獲得橫向分辨率在20～60 nm的超分辨圖像，但高達GW/cm2的損耗光功率制約了其在活細胞上的應用，同時受激光器輸出波長的限制，只能對特定熒光染料進行成像，例如高光穩定性熒光蛋白(Phtotostabe Fluorescent Proteins)；(2)基于單分子定位的超分辨顯微成像技術(Single Molecule Localization Microscopy,SMLM)，包括光激活定位顯微技術(Photoactivated Localization Microscopy,PALM)[6-8]和隨機光學重構顯微技術(Stochastic Optical Reconstruction Microscopy,STORM)[9-11]，該技術的圖像分辨率通常在10～30 nm之間，雖然該類技術對激發光強的要求不高(kW/cm2)，但是為了采集一幅超分辨圖像需要對同一樣品進行近萬次曝光，較低的時間分辨率同樣制約了其在活細胞動態觀察中的應用，除此之外，該類技術對熒光染料的要求也比較高，所選的熒光染料必須具有理想的“激發-淬滅”效率，例如光開關熒光蛋白(Photoswitchable Fluorescent Proteins)；(3)基于光學傳遞函數擴展的結構光照明顯微技術(Structured Illumination Microscopy,SIM)[12-14]，其超分辨圖像的橫向分辨率通常在50～120 nm之間，較低的激發光強(W/cm2)、對熒光染料的非特異性需求、快速的寬場成像優勢，使得SIM技術是這三類技術中最適合，也是目前在活細胞超分辨光學顯微成像方面應用最多的技術。

In this paper, the history of structured illumination microscopy is systematically reviewed, and the commonalities and differences between the basic principles and methods of super-resolution microscopy with wide-field structured illumination and super-resolution microscopy with point-scanning structured illumination are compared in detail. The two latest technologies of single-photon excitation super-resolution microscopy based on spectral resolution and two-photon excitation super-resolution microscopy combined with adaptive optics have been highlighted. Finally, the development prospect of structured illumination super-resolution imaging technology is briefly discussed.

本文系統回顧了結構光照明顯微技術的發展歷程，詳細對比了寬場結構光照明超分辨顯微成像和點掃描結構光照明超分辨顯微成像的基本原理與實現方法的共性和差異，重點介紹了本課題組開發的基于光譜分辨的單光子激發超分辨顯微鏡和結合自適應光學的雙光子激發超分辨顯微鏡兩大最新技術，最后，我們簡要討論了結構光照明超分辨成像技術的發展方向。

Fig.1 WF-SIM technologies based on coherent illumination and incoherent illumination. WF-SIM technology based on coherent illumination includes two-dimensional structured illumination microscopy(2D-SIM) and three-dimensional structured illumination microscopy(3D-SIM) 圖1 基于相干光照明與非相干光照明的SIM技術。基于相干光照明的SIM技術包括：二維結構光照明顯微鏡(2D-SIM)與三維結構光照明顯微鏡(3D-SIM)

2 Wide-field structured illumination microscopic technology

寬場結構光照明顯微成像技術

Structural illumination microscopy has been presented to scientists in wide field imaging since its invention. It is also called Wide Field Structured Illumination Microscopy(WF-SIM)[15-16](see Fig.1). In 1999, prof. Heintzmann first proposed the use of laterally modulated light to illuminate samples, using modulated illumination light to encode high spatial frequency information that cannot be detected by the original objective lens into a detectable low-frequency image. Once the intensity distribution of the modulating illumination light field and the finally detected low-frequency encoded fringes, in which the sample′s high spatial frequency information is superimposed, are known, the original high-frequency information of the sample can be obtained by calculation. They obtained super-resolution images using fluorescent beads in experiments, and called this technique Lateral Modulated Excitation Microscopy(LMEM)[12]. In 2000, Professor Gustafsson proposed the classical Two-dimensional Structured Illumination Microscopy(2D-SIM) on the basis of Heintzmannetal.[13-14], which is also the most familiar structure today. They obtained about two times higher lateral resolution(～120 nm) than the conventional wide-field microscope through experiments, and applied this technique for the first time to fixed biological samples. The structured illumination microscopy at this time can only increase the resolution in the lateral direction but not in the longitudinal direction. In 2001, Frohn et al. extended this technology to three-dimensional space and proposed a theoretical model to improve three-dimensional full spatial resolution by using three-dimensional structured illumination[17]. However, this scheme was not verified by Gustafssonetal. until 2008. This is the unique Three Dimensional Structured Illumination Microscopy(3D-SIM)[18]. Due to the linear fluorescence excitation characteristics, the resolution of either 2D-SIM or 3D-SIM technology can only be increased by two times but not indefinitely. In 2002, Professor Heintzmann introduced non-linear fluorescence excitation technology into structured illumination. Theoretically, the spectrum space of SIM was further expanded and Saturated Patterned Excitation Microscopy(SPEM) based on nonlinear fluorescence excitation was proposed[19]. In 2005, Gustafssonetal. experimentally verified the feasibility of the SPEM technique and obtained imaging results with a lateral resolution of ～40 nm, which is referred to as Saturated Structured Illumination Microscopy(SSIM)[20]. Due to the requirement of higher excitation light intensity(MW/cm2), the early SPEM/SSIM technology has a very strong non-linear photo-bleaching effect, making it difficult to reconstruct ideal super-resolution images and is not suitable for biological samples. In 2008, Hirvonenetal. proposed the theory of non-linear structured illumination based on photoswitchable fluorescent protein[21], which was subsequently verified by Regoetal. in 2012. In this experiment, non-linear fluorescence excitation was achieved with an excitation light intensity reduction of 6 orders of magnitude, and a lateral resolution of ～50 nm was obtained in cell imaging[22]. From 2D to 3D, from linear to nonlinear, after the theory of SIM technology was perfected, scientists committed to apply SIM technology to live cell biological imaging. In 2008, Schermellehetal. developed the multi-color SIM technology and applied it to the study of nuclear membrane[23]. In 2009, Kneretal. implemented a rapid 2D-SIM imaging technology and applied it to live cell studies. At ～100 nm resolution, 11 super-resolution images per second can be taken[24]. In 2011, Shaoetal. used 3D-SIM technology to achieve rapid 3D live-cell imaging[25]. In 2015, Lietal. proposed a novel non-linear SIM technology combined with pattern activation to obtain a spatial resolution of 62 nm in live cell imaging[26].

結構光照明顯微術自提出以來主要是以寬場成像的方式展現在科學家們面前的，可以稱之為寬場結構光照明顯微成像技術(Wide Field Structured Illumination Microscopy,WF-SIM)[15-16](見圖1)。1999年，Heintzmann教授首次提出了這種采用橫向調制光照明樣品，利用調制照明光將原本物鏡探測不到的高空間頻率信息編碼至可探測到的低頻圖像中，如果知道調制照明光場的強度分布和最終探測到的疊加了樣品高空間頻率信息的低頻編碼條紋后，樣品原本的高頻信息也就可以通過計算的方式獲得。他們利用熒光小球在實驗中獲得了超分辨率圖像，并把這種技術稱為橫向調制激發顯微鏡(Laterally Modulated Excitation Microscopy,LMEM)[12]。2000年，Gustafsson教授在Heintzmann等人的基礎上提出了經典的二維結構光照明顯微成像技術(Two Dimensional Structured Illumination Microscopy,2D-SIM)[13-14]，也是我們現如今最為熟知的結構光照明顯微術，通過實驗他們獲得了較普通寬場顯微鏡提高約兩倍的橫向分辨率(～120 nm)，并首次將這一技術應用于固定處理的生物樣品中。此時的結構光照明顯微術只能在橫向上提高分辨率，不能在縱向上提高分辨率。2001年，Frohn等人把這一技術拓展到三維空間，提出了利用三維結構光照明提高三維全空間分辨率的理論模型[17]，但直至2008年，此方案才由Gustafsson等人實驗驗證，這便是獨特的三維結構光照明顯微成像技術(Three Dimensional Structured Illumination Microscopy,3D-SIM)[18]。受線性熒光激發特性的影響，無論是2D-SIM還是3D-SIM技術的分辨率只能提高兩倍，并不能無限提高。2002年，Heintzmann教授將非線性熒光激發技術引入到結構光照明成像中，從理論上進一步擴展SIM的頻譜空間，提出了基于非線性熒光激發的飽和結構光激發顯微鏡(Saturated Patterned Excitation Microscopy,SPEM)[19]。2005年，Gustafsson等人實驗驗證了SPEM技術的可行性，獲得了橫向分辨率～40 nm的成像結果，并將之稱為飽和結構光照明顯微鏡(Saturated Structured Illumination Microscopy,SSIM)[20]。由于需要較高的激發光強(MW/cm2)，早期的SPEM/SSIM技術存在著非常強的非線性光漂白效應，很難重建出理想的超分辨圖像，也不適用于生物樣品。2008年，Hirvonen等人提出了基于光開關熒光蛋白實現非線性結構光照明的理論[21]，隨后于2012年，由Rego等人實驗驗證，他們在激發光強降低6個數量級的條件下實現了非線性熒光激發，在細胞成像中獲得了～50 nm的橫向分辨率[22]。由二維到三維，由線性到非線性，在SIM技術的理論被完善后，科學家們致力于將SIM技術應用于活細胞生物成像。2008年，Schermelleh等人開發了多色彩的SIM技術并把它應用于細胞核膜的研究[23]。2009年，Kner等人實現了快速的2D-SIM成像技術并應用于活細胞研究，在～100 nm分辨率的情況下，每秒可拍攝11幅超分辨圖像[24]。更進一步，到2011年，Shao等人利用3D-SIM技術實現了快速的三維活細胞成像[25]。2015年，Li等人提出了結合圖案激活的新型非線性SIM技術，在活細胞成像中獲得了62 nm的空間分辨率[26]。

Any optical system can be regarded as a low-pass filter whose spatial frequency bandwidth(space cutoff frequency:kc=2NA/λ) can be determined by the Optical Transfer Function(OTF). Information above this spatial frequency cannot be passed. The SIM technology encodes high spatial frequency structural information in the sample into a low spatial frequency image by spatial frequency mixing to achieve optical imaging beyond the diffraction limit.

任何一個光學系統可以看作為一個低通濾波器，其可通過的空間頻率帶寬(空間截止頻率：kc=2NA/λ)由光學傳遞函數(Optical Transfer Function,OTF)決定，高于這一空間頻率的信息都不可以通過。SIM技術通過空間頻率混合的方式將樣品中高空間頻率的結構信息編碼至低空間頻率的圖像來實現超過衍射極限的光學成像。

WF-SIM technology uses modulated light formed by two or three laser beams interference as sample excitation light(for the sake of simplicity, the following only analyzes the technical principles of 2D-SIM). If the excitation light intensity is weak, due to the linear fluorescence excitation effect, the fluorescence emission intensity is linearly positively correlated with the excitation light intensity, and the fluorescence distribution acquired on the detector image plane can be expressed as[12-13]:

WF-SIM技術采用兩束或三束激光干涉形成的調制光作為樣品激發光(為了簡單明了，以下僅分析2D-SIM的技術原理)。當激發光強較弱時，由于線性熒光激發效應，熒光發射強度與激發光強度呈線性正相關，探測器像面所采集的熒光分布可表示為[12-13]：

i(r)=[s(r)×e(r)]?h(r) ,

(1)

wheree(r)=1+cos(kc×r+φ) is the excitation light distribution function,s(r) is the sample distribution function labeled with fluorescence, andh(r) is the point spread function(PSF) of the system. In order to analyze the effect of the WF-SIM technology on the spatial frequency expansion of the optical system, the above equation is to be Fourier transformed:

式中，e(r)=1+cos(kc×r+φ)是激發光分布函數，s(r)是標記了熒光的樣品分布函數，h(r)是系統的點擴散函數(point spread function,PSF)。為了分析WF-SIM技術對光學系統空間頻率的擴展，需要將上式進行傅里葉變換：

I(k)=[S(k)?E(k)]×H(k)]=

(2)

It can be seen from the above equation that by applying structured illumination to the sample plane with a sinusoidal modulation distribution, the WF-SIM technology moves the high spatial frequency information originally blocked by the optical system into the passband detectable by the optical system OTF, which expands the spatial cut-off frequency of the WF-SIM system tokWF_SIM≤2kc.

可以看出，通過對樣品面施以正弦調制分布的結構光照明，WF-SIM技術將原本被光學系統截止掉的高空間頻率信息k+ke搬移進光學系統OTF可探測的通帶內，使得WF-SIM系統的空間截止頻率擴展為kWF_SIM≤2kc。

When the excitation light intensity increases to a certain degree, the fluorescent molecules enter the saturated excitation state, and the fluorescence emission intensity is nonlinearly positively correlated with the excitation light intensity(see Fig.2). At this time, due to the nonlinear fluorescence excitation effect[19,22]:

當激發光強增加到一定程度時，熒光分子進入飽和激發狀態，熒光發射強度與激發光強度呈非線性正相關(見圖2)，此時由于非線性熒光激發效應[19,22]：

s(r)×F[e(r)]=s(r)×[a0+

a1e(r)+a2e(r)2+…+ane(r)n] ,

(3)

whereF[e(r)] is the non-linear response of the sample to the structured illumination, which contains the high-order frequencies of the structured illumination. These higher-order frequencies shift the structural information of the higher spatial frequency of the sample to the pass band of the optical system, further expanding the spatial frequency of the WF -SIM.

式中，F[e(r)]表示樣品對照明結構光的非線性響應，它包含了照明結構光的高階頻率，這些高階頻率會將樣品更高空間頻率的結構信息搬移到光學系統的通帶內，進一步拓展WF-SIM的空間頻率。

Fig.2 Principle of nonlinear SIM technology[15]. (a)Left graph shows that the fluorescence intensity tends to be saturated with the increase of the illumination intensity. Right graph shows the distribution of the fluorescence signal at different saturation states in spatial domain, which gradually presents high-order harmonic signals. (b)Figure below shows the distribution of high-order harmonic components at different saturation states in Fourier domain, reflecting the emergence and increase of higher order harmonic components. 圖2 非線性SIM技術原理[15]。(a)左圖表示隨著照明強度增加，激發出的熒光強度逐漸趨于飽和；右圖表示不同飽和狀態下激發出的熒光信號在時域空間的分布，逐漸表現出高階諧波信號；(b)下圖表示的是不同飽和狀態下高階諧波分量在頻域空間的分布，體現更高頻分量的出現與增加

According to the different lighting sources, structured light generation devices, and detection signals, the WF-SIM technology can be very different when the specific experimental system is built(see Tab.1). When the illumination light is a coherent laser[13-14,18,22-27,30-34], if incident light is incident on the structured light generating device(e.g., transmission grating, digital micromirror array, phase spatial light modulator), the device of the periodic structure diffracts the incident light into diffracted light including 0 order, ±1 order, ±2 order,etc., and the structured illumination can be formed on the sample plane by mutual interference of diffracted lights of different orders. The 2D-SIM technology obstructs 0-order and other orders, allowing only ±1 orders of two-beam diffracted light to interfere. The 3D-SIM technique is to allow 0-grade, ±1-level three-beam interference(see Fig.1a). When the illumination light is an incoherent light source such as a high-pressure mercury lamp or LED[28-29], according to object-image conjugate relation and low-pass permeability of the optical system, the conjugate image of the structured light generation device, which includes various frequency informations, may theoretically exist only fundamental and first-order harmonic information after passing through the optical system, and project structured light illumination at the sample plane(see Fig.1b).

WF-SIM技術的實現方法根據照明光源、結構光產生裝置以及探測信號的不同，在具體實驗系統構建時會有很大不同(見表1)。當照明光為相干性的激光時[13-14,18,22-27,30-34]，入射光照射到結構光產生裝置(例如，透射光柵、數字微反射鏡陣列、相位空間光調制器)上，該周期結構的裝置會將入射光衍射成為包含0級、±1級、±2級等的衍射光，借由不同級次衍射光的相互干涉便可在樣品面形成結構光照明。2D-SIM技術便是將0級與其他級次遮擋，只讓±1級兩束衍射光干涉。3D-SIM技術是讓0級，±1級三束光干涉(見圖1a)。當照明光為非相干性的光源如高壓汞燈或LED時[28-29]，根據物像共軛關系與光學系統的低通透過性，包含各種頻率信息的結構光產生裝置的共軛像在通過光學系統后理論上可以只有基頻與一階諧波信息存在，在樣品面投射結構光照明(見圖1b)。

Tab.1 Implementation methods of WF-SIM technology

Through the digital reconstruction of multiple original images, WF-SIM technology can realize super-resolution images. The quality of the reconstructed image may be degraded due to fluctuations in illumination intensity, sample drift, sample bleaching, inaccurate capture of the period and direction of the illumination structure, weak modulation contrast of the illumination structure light, and phase difference jitter between the original images. Fortunately, with the help of a good reconstruction algorithm, the posterior estimates of these influencing factors can be accurately extracted from the original image, and relatively restore the high frequency information in the sample with high-fidelity[29,35-58]. In recent years, WF-SIM technology has made great progress both in algorithms and hardware. In addition to providing the same multi-color fluorescence excitation function as confocal technology, the imaging speed of 2D-SIM technology has broken through to 79 frames/s. @16.5 μm2[32]. The latest non-linear SIM technology achieves a spatial resolution of 62 nm@100 W/cm2[26](see Fig.3) with live cell imaging at 20 consecutive time points. In addition, it does not require special sample preparation and can be used with any fluorescent dye. All in all, WF-SIM technology has become the most favorable super-resolution technology for live-cell imaging, and has become more and more popular in the fields of life sciences, biomedicineetc.

WF-SIM技術實現超分辨圖像需要通過多幅原始圖像的數字重建，重建圖像的質量可能因為照明強度的波動、樣品漂移、樣品漂白、照明結構光周期與方向的不準確捕捉、照明結構光的弱調制對比度、各原始圖像之間的相位差抖動等因素變差。幸運的是，借助優良的重建算法可以從原始圖像中準確的提取這些影響因素的后驗估計值，進而比較真實的還原樣品中的高頻信息[29,35-58]。近幾年WF-SIM技術無論是在算法還是硬件上都有了長足的進步，除了可以提供與共聚焦技術一樣的多色熒光激發功能外，2D-SIM技術的成像速度已突破到了79 frame/s@16.5 μm2[32]，最新的非線性SIM技術還在連續20個時間點的活細胞成像中獲得了62 nm的空間分辨率@100 W/cm2[26](見圖3)，再加上它不需要特殊的樣品制備，可以與任何熒光染料一起使用，WF-SIM技術已經成為最適合于活細胞成像的超分辨技術，在生命科學、生物醫學等領域已經越來越普及。

Fig.3 Linear and nonlinear SIM techniques applied to live cell imaging[26] 圖3 線性與非線性SIM技術應用于活細胞成像[26]

3 Point scanning structured illumination microscopy imaging technology

點掃描結構光照明顯微成像技術

WF-SIM technology has many advantages. For example, it has no restrictions on fluorescent dyes, and almost all commonly used dyes can be used for imaging, at the same time, it is a wide-field imaging technology that can simultaneously meet large-scale and high-speed imaging requirements, which greatly facilitates biological research. However, wide field imaging also limits its application on thick tissue samples. The power density of the wide-field excitation light is weak, and its excitation light field is susceptible to tissue scattering and cannot penetrate the tissue surface for three-dimensional imaging. So far, these results have been limited to single(layer) cell imaging and it has not been possible to perform super-resolution imaging of thick tissue. In recent years, there have been new changes in structural illumination microscopy, which is based on spatial frequency mixing to achieve an optical transfer function, forming a technique called Point Scanning Structured Illumination Microscopy(PS-SIM)[59](see Fig.4 and Fig.5). In 2010, Mülleretal. proposed a structured illumination by using an Airy spot focused by an objective lens instead of the grating required for a conventional wide-field SIM, by scanning the Airy disk and recording each of the Airy disk-excited fluorescence images, a super-resolution image was reconstructed. This is called Image Scanning Microscopy(ISM)[60]. However, since this technique requires the collection of 62 500 original images and it takes more than 10 minutes to reconstruct a super-resolution image(10 μm×10 μm), it′s not very practical. In 2012, Yorketal. used the Digital Micromirror Device(DMD) on the basis of the ISM to simultaneously scan multiple focused Airy disks to increase the imaging speed of the ISM to the second level, and for the first time, the imaging depth of super-resolution technology in biological samples is extended to 45 μm, while the resolution can be maintained at 145 nm. This technique is called Multifocal Structured Illumination Microscopy(MSIM)[61]. Afterwards, under the prompt of MSIM technology, Schulzetal. collected and recorded the images of each rotation angle in a Confocal Spinning-Disk Microscope(CSD), the pixel reassignment operation is performed after the images of all the angles are collected, and the super-resolution imaging is also realized[62]. Both the ISM technology and the MSIM technology need to collect a large number of original images first, and then use post-image reconstruction(pixel reassignment[63]) to generate super-resolution images. At this time, the PS-SIM technology does not have the capability of real-time imaging. In addition, post-image reconstruction will inevitably introduce subjective human factors, and it will also cause artifacts in the final image due to improper image acquisition. In 2013, the De Luca group, the Heintzmann group, and the Shroff group almost simultaneously proposed to perform the shifting and scaling operations required for image reconstruction in optical instead of digital. Specifically, the De Luca team and Prof. Heintzmann introduced optical secondary scanning on the basis of ISM technology to achieve pixel reassignment operations in digital processing. De Luca group called this technique RE-scan confocal microscopy(RE-scan)[64], and Heinzmann group called this technique Optical Photon Reassignment microscopy(OPRA)[65]. The Shroff team combined scanning galvanometers with microlens arrays on the basis of MSIM to realize redistribution of signal photons on the image plane. They named the technology as Instant Structure Illumination Microscopy(iSIM)[66]. With this technology, imaging speeds up to 100 frames/s can be achieved in live cell imaging, thus real-time superresolution imaging(video rate imaging) of biological samples is realized.

Fig.4 Principle rinciple and technology of single PS-SIM. (a)Image scanning microscopy (ISM). (b) Optical photon reassignment microscopy(OPRA)/RE-scan confocal microscopy(RE-scan) 圖4 單點掃描結構光照明成像原理與技術。(a)圖像掃描顯微鏡(ISM)，(b)光學光子重定位顯微鏡(OPRA)/二次掃描共聚焦顯微鏡(RE-scan)

Fig.5 Principle and technology of Multi-PS-SIM technology. (a)Multifocal structured illumination microscopy(MSIM); (b)Instant structured illumination microscopy(iSIM) 圖5 多點掃描結構光照明成像原理與技術。(a)多焦點結構光照明顯微鏡(MSIM); (b)瞬時結構光照明顯微鏡(iSIM)

WF-SIM技術有不少的優點，包括：它對熒光染料沒有任何限制，幾乎所有常用的染料都可以用來成像，同時，它是寬場成像技術，可同時滿足大范圍、高速度成像的需求，這都大大方便了生物學研究。不過也正是因為寬場成像限制了其在厚組織樣品上的應用。寬場激發光的功率密度較弱，而且其激發光場易受組織散射的影響，無法穿透組織表面進行三維成像，所以到目前為止，這些成果都局限在單個(層)細胞成像上，還無法對厚組織進行超分辨成像。近幾年，基于空間混頻實現光學傳遞函數擴展的結構光照明顯微術產生了新的變化，形成了我們稱之為點掃描結構光照明顯微成像的技術(Point Scanning Structured Illumination Microscopy,PS-SIM)[59](見圖4和圖5)。2010年，Müller等人提出一種利用經物鏡聚焦形成的艾里斑代替傳統寬場SIM所需的光柵來實現結構化照明，通過掃描艾里斑，并記錄每個艾里斑激發的熒光圖像，進而重建出超分辨圖像的技術，他們將這種技術稱之為圖像掃描顯微鏡(Image Scanning MicroscopyISM)[60]。但由于該技術需要采集6.25萬張原始圖像，花費10 min以上，才能重建出一張超分辨圖像(10 μm×10 μm)，其實用性不是很高。2012年，York等人在ISM的基礎上使用數字微反射鏡陣列(Digital Micromirror Device,DMD)讓多個聚焦艾里斑同時掃描成像，將ISM的成像速度提高到秒級，并首次將超分辨技術在生物樣品中的成像深度拓展到45 μm，分辨率還能維持在145 nm。他們將這種技術稱之為多焦點結構光照明顯微術(Multifocal Structured Illumination Microscopy,MSIM)[61]。而后Schulz等人在MSIM技術的提示下，在轉盤式共聚焦顯微鏡(Confocal Spinning-Disk Microscope,CSD)中，對每個旋轉角度的圖像進行采集記錄，待所有角度的圖像都采集完成后進行像素重定位(pixel reassignment)操作，同樣實現超分辨成像[62]。無論是ISM技術還是MSIM技術都需要先采集大量的原始圖像，再經過后期圖像重建(像素重定位[63])來生成超分辨圖像，此時的PS-SIM技術并不具備實時成像的潛力，除此之外，后期圖像重建勢必會引入主觀人為因素，也會因為圖像采集不當造成最終圖像存在偽影。2013年，De Luca小組、Heintzmann小組與Shroff小組幾乎同時提出在光學上完成數字圖像重建中所需的像素移動和縮放操作。具體實現上，De Luca小組與Heintzmann教授在ISM技術的基礎上引入光學二次掃描來實現數字處理中像素重定位的操作，De Luca小組把這種技術稱之為二次掃描共聚焦顯微鏡(RE-scan confocal microscopy,RE-scan)[64]，Heintzmann小組把這種技術稱之為光學光子重定位顯微鏡(Optical Photon Reassignment microscopy,OPRA)[65]。Shroff小組則是在MSIM的基礎上結合掃描振鏡與微透鏡陣列實現信號光子在像面上的重新分布，他們將該技術命名為瞬時結構光照明超分辨顯微成像技術(Instant Structure Illumination Microscopy,iSIM)[66]，利用該技術，他們在活細胞成像中實現最快可達到100 frames/s的成像速度，真正實現了生物樣品的實時超分辨成像(視頻速度成像)。

Unlike the WF-SIM technology, which uses modulated light with a specific high spatial frequency, the PS-SIM technology uses an excitation point that contains all spatial frequencies within the cutoff frequency of the optical system as the illumination structured light, and the point spread function of the excitation point ishex(r). Similarly, when there is only a linear fluorescence excitation effect, after one of the fluorescent molecules in the sample is excited, the distribution function of the detector image plane can be expressed as:

不同于WF-SIM技術中采用特定高空間頻率的調制結構光照明，PS-SIM技術則采用包含了光學系統截止頻率以內所有空間頻率的激發光斑作為照明結構光，激發光斑的點擴散函數為hex(r)。同樣地，當只存在線性熒光激發效應時，樣品中的一個熒光分子δ(r)被激發后，在探測器像面的分布函數可表示為：

?δ(r) ,

(4)

wherehem(r) is the point spread function of the emitted fluorescence,d(r) is the pixel distribution function of the detector, andbis the pixel pitch. In fact, due to the optical system magnification, the detector′s pixel size is much smaller thanhem(r), i.e.d(r)～δ(r). Perform the Fourier transform on the above formula, we get:

式中，hem(r)是發射熒光的點擴散函數，d(r)是探測器的像素分布函數，b是像素間距。實際上由于光學系統放大倍率的存在，探測器的像素大小遠小于hem(r)，即d(r)～δ(r)。對上式進行傅里葉變換：

?[Hem(k)×exp-i2π(k×nb)]} ,

(5)

Based on the spreading effect of the convolutions of the two functions, it can be seen that the spatial frequency of the image formed by the excitation of a fluorescent molecule can be extended tokex+kem, i.e. PS-SIM technology can theoretically extend the spatial frequency of the optical system tokPS-SIM≤2kc. However, if the fluorescence molecular signals corresponding to each scanning laser spot are directly arranged in two-dimensional images, the resolution of the optical system cannot be improved. This is due to the fact that in the PS-SIM technique, a fluorescent moleculeδ(r) is collected on the detector by the pixels co-optical axis of the fluorescent moleculeδ(r) and the pixels off-optical axis of the fluorescent moleculeδ(r). The signal is mainly from the peak signal ofhex(r)×[hem(r)×d(r-n×b)], that is to say, the effective signal collected by the pixel displaced byn×bdistances from the fluorescent moleculeδ(r) is actually from the pixel displaced byn×b/2 distances from the fluorescent moleculeδ(r). Therefore, before arranging the fluorescence molecular signals in a direct sequence into a two-dimensional image, the fluorescence molecular signal must be pixel reassigned. The distribution function of the fluorescent moleculesδ(r) after reassignment at the detector plane can be expressed as:

?δ(r) .

(6)

Theoretically, PS-SIM technology can also use nonlinear fluorescence excitation effect to achieve higher spatial frequency expansion, but there is no relevant report.

理論上，PS-SIM技術也可以利用非線性熒光激發效應實現更高空間頻率的擴展，但目前并沒有文章報道。

The realization of PS-SIM technology is mainly based on the point scanning illumination microscopy technique, therefore, it can be well compatible with single-photon excitation and two-photon excitation. Depending on the excitation mode, scanning device, and photon reassignment method, PS-SIM technique also has different combinations in hardware system construction(see Tab.2). In single-photon excitation[60-66], PS-SIM technology has higher light collection efficiency, higher imaging resolution, and better image signal-to-noise ratio than confocal technology. However, since PS-SIM technology relies on the area array detector to collect fluorescence signals, each scanning point signal needs to be detected by each pixel in a spatially split manner. The spit signal itself is very weak, plus the “pixel dwell time” of the spot scanning technology, the camera readout time, and the time of multiple acquisitions of the original image, making the imaging speed of the PS-SIM technology extremely slow, and the initial imaging speed of ISM is only 0.001 6 Hz@10 μm2. In addition, ISM need digital pixel reassignment and deconvolution to achieve superresolution images. The quality of the reconstructed image can be affected by the quality of the reconstruction algorithm and any factors in imaging collecting(see Fig.4a). In recent years, the optimization of the algorithm and the innovation of hardware technology have greatly improved the imaging speed and imaging quality of PS-SIM[69-77]. Replacing digital pixel reassignment with optical photon reassignment, so-called OPRA/RE-scan, which enables superresolution imaging in one acquisition process, avoiding the time consumption of multiple acquisitions of the original image, but still limited by the “pixel dwell time” of the point scanning technique(see Fig.4b). Multi-point scanning provides a mode of “parallel excitation” that avoids the “pixel dwell time” limitation of single-point scanning techniques. MSIM scans samples by using sparse two-dimensional illumination patterns generated by high-speed DMDs. Although digital photon reassignment is still used to obtain super-resolution images, the imaging speed is increased to 1 Hz@48 μm2(see Fig.5a). Further combining multifocal scanning with optical photo reassignment, iSIM increases the imaging speed to 100 Hz and is only limited by the camera read time(see Fig.5b).

PS-SIM技術的實現方法主要是基于點掃描成像技術，因此，它可以很好的兼容單光子激發與雙光子激發。依據激發模式、掃描裝置、光子重定位方式的不同，PS-SIM技術在硬件系統構建上也有著不同的組合(見表2)。單光子激發時[60-66]，PS-SIM技術比共聚焦技術具有更高的光收集效率、更高的成像分辨率及更好的圖像信噪比，但由于PS-SIM技術依靠面陣探測器收集熒光信號，每一個掃描點的信號需按空間展開的方式被各像素探測，分散的信號本身就很弱，再加上點掃描技術的“像素停留時間”、相機的讀出時間、多次采集原始圖像的時間，使得PS-SIM技術的成像速度極其緩慢，最初ISM的成像速度只有0.001 6 Hz@10 μm2。除此外，ISM需要借助數字式的光子重定位與減卷積算法實現超分辨圖像，重建算法的優劣與成像過程中的任何因素都會影響重建圖像的質量(見圖4a)。近幾年減卷積算法的優化與硬件技術的革新使得PS-SIM技術在成像速度與成像質量上都有了極大的提升[69-77]。將數字方式的光子重定位換成光學方式的光子重定位，便是OPRA/RE-scan，它們可在一次采集過程中實現超分辨成像，避免了多次采集原始圖像時間上的消耗，但仍然受點掃描技術的“像素停留時間”限制(見圖4b)。多點掃描提供了一種“并行激發”方式，可以避免單點掃描技術的“像素停留時間”限制，MSIM通過采用高速DMD產生的稀疏二維照明圖案來掃描樣品，雖然仍使用數字式光子重定位獲得超分辨圖像，卻使得成像速度提升到了1 Hz@48 μm2(見圖5a)，進一步，將多點掃描與光學方式的光子重定位結合，iSIM將成像速度提升到了100 Hz，僅僅受相機讀出時間的限制(見圖5b)。

Tab.2 Implementation methods of PS-SIM technology

Confocal technology has always been the main tool in live-cell imaging, with a relatively high resolution and a relatively good imaging speed. Single-photon excited PS-SIM technology has surpassed confocal technology after combining multifocal scanning with optical photo reassignment. It not only has super-resolution imaging capability, but also has several times greater imaging speed than existing confocal technology. In addition, it remains the multicolor imaging property of the confocal technology. With multicolor super-resolution imaging, biologists can more accurately capture the interactions between different structures at the same location in living cells. There are two main implementations of multi-color imaging. The first is multi-spectral excitation, which performs multiple detections of the fluorescence signal excited by each spectral line and the images were merged to realize multicolor imaging. MSIM and iSIM used this method. However, this technique requires multiple switching of the laser wavelength in the multicolor imaging process, which inevitably causes spatial misalignment in the image, and is not suitable for fast dynamic imaging such as calcium imaging. The second is monochromatic excitation, which adopts multi-channel detection. Fluorescence signals of different wavelengths are separated by a spectroscopic device (such as a grating, a prism, or a filter group), although separate fluorescent signals can be collected from different detection channels. However, due to the wide spectrum of fluorescent proteins and their cross excitation and emission spectra, this technique inevitably causes the spectral crosstalk of the fused image. We have established a spectrally resolved single-photon excitation super-resolution microscope combining the RE-scan technique and the spectral unmixing principle to achieve multi-color superresolution microscopy under single excitation conditions[78](see Fig.6). It ensures the spectral purity of the multicolor image after fusion. Moreover, since multi-color super-resolution imaging results can be obtained with one imaging, the technique is also very useful in rapid dynamic imaging, such as calcium imaging.

共聚焦技術之所以一直是活細胞成像中的主力工具，除了比較合適的成像速度外就是相對高的分辨率。單光子激發的PS-SIM技術在結合多點掃描與光學方式的光子重定位后已經完全超越共聚焦技術，不僅具有超分辨成像的能力，而且成像速度是現有共聚焦技術的幾倍以上，除此之外還保留了共聚焦技術的多色成像功能。借助多色超分辨成像，生物學家們可以更準確的捕捉到活細胞中同一位置不同結構之間的相互作用。多色成像主要有兩種實現方式，第一種是多譜線激發，對每個譜線激發的熒光信號進行多次探測，將圖像融合實現多色成像，MSIM與iSIM便是采用這種方式。但是這種技術在多色成像過程中需要多次切換激光波長，會不可避免地造成圖像上的空間錯位，而且也不適合于鈣離子等的快速動態成像。第二種是單色激發，多通道探測的方式。通過分光器件(例如光柵、棱鏡或者濾光片組)將不同波長的熒光信號分離，雖然可以從不同的探測通道上收集分離的熒光信號，但由于熒光蛋白寬譜帶且相互交叉的激發與發射譜，這種技術不可避免地會造成融合圖像的光譜串擾。我們結合RE-scan技術與光譜解混(spectral unmixing)原理建立了一種光譜分辨的單光子激發超分辨顯微鏡，實現了單次激發條件下的多色超分辨顯微成像[78](見圖6)，不僅確保了多色融合后圖像的光譜純凈性，而且由于我們是一次成像便可獲得多色超分辨成像結果，該技術在鈣離子等的快速動態成像中有很強的應用價值。

Fig.6 Single-photon excitation superresolution microscopy imaging based on spectral resolution[78]. SYTO 82 and LysoTracker Red respectively label the nuclei(red in the figure) and lysosomes(green in the figure) of bEnd3-type live cells; (a,e) are normal RE-scan super-resolution images; (d,h) are spectrally resolved RE-scan super-resolution images; (b,f) and (c,g) are the nucleus and lysosomes isolated by spectral unmixing; i is the fluorescence spectrum of two dyes 圖6 基于光譜分辨的單光子激發超分辨顯微成像[78]。SYTO 82與LysoTracker Red分別標記了bEnd3型活細胞的細胞核(圖中紅色)與溶酶體(圖中綠色)；(a,e)普通的RE-scan超分辨圖像； (d,h)基于光譜分辨的RE-scan超分辨圖像；(b,f)和(c,g)分別為光譜解混分離出的細胞核和溶酶體；(i)為兩種染料的熒光光譜

The greatest advantage of the PS-SIM technology is its combination with two-photon excitation. This technique expands super-resolution technology, which can only observe cell thickness, into super-resolution technology that can observe tissue thickness[67-68,79]. Whether it is WF-SIM technology, PS-SIM technology based on single photon excitation, or other super-resolution techniques, one of the biggest problems is that the distribution of light field and the intensity of the excitation light are easily impact by the scattering of the tissue. As the depth of imaging increases, the light field gradually deforms and the light intensity gradually decreases. Two-photon excitation uses a longer-wavelength near-infrared laser as the excitation light source, fundamentally reducing the scattering of the excitation light by the tissue. The two-photon excitation-based MSIM and OPRA/RE-scan achieve a lateral resolution of 145 nm and an axial resolution of 400 nm in the living body. The imaging depth exceeds 100 μm and the imaging speed is close to 1 Hz. At present, there is no report on iSIM based on two-photon excitation, but theoretically it can also be as fast as iSIM under single photon excitation, and it can also perform large depth tissue ultra-resolution imaging like the two-photon excitation-based MSIM and OPRA/RE-scan.

PS-SIM技術被提出后最大的優勢便是與雙光子激發相結合，將只能觀察細胞厚度的超分辨技術拓展為可以觀察組織厚度的超分辨技術[67-68,79]。無論是WF-SIM技術，基于單光子激發的PS-SIM技術，還是其他超分辨技術，一個最大的問題就是激發光的光場分布與光強大小容易受組織散射的影響，隨著成像深度的增加，光場逐漸變形，光強逐漸變小。雙光子激發使用較長波長的近紅外激光作為激發光源，從根本上降低了組織對激發光的散射。基于雙光子激發的MSIM與OPRA/RE-scan在活體組織中實現了145 nm的橫向分辨率和400 nm的軸向分辨率，成像深度超過100 μm，成像速度接近1 Hz。目前，基于雙光子激發的iSIM并沒有文章報道，但理論上它也可以像單光子激發下的iSIM一樣快速，像基于雙光子激發的MSIM與OPRA/RE-scan一樣進行大深度的組織超分辨成像。

Fig.7 Two-photon excitation superresolution microscopy combining with adaptive optics[79]. a, b, c, d and e, f are the fluorescence cytoskeleton images taken from two-photon excited super-resolution microscope, two-photon excited super-resolution microscope with adaptive optics and two-photon excited super-resolution microscope with adaptive optics and deconvolution analysis; g-l are respectively enlarged views of corresponding area in figure e; m represents the latral and axial resolutions of the system; n represents the wave front phase diagram before(left)and after(right) the AO correction 圖7 結合自適應光學的雙光子激發超分辨顯微成像[79]。a、b、c、d及e、f分別為普通雙光子激發超分辨顯微鏡、基于自適應光學的雙光子激發超分辨顯微鏡與基于自適應光學的雙光子激發超分辨顯微鏡，并結合圖像減卷積處理后的細胞骨架成像結果；g～l分別為e圖對應區域的放大圖；m表示系統的橫向與縱向分辨率；n表示自適應校正前后的波前相位圖

Although two-photon excitation does increase the depth of penetration of PS-SIM in super-resolution imaging, as the depth of imaging deepens, the shape of the excitation point will still be distorted, when imaging at large depths, the excited fluorescence signal is also more susceptible to scattering. To solve this problem, adaptive optics based on PS-SIM technology is introduced and a two-photon excitation super-resolution microscope based on adaptive optics[79]is proposed(see Fig.7). The system combines both super-resolution optical microscopy imaging capability and large-depth 3D imaging capability enabling the penetration depth of super-resolution imaging to increase to 250 μm, while the lateral resolution still maintains at 176 nm, and the longitudinal resolution at 729 nm. Using this technique, high-resolution 3D imaging research is conducted on cells, nematode embryos and larvae, fruit fly slices, and zebrafish embryos, and the imaging results are far superior to conventional two-photon imaging. Because this technique improves photon utilization efficiency and thus reduces the required laser power, it allows developmental biologist to perform high-resolution, three-dimensional, dynamic observations of the development of nematode embryos in up to one hour of continuous three-dimensional imaging.

雖然使用雙光子激發確實提高了PS-SIM技術在超分辨成像下的穿透深度，但是隨著成像深度的加深，激發光斑的形狀仍然會發生變形，相對的，在大深度成像時，激發出來的熒光信號也更容易受散射影響。為了解決這一問題，我們在PS-SIM技術的基礎上引入了自適應光學，提出了基于自適應光學的雙光子激發超分辨顯微鏡[79](見圖7)。該系統同時具備超分辨光學顯微成像功能和大深度三維成像能力，使光學超分辨成像深度推進至250 μm，橫向分辨率依然能保持在176 nm、縱向分辨率保持在729 nm。利用該技術，我們對細胞、線蟲胚胎及幼蟲、果蠅腦片和斑馬魚胚胎開展了高清晰三維成像研究，成像效果遠優于傳統雙光子成像。由于該技術提高了光子利用效率，從而降低了所需激光功率，可以允許發育學家在長達1個小時的連續三維成像中對線蟲胚胎的發育過程開展高清晰的三維動態觀測。

4 Summary and outlook

總結與展望

Confocal microscopy has always been a necessary tool for scientific researchers in the field of life sciences and biomedicine. In recent years, with the improvement of hardware and software in super-resolution technology, structured illumination super-resolution microscopy has become the most favored technology that can completely replace confocal microscopy. At the live-cell imaging level, WF-SIM technology and single-photon excited PS-SIM technology can provide image information far beyond confocal resolution and imaging speed. In vivo imaging, PS-SIM technology combined with two-photon excitation can provide super-resolution image information at large imaging depth, which is not available with traditional confocal technology and two-photon technology. In addition, for specific applications, the SIM technology can also be integrated with other imaging technologies, such as the integration of SIM technology with light sheet illumination technology in embryonic development, the integration of SIM technology with plasma structure in material chemistry.

無論是在生命科學領域還是生物醫學領域，共聚焦顯微鏡一直是科研工作者的必備工具。近幾年，隨著超分辨技術在硬件與軟件上的完善，結構光照明超分辨顯微鏡已成為目前最被看好的可以完全取代共聚焦顯微鏡的技術。活細胞成像層面，WF-SIM技術和單光子激發的PS-SIM技術可以提供遠超共聚焦分辨率與成像速度的圖像信息。活體成像方面，結合雙光子激發的PS-SIM技術則可以提供大成像深度下的超分辨圖像信息，這是傳統共聚焦技術與雙光子技術都不具備的。除此之外，針對特殊的應用，SIM技術還可以與其他成像技術融合，例如針對胚胎發育學，SIM技術與光片照明技術的融合，針對材料化學，SIM技術與等離子體結構的融合。

After nearly two decades of development, the potential of WF-SIM technology in cell imaging applications has been fully tapped. A researcher with experience in the development of optical instruments can build a set of 2D-SIM or 3D-SIM systems after spending a short period of system construction training and spending less money and effort. A non-optical professional can also use the WF-SIM system to obtain excellent super-resolution images after 1-2 days of sample preparation and system calibration training [57]. In contrast, the PS-SIM technology that has been developed for less than a decade has very large space for development. (1)PS-SIM can not only be well integrated in existing confocal microscopes, but also easy to operate, however, further improvement in resolution is required. One possible implementation method is to combine the photoswitchable fluorescent proteins using nonlinear fluorescence excitation effect. By controling the possibility that the photoswitchable fluorescent proteins is activated with activating light for the first time, and then exciting the excited protein with the excitation light. The resolution of PS-SIM technology can improve again. (2)PS-SIM technology can be perfectly combined with two-photon technology, using strong anti-scattering capability of two-photon in the biological tissue to achieve two-photon super-resolution imaging depth of 250 μm. However, this is still extremely limited compared to the 1 mm imaging depth reported by ordinary two-photon technology. PS-SIM technology can further combine three-photon technology to realize three-photon super-resolution imaging with greater depth of detection. In addition, PS-SIM technology can also be combined with new large-depth imaging technologies such as photoacoustic and OCT to give full play to the advantages of each technology and achieve new breakthroughs in resolution and imaging depth of each technology. (3)Further promote the implementation of iSIM based on two-photon excitation, and further enhance the three-dimensional imaging(volumetric imaging) speed of the microscope under the premise of improving the resolution and detection depth, making this technology play an unprecedented role in the study of neuroscience and immunology.

經過近20年的發展，WF-SIM技術在細胞成像應用的潛力已經被充分挖掘。一位有光學儀器開發經驗的研究人員在經歷短時間的系統構建培訓后花費較少的財力與精力就可以搭建起一套2D-SIM或者3D-SIM系統。一位非光學專業的研究人員也可以在接受1～2天的樣品制備與系統標定培訓后使用WF-SIM系統獲得優秀的超分辨圖像[57]。相比之下，提出至如今不到十年的PS-SIM技術卻還有很大的發展空間。(1)PS-SIM不僅可以很好的融合于現有共聚焦顯微鏡中，而且操作方便，但是需要進一步提高分辨率。一種可能的實現方法是結合光開關熒光蛋白，利用非線性熒光激發效應，通過激活光第一次照射控制光開關熒光蛋白被激活的可能性，再通過激發光激發這些待激發的蛋白就可以實現PS-SIM技術分辨率再提高。(2)PS-SIM技術可以完美的與雙光子技術結合，利用雙光子在生物組織的強抗散射能力實現250 μm成像深度下的雙光子超分辨成像。但這與普通雙光子技術報道的1 mm成像深度相比還是極為有限。PS-SIM技術可進一步結合三光子技術實現更大探測深度的三光子超分辨成像。除此之外，PS-SIM技術還可以與光聲，OCT等新型大深度成像技術結合，充分發揮各技術的優勢，實現各技術分辨率與成像深度的新突破。(3)進一步推進基于雙光子激發的iSIM實現，在提升分辨率與探測深度的前提下，進一步提升顯微鏡的三維成像(體成像)速度，使這一技術在神經科學，免疫學等的研究中發揮前所未有的效用。

參考文獻：

[1] ABBE E. Contributions to the theory of the microscope and that microscopic perception[J].Arch.Microsc.Anat,1873,9:413-468.

[2] HUANG B,BATES M,ZHUANG X. Super-resolution fluorescence microscopy[J].AnnualReviewofBiochemistry,2009,78:993-1016.

[3] SCHERMELLEH L,HEINTZMANN R,LEONHARDT H. A guide to super-resolution fluorescence microscopy[J].TheJournalofCellBiology,2010,190(2):165-175.

[4] HELL S W,WICHMANN J. Breaking the diffraction resolution limit by stimulated emission: stimulated-emission-depletion fluorescence microscopy[J].OpticsLetters,1994,19(11):780-782.

[5] KLAR T A,JAKOBS S,DYBA M,etal.. Fluorescence microscopy with diffraction resolution barrier broken by stimulated emission[J].ProceedingsoftheNationalAcademyofSciences,2000,97(15):8206-8210.

[6] BETZIG E,PATTERSON G H,SOUGRAT R,etal.. Imaging intracellular fluorescent proteins at nanometer resolution[J].Science,2006,313(5793):1642-1645.

[7] HESS S T,GIRIRAJAN T P K,MASON M D. Ultra-high resolution imaging by fluorescence photoactivation localization microscopy[J].BiophysicalJournal,2006,91(11):4258-4272.

[8] SHROFF H,GALBRAITH C G,GALBRAITH J A,etal.. Live-cell photoactivated localization microscopy of nanoscale adhesion dynamics[J].NatureMethods,2008,5(5):417-423.

[9] RUST M J,BATES M,ZHUANG X. Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy(STORM)[J].NatureMethods,2006,3(10):793-795.

[10] BATES M,HUANG B,DEMPSEY G T,etal.. Multicolor super-resolution imaging with photo-switchable fluorescent probes[J].Science,2007,317(5845):1749-1753.

[11] HUANG B,WANG W,BATES M,etal.. Three-dimensional super-resolution imaging by stochastic optical reconstruction microscopy[J].Science,2008,319(5864):810-813.

[12] HEINTZMANN R,CREMER C. Laterally modulated excitation microscopy: improvement of resolution by using a diffraction grating[J].Proc.SPIE,1999,3568:185-196.

[13] GUSTAFSSON M G L,AGARD D A,SEDAT J W. Doubling the lateral resolution of wide-field fluorescence microscopy using structured illumination[J].Proc.SPIE,2000,3919:141-150.

[14] GUSTAFSSON M G L. Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy[J].JournalofMicroscopy,2000,198(2):82-87.

[15] WICKER K. Super-resolution fluorescence microscopy using structured illumination[J].Super-ResolutionMicroscopyTechniquesintheNeurosciences,2014:133-165.

[16] REGO E H,SHAO L. Practical structured illumination microscopy[J].AdvancedFluorescenceMicroscopy:MethodsandProtocols,2015:175-192.

[17] FROHN J T,KNAPP H F,STEMMER A. True optical resolution beyond the Rayleigh limit achieved by standing wave illumination[J].ProceedingsoftheNationalAcademyofSciences,2000,97(13):7232-7236.

[18] GUSTAFSSON M G L,SHAO L,CARLTON P M,etal.. Three-dimensional resolution doubling in wide-field fluorescence microscopy by structured illumination[J].BiophysicalJournal,2008,94(12):4957-4970.

[19] HEINTZMANN R,JOVIN T M,CREMER C. Saturated patterned excitation microscopy-a concept for optical resolution improvement[J].JOSAA,2002,19(8):1599-1609.

[20] GUSTAFSSON M G L. Nonlinear structured-illumination microscopy: wide-field fluorescence imaging with theoretically unlimited resolution[J].ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica,2005,102(37):13081-13086.

[21] HIRVONEN L,MANDULA O,WICKER K,etal.. Structured illumination microscopy using photoswitchable fluorescent proteins[J].Proc.SPIE,2008,6861:68610L.

[22] REGO E H,SHAO L,MACKLIN J J,etal.. Nonlinear structured-illumination microscopy with a photoswitchable protein reveals cellular structures at 50-nm resolution[J].ProceedingsoftheNationalAcademyofSciences,2012,109(3):E135-E143.

[23] SCHERMELLEH L,CARLTON P M,HAASE S,etal.. Subdiffraction multicolor imaging of the nuclear periphery with 3D structured illumination microscopy[J].Science,2008,320(5881):1332-1336.

[24] KNER P,CHHUN B B,GRIFFIS E R,etal.. Super-resolution video microscopy of live cells by structured illumination[J].NatureMethods,2009,6(5):339-342.

[25] SHAO L,KNER P,REGO E H,etal.. Super-resolution 3D microscopy of live whole cells using structured illumination[J].Naturemethods,2011,8(12):1044.

[26] LI D,SHAO L,CHEN B C,etal.. Extended-resolution structured illumination imaging of endocytic and cytoskeletal dynamics[J].Science,2015,349(6251):aab3500.

[27] CHANG B J,CHOU L J,CHANG Y C,etal.. Isotropic image in structured illumination microscopy patterned with a spatial light modulator[J].OpticsExpress,2009,17(17):14710-14721.

[28] DAN D,LEI M,YAO B,etal.. DMD-based LED-illumination Super-resolution and optical sectioning microscopy[J].ScientificReports,2013,3.

[30] F?RSTER R,LU-WALTHER H W,JOST A,etal.. Simple structured illumination microscope setup with high acquisition speed by using a spatial light modulator[J].OpticsExpress,2014,22(17):20663-20677.

[31] LU-WALTHER H W,KIELHORN M,F RSTER R,etal.. fastSIM:a practical implementation of fast structured illumination microscopy[J].MethodsandApplicationsinFluorescence,2015,3(1):014001.

[32] SONG L,LU-WALTHER H W,F RSTER R,etal.. Fast structured illumination microscopy using rolling shutter cameras[J].MeasurementScienceandTechnology,2016,27(5):055401.

[33] YOUNG L J,STR?HL F,KAMINSKI C F. A guide to structured illumination TIRF microscopy at high speed with multiple colors[J].JournalofVisualizedExperiments:JoVE,2016(111).

[34] 文剛,李思黽,楊西斌,等.基于激光干涉的結構光照明超分辨熒光顯微鏡系統[J].光學學報,2017,37(3):32-42.

WEN G,LI S M,YANG X B,etal.. Super-resolution fluorescence microscopy system by structured illumination based on laser interference[J].ChineseJournalofOptics,2017,37(3):32-42.(in Chinese)

[35] SCHAEFER L H,SCHUSTER D,SCHAFFER J. Structured illumination microscopy: artefact analysis and reduction utilizing a parameter optimization approach[J].JournalofMicroscopy,2004,216(2):165-174.

[36] SHROFF S A,FIENUP J R,WILLIAMS D R. Phase-shift estimation in sinusoidally illuminated images for lateral superresolution[J].JOSAA,2009,26(2):413-424.

[37] SHROFF S A,FIENUP J R,WILLIAMS D R. Lateral superresolution using a posteriori phase shift estimation for a moving object: experimental results[J].JOSAA,2010,27(8):1770-1782.

[38] O'HOLLERAN K,SHAW M. Polarization effects on contrast in structured illumination microscopy[J].OpticsLetters,2012,37(22):4603-4605.

[39] WICKER K. Non-iterative determination of pattern phase in structured illumination microscopy using auto-correlations in Fourier space[J].OpticsExpress,2013,21(21):24692-24701.

[40] WICKER K,MANDULA O,BEST G,etal.. Phase optimisation for structured illumination microscopy[J].OpticsExpress,2013,21(2):2032-2049.

[41] WICKER K. Non-iterative determination of pattern phase in structured illumination microscopy using auto-correlations in Fourier space[J].OpticsExpress,2013,21(21):24692-24701.

[42] AYUK R,GIOVANNINI H,JOST A,etal.. Structured illumination fluorescence microscopy with distorted excitations using a filtered blind-SIM algorithm[J].OpticsLetters,2013,38(22):4723-4726.

[43] O'HOLLERAN K,SHAW M. Optimized approaches for optical sectioning and resolution enhancement in 2D structured illumination microscopy[J].BiomedicalOpticsExpress,2014,5(8):2580-2590.

[44] CHU K,MCMILLAN P J,SMITH Z J,etal.. Image reconstruction for structured-illumination microscopy with low signal level[J].OpticsExpress,2014,22(7):8687-8702.

[45] CHAKROVA N,HEINTZMANN R,RIEGER B,etal.. Studying different illumination patterns for resolution improvement in fluorescence microscopy[J].OpticsExpress,2015,23(24):31367-31383.

[47] BALL G,DEMMERLE J,KAUFMANN R,etal.. SIMcheck: a toolbox for successful super-resolution structured illumination microscopy[J].ScientificReports,2015,5.

[48] F?RSTER R,WICKER K,MüLLER W,etal.. Motion artefact detection in structured illumination microscopy for live cell imaging[J].OpticsExpress,2016,24(19):22121-22134.

[49] ZHOU X,LEI M,DAN D,etal.. Image recombination transform algorithm for superresolution structured illumination microscopy[J].JournalofBiomedicalOptics,2016,21(9):096009-096009.

[50] CHAKROVA N,RIEGER B,STALLINGA S. Deconvolution methods for structured illumination microscopy[J].JOSAA,2016,33(7):B12-B20.

[51] PEREZ V,CHANG B J,STELZER E H K. Optimal 2D-SIM reconstruction by two filtering steps with Richardson-Lucy deconvolution[J].ScientificReports,2016,6:37149.

[52] YANG Q,CAO L,ZHANG H,etal.. Method of lateral image reconstruction in structured illumination microscopy with super resolution[J].JournalofInnovativeOpticalHealthSciences,2016,9(3):1630002.

[53] LAL A,SHAN C,XI P. Structured illumination microscopy image reconstruction algorithm[J].IEEEJournalofSelectedTopicsinQuantumElectronics,2016,22(4):50-63.

[54] MüLLER M,M?NKEM?LLER V,HENNIG S,etal.. Open-source image reconstruction of super-resolution structured illumination microscopy data in ImageJ[J].NatureCommunications,2016,7.

[55] 周興,但旦,千佳,等.結構光照明顯微中的超分辨圖像重建研究[J].光學學報,2017,37(3):10-21.

ZHOU X,DAN D,QIAN J,etal.. Super-resolution reconstruction theory in structured illumination microscopy[J].ChineseJournalofOptics,2017,37(3):10-21.(in Chinese)

[56] 趙天宇,周興,但旦,等.結構光照明顯微中的偏振控制[J].物理學報,2017,66(14):295-305.

ZHAO T Y,ZHOU X,DAN D,etal.. Polarization control methods in structured illumination microscopy[J].ChineseJournalofPhysics,2017,66(14):295-305.(in Chinese)

[57] DEMMERLE J,INNOCENT C,NORTH A J,etal.. Strategic and practical guidelines for successful structured illumination microscopy[J].Nat.Protoc,2017.

[58] KRAUS F,MIRON E,DEMMERLE J,etal.. Quantitative 3D structured illumination microscopy of nuclear structures[J].NatureProtocols,2017,12(5):1011-1028.

[59] STR?HL F,KAMINSKI C F. Frontiers in structured illumination microscopy[J].Optica,2016,3(6):667-677.

[60] MüLLER C B,ENDERLEIN J. Image scanning microscopy[J].PhysicalReviewLetters,2010,104(19):198101.

[61] YORK A G,PAREKH S H,DALLENOGARE D,etal.. Resolution doubling in live, multicellular organisms via multifocal structured illumination microscopy[J].NatureMethods,2012,9(7):749-754.

[62] SCHULZ O,PIEPER C,CLEVER M,etal.. Resolution doubling in fluorescence microscopy with confocal spinning-disk image scanning microscopy[J].ProceedingsoftheNationalAcademyofSciences,2013,110(52):21000-21005.

[63] SHEPPARD C J R,MEHTA S B,HEINTZMANN R. Superresolution by image scanning microscopy using pixel reassignment[J].OpticsLetters,2013,38(15):2889-2892.

[64] DE LUCA G M R,BREEDIJK R M P,BRANDT R A J,etal.. Re-scan confocal microscopy: scanning twice for better resolution[J].BiomedicalOpticsExpress,2013,4(11):2644-2656.

[65] ROTH S,SHEPPARD C J R,WICKER K,etal.. Optical photon reassignment microscopy(OPRA)[J].OpticalNanoscopy,2013,2(1):5.

[66] YORK A G,CHANDRIS P,DALLENOGARE D,etal.. Instant super-resolution imaging in live cells and embryos via analog image processing[J].NatureMethods,2013,10(11):1122-1126.

[67] INGARAMO M,YORK A G,WAWRZUSIN P,etal.. Two-photon excitation improves multifocal structured illumination microscopy in thick scattering tissue[J].ProceedingsoftheNationalAcademyofSciences,2014,111(14):5254-5259.

[68] WINTER P W,YORK A G,DALLENOGARE D,etal.. Two-photon instant structured illumination microscopy improves the depth penetration of super-resolution imaging in thick scattering samples[J].Optica,2014,1(3):181-191.

[69] WEISSHART K. The basic principle of airyscanning[J].ZeissTechnologyNote,2014:22.

[70] STR?HL F,KAMINSKI C F. A joint Richardson-Lucy deconvolution algorithm for the reconstruction of multifocal structured illumination microscopy data[J].MethodsandApplicationsinFluorescence,2015,3(1):014002.

[71] AZUMA T,KEI T. Super-resolution spinning-disk confocal microscopy using optical photon reassignment[J].OpticsExpress,2015,23(11):15003-15011.

[72] CURD A,CLEASBY A,MAKOWSKA K,etal.. Construction of an instant structured illumination microscope[J].Methods,2015,88:37-47.

[73] MCGREGOR J E,MITCHELL C A,HARTELL N A. Post-processing strategies in image scanning microscopy[J].Methods,2015,88:28-36.

[74] SIVAGURU M,URBAN M A,FRIED G,etal.. Comparative performance of airyscan and structured illumination superresolution microscopy in the study of the surface texture and 3D shape of pollen[J].MicroscopyResearchandTechnique,2016.

[75] ROTH S,HEINTZMANN R. Optical photon reassignment with increased axial resolution by structured illumination[J].MethodsandApplicationsinFluorescence,2016,4(4):045005.

[76] SHEPPARD C J R,ROTH S,HEINTZMANN R,etal.. Interpretation of the optical transfer function: Significance for image scanning microscopy[J].OpticsExpress,2016,24(24):27280-27287.

[77] DE LUCA G M R,DESCLOS E,BREEDIJK R M P,etal.. Configurations of the Re-scan Confocal Microscope(RCM) for biomedical applications[J].JournalofMicroscopy,2017,266(2):166-177.

[78] CHEN L C,ZHENG W,etal.. Multicolor re-scan super-resolution imaging of live cells(unpublished).

[79] ZHENG W,WU Y,WINTER P,etal.. Adaptive optics improves multiphoton super-resolution imaging[J].NatureMethods,2017,14(9):869-872.