Confocal vs. Widefield Microscopy Comparison

Research will be limited to:
- Method: Confocal Microscopy
- Method: Widefield Microscopy
- Property: Resolution
- Property: Optical Sectioning
- Property: Photobleaching Speed

Strategy: I will conduct a comparative analysis of confocal and widefield fluorescence microscopy by focusing on the fundamental physical trade-offs between lateral/axial resolution, the mechanism of optical sectioning (or lack thereof), and the kinetics of photobleaching. The search will target quantitative benchmarks and theoretical limits for both systems to provide a rigorous PI-level comparison.
Queries:
- confocal vs widefield microscopy resolution optical sectioning comparison
- photobleaching rates confocal vs widefield fluorescence microscopy
- axial resolution limits widefield vs laser scanning confocal microscopy
- point spread function comparison widefield and confocal microscopy
- signal-to-noise ratio and phototoxicity widefield vs confocal microscopy

Executing 5 concurrent searches:
- confocal vs widefield microscopy resolution optical sectioning comparison
- photobleaching rates confocal vs widefield fluorescence microscopy
- axial resolution limits widefield vs laser scanning confocal microscopy
- point spread function comparison widefield and confocal microscopy
- signal-to-noise ratio and phototoxicity widefield vs confocal microscopy

Found 6 papers. Key excerpts:

Confocal imaging capacity on a widefield microscope using a spatial light modulator — Yao L. Wang et al., 2020

IntroductionWidefield epifluorescence microscopes have driven numerous advances in the biological sciences and are ubiquitous in laboratories. Despite their powerful capabilities, broad accessibility, and relatively low cost, widefield microscopes cannot exclude out-of-focus or scattered light. In sparsely-populated or sparsely-labelled samples, this weakness has relatively minor impact: The illumination light is focused onto the plane being observed, so out-of-focus objects are illuminated by a lower intensity of light and in-focus objects are more likely to dominate images. The light from these out-of-focus objects, however, is not excluded. It remains diffusely in the image and interferes with imaging. This weakness has spurred the development of scanning techniques such as confocal microscopy, which can reject both out-of-focus and scattered light [1]. The key component in confocal microscopes is a pinhole in the emission path, which excludes out-of-focus light. Point scanning, in combination with the pinhole, also excludes scattered light as only imaging light from the excited location is included in the final image. Exclusion of unwanted light allows confocal to have significantly better resolution and optical sectioning ability than widefield microscopy. The disadvantages of confocal microscopy are greatly increased complexity, moving parts, requirement of synchronization, and consequently, increased cost.In our study, we developed simple and inexpensive methods to reduce and exclude scattered and out-of-focus light in a standard widefield epifluorescence microscope. In our approach, we spatially modulate the illumination light and postprocess captured images. Our technique capitalizes on pixelated arrays, such as the spatial light modulator (SLM) and the digital micromirror device (DMD), which modulate the intensity of transmitted and/or reflected light. A transmissive SLM between two crossed polarizers selectively transmits varying intensities of light through each array element. Many projectors use these SLM devices to display an image using incoherent light from a projector bulb. As previously reviewed [2], SLMs and DMDs at the field or aperture stop can control the spatial distribution of the excitation light [3], select the sample location that is observed [4], exclude out-of-focus light [5, 6], and perform structured illumination [7]. Pixelated arrays represent a flexible and cost-effective approach [8] to spatially modulate a light distribution for multiple

applications.Extending the efforts of prior studies, here we show that a transmissive SLM can selectively illuminate locations in the sample with < 500 nm resolution when placed at the field stop of an epifluorescence microscope. Together with minimal postprocessing, we demonstrate optical sectioning (i.e., confocal) capability on a typical widefield microscope using an inexpensive SLM add-on. We characterize images taken on our setup and compare them to images from widefield and confocal microscopes. We demonstrate significantly clearer in vivo imaging of neurons in C. elegans compared to typical widefield microscopy.
ResultsBasic configuration and conceptThe modern widefield epifluorescence microscope (see Fig. 1a) is centered around an objective that focuses Köhler illumination light onto a sample and captures emission light from the sample. The emission light is imaged by a tube lens, often onto a camera array. Many widefield microscopes have additional optics in the excitation beampath, represented by two lenses in Fig. 1a, that shape and optimize the excitation light. These optics create two planes, the field and aperture (not shown) stops, where masks can be inserted to control the extent and angle of the excitation, respectively.Download figureOpen in new tabFigure 1. Optics and light beampaths for microscope setups.(a) Widefield illumination optics create field stop plane, whose light distribution is projected to focal plane in sample via objective. Camera images emission light from focal plane. (b) Introduction of SLM at field stop allows arbitrary illumination patterns at focal plane. (c) Confocal microscope optics scan illumination beam and utilize pinhole to exclude out-of-focus emission light. (d) Widefield axial illumination profile. Wide intensity peak around focal plane. (e) Scanned axial illumination profile. Narrow intensity peak around focal plane.In contrast to the simplicity of widefield microscopes, confocal microscopes (simplified diagram in Fig. 1c) utilize complex sets of optics in both the excitation and emission beampaths. The excitation optics sweep the excitation light across the sample to image each location in turn. The optics depend on the type of confocal microscope: Laser excitation is typically scanned with mirrors [9]. Spinning disk excitation illuminates multiple, distant points in the sample simultaneously by passing illumination light through a Nipkow disk containing multiple holes [10]. In

general, the emission optics include the key optic that distinguishes confocal from widefield microscopy: a pinhole positioned at a conjugate focal plane in the emission beampath. The pinhole strongly filters out light that does not originate from the focal plane, allowing clear imaging of a two-dimensional slice in the bulk without stray light from other depths (e.g., optical sectioning). Because the emission light comes from different locations in the sample, it can exit the objective at various angles and must be “descanned”. Practically, emission light is typically counterpropagated through the same optics used to scan the excitation beam (not shown in figure for simplicity). It is then spectrally separated by a dichroic mirror and filtered by the pinhole prior to detection by a photomultiplier tube (PMT) for laser scanning confocal or by a camera array for spinning disk confocal.In our setup, the SLM is inserted into the field stop position, so that the transmitted light distribution is projected at the focal plane in the sample (see Fig. 1b). The key functional difference between widefield and SLM scanned illumination is that SLM scanning concentrates the illumination to the focal plane. The underlying mechanism is shown in Fig. 1de. The different beam profiles of the widefield and SLM scanned configurations leads to different intensity profiles in z. The intensity is I = P / A = P / πr2, where I, P, A, and r are the optical intensity, optical power, beam cross-sectional area, and beam radius, respectively. The widefield illumination beam is purposely diffuse so that it illuminates the entire field of view at the focal plane. Because the focal area remains large at the focal plane (dotted line), the radius changes relatively little near the focal plane, and the intensity profile in the axial direction is relatively wide. Conversely, the SLM-scanned beam can illuminate a small region of the focal plane (corresponding to transmission through a single SLM element). Near the focal plane, the radius comes to a sharp minimum. Thus, assuming the same numerical aperture as the widefield configuration, the intensity profile in the axial direction is relatively narrow. By concentrating the illumination at the focal plane, we reduce the emission light from out-of-focus objects and enhance the ability of the microscope to optically section the sample, even without true conf

Fast widefield imaging of neuronal structure and function with optical sectioning in vivo — Ziwei Li et al., 2020

INTRODUCTION The nervous system is composed of a complex collection of cells that communicate via electrical and chemical signals and function as networks to process information ( 1 ). To understand their structure and function physiologically, it is essential to study them in living animals. Noninvasive and capable of resolving subcellular structures, optical microscopy has been extensively applied in the field of neuroscience. For in vivo imaging, the most popular methods are confocal fluorescence microscopy and multiphoton fluorescence microscopy, both of which are capable of optically sectioning three-dimensional (3D) samples and extracting information from the focal plane of the microscope objective ( 2 – 5 ). In contrast, the application of standard widefield fluorescence microscopy, in which the entire sample is illuminated and the emitted fluorescence is collected by an objective lens and imaged with a camera, is usually confined to in vitro samples such as cultured cells or thin tissue sections. This is because when a widefield microscope is used to image thick samples, emitted fluorescence photons from both in-focus and out-of-focus structures arrive at the imaging camera. Obscuring the in-focus information, the out-of-focus background makes it difficult to visualize subcellular structures such as synapses. In other words, despite its simplicity in hardware implementation and fast imaging speed, the lack of optical sectioning of standard widefield fluorescence microscopy limits its application in complex 3D samples such as the brain in vivo. One approach that imparts optical sectioning capability to widefield fluorescence microscopy uses structured illumination (SI). When a sample is illuminated with a structured pattern through the microscope objective, the high-spatial frequency illumination contrast is only preserved at the focal plane and, thus, modulates the in-focus signal, while the out-of-focus signal remains unmodified. Reconstruction algorithms take advantage of this difference in signal modulation to reject out-of-focus signal and retrieve in-focus information, thus enabling optical sectioning of 3D samples. One popular implementation of this optical-sectioning SI microscopy (OS-SIM) was proposed by Neil et al. ( 6 ), where three images with phase-shifted sinusoidal illumination patterns are used to computationally reconstruct an optically sectioned image. Powerful in practice, relatively low cost when compared with confocal and multiphoton fluorescence microscopy, and commercially available (e.g., ApoTome,

Zeiss), OS-SIM is nevertheless rarely used for in vivo imaging because of several factors: the often low signal-to-noise ratio (SNR) of in vivo samples, the motion-induced artifacts, and the sample-induced wavefront distortion of the image-forming light. Without addressing these issues, the reconstructed images are prone to low contrast, reduced resolution, and severe artifacts. Here, we describe an optimized OS-SIM method for in vivo imaging. We developed a new reconstruction algorithm that suppresses noise and corrects for sample motion. We further combined adaptive optics (AO) ( 7 ) with OS-SIM to measure and correct sample-induced aberrations. Applying our refined OS-SIM method to live mouse, zebrafish, and Drosophila larvae, we demonstrated high-speed, synapse-resolving imaging of neuronal structure and function in vivo. RESULTS A refined OS-SIM method incorporating AO The original OS-SIM method ( 6 ), hereby referred to as “basic OS-SIM” takes three images I 0 , I 1 , and I 2 with sinusoidal illumination patterns ( Fig. 1 , inset 1; see fig. S1 for detailed optical layout) of the same orientation but equally spaced phases 0°, 120°, and 240°. It reconstructs an OS-SIM image using Eq. (1) I basic   SIM = ( I 0 − I 1 ) 2 + ( I 1 − I 2 ) 2 + ( I 2 − I 0 ) 2 (1) Fig. 1 Simplified AO OS-SIM schematic diagram. Blue dashed polygon, OS-SIM module; red dashed polygon, AO module. Inset 1: Two-beam interference generates patterned illumination. Inset 2: Direct wavefront measurement with a Shack-Hartmann (SH) sensor composed of a lenslet array and a camera. Inset 3: A segmented deformable mirror corrects aberration by controlling the piston, tip, and tilt positions of each segment. SLM, spatial light modulator; Di, dichroic mirror; OBJ, objective; DM, deformable mirror; MM, movable mirror. See fig. S1 for detailed optical paths. In this “square law detection” method, out-of-focus (unmodulated) signal is discarded by the pairwise subtractions, and

the effects of the nonuniform illumination patterns are removed by summing the in-focus signals. Straightforward to implement, this method nevertheless has several limitations, as shown in an example dataset on a fixed cortical tissue section from the brain of a Thy1 –GFP (green fluorescent protein) line M mouse ( 8 ) (raw and reconstructed images using basic SIM; Fig. 2, A and B ). First, the squared subtractions in Eq. (1) make fluctuating noise all positive (“positive bias”) and decrease SNR of the reconstructed images ( 9 ). This particularly affects images of fine structures, because the attenuation of optical transfer function (OTF) amplitude at high spatial frequency makes their signal more sensitive to the presence of noise. Fig. 2 Refined OS-SIM algorithm. Example dataset from fixed mouse brain slice ( Thy1 -GFP line M). ( A ) Three raw images taken at a depth of 25 μm with spatially shifted illumination patterns. ( B ) Basic SIM image reconstructed with Eq. (1) . ( C ) Low-pass (LP)–filtered basic SIM image. ( D ) Widefield image calculated with Eq. (2) . ( E ) High-pass (HP)–filtered widefield image. ( F ) Refined SIM image from Eq. (3) , calculated as a weighted summation of LP( I basic SIM ) and HP( I WF ). ( G ) Comparison of basic SIM and refined SIM reconstructions. (A to F) Images of a single optical section; (G) maximum intensity projection over 8 μm. Scale bars, 2 μm (A to F) and 5 μm (G). Inspired by HiLo microscopy ( 10 – 12 ), in our refined OS-SIM method, to suppress high-frequency noise while maintaining optical sectioning, the basic OS-SIM image was low-pass filtered with LP σ = exp[−( k x 2 + k y 2 )/2σ 2 ], where k x and k y are spatial frequencies, and σ is the standard deviation (SD) of the Gaussian function, defined as the crossover frequency ( Fig. 2C ). To restore high-spatial frequency information to the image, we calculated the widefield image I WF as the average of the three raw images

The TriScan: fast and sensitive 3D confocal fluorescence imaging using a simple optical design — Robin Van den Eynde et al., 2023

1 IntroductionFluorescence microscopy is a technique of choice in the life and material sciences. More and more applications require the visualization of optically-large samples such as tissues and tissue models in a way that is compatible with high-throughput and high-content operation. Confocal microscopy is a technique of choice for 3D imaging, though it is intrinsically slow because the fluorescence is collected only from a tiny diffraction-limited volume that must be scanned across the sample (figure 1).Download figureOpen in new tabFigure 1: Comparison of classic fluorescence microscopy techniques.Widefield systems are very fast because they illuminate the full field-of-view at once, but provide poor optical sectioning. Confocal systems provide good sectioning but are slow because they must scan a small focus volume over the entire sample. Line-scan confocal systems provide similar sectioning but are much faster since one of the scan directions is eliminated.Several strategies have been developed to deliver faster high-performance 3D imaging. Spinning-disk confocal parallelizes the imaging process by simultaneously illuminating the sample through many pinholes at once, though the resulting systems are optically and mechanically complex, suffer from issues such as pinhole cross-talk and low light coupling efficiencies, and are difficult to combine with more advanced features such as adjustable spectral detection. Light-sheet microscopy delivers much faster imaging by combining widefield detection with illumination of the sample using an orthogonal ‘sheet’ of light, though this requires the introduction of a second objective which limits the numerical aperture of the imaging and also complicates the use of common sample formats such as multiwell plates. These restrictions can be mitigated by introducing mirrors into custom sample holders [1] or via the introduction of unconventional imaging geometries or custom optical elements [2–5], but at a cost in complexity and hardware that can reduce versatility, performance, and ease of interpretation, though with the advantage of a reduced phototoxicity. These sophisticated instruments enable highly fascinating experiments, but are less suited to routine ‘workhorse’ measurements such as diagnostic imaging or screening, where straightforward operation and a lower cost are important.An alternative approach for fast 3D imaging combines the confocal principle with different geometries for the illumination and optical aperature. In line-scan confocal microscopy, the excitation light is focused into a line rather than a spot and the circular pinhole is replaced with a slit aperture [

6]. In this way, a full line can be acquired from the sample at once, providing a dramatic speedup since it eliminates an entire scan axis. A 512 by 512 pixel acquisition, for example, can be sped up by a factor 512 in principle. The use of a slit aperture does have consequences for the imaging performance, in principle leading to an asymmetric PSF and reduced z-sectioning, though these effects are small [7] since typical confocal microscopes are not operated with infinitesimally small pinholes. A number of such instruments have already been developed, including the use of line detectors [8]. Alternatively, the fluorescence line can be rescanned over standard 2D cameras [9] that provide much higher detection effiencies. However, these approaches have relied on the use of two synchronized scanning mirrors to perform the necessary scanning-descanning of the light over the sample and subsequent rescanning over the detector. This limits the scanning mirrors to slower variants, reducing the imaging speed while also imposing synchronization overhead in the hardware and measurement control software. Such slow scanning speeds are also problematic when fast stochastic events must be sampled, as is the case in techniques such as single-molecule localization microscopy (SMLM) or single-particle tracking (SPT).We reasoned that this design could be simplified and the performance increased by adopting an all-optical synchronization that permits the use of a single, very fast scanning mirror. Intriguingly, such an approach was already proposed in the very initial publication describing confocal line-scan imaging [6], though in a format that is non-straightforward to build and operate. In this contribution, we describe the ‘TriScan’, a line-scanning confocal that can combine the speed, sensitivity, and convenience of widefield imaging with the optical sectioning of a one-photon confocal, using a comparatively simple optical layout. By delivering a strong imaging performance within a self-contained and inexpensive format, the TriScan offers exciting opportunities as a workhorse system for the fast imaging of 3D samples across a variety of use cases.
2 Results and DiscussionA conceptual scheme of our TriScan is shown in figure 2. The essence of the design is to combine a single scanning mirror with a threefold passage of the light over this mirror. The first passage scans the excitation line over the sample, the second passage descans the emission so that

it passes through a slit aperture, and the third passage scans the light over the camera. Figure 3 shows the concrete optical design in which this triple passage is realized by introducing a dichroic mirror that selectively transmits the emission and causes it to be reflected by a broadband mirror. Such an arrangement does attenuate the emission somewhat by requiring a double pass of the emission through the dichroic mirror. This can be avoided by removing this dichroic mirror and replacing the back mirror with a dichroic mirror that can reflect the fluorescence for incident angles close to zero degrees, whereupon the excitation light is introduced such that it is transmitted by this dichroic.Download figureOpen in new tabFigure 2: Conceptual scheme of the TriScan microscope.Excitation light is shaped into a line and scanned over the sample. The resulting emission is descanned and reflected by a broadband mirror positioned at a slight angle, after which is scanned again over the camera, forming an image of the sample. The optical synchronization inherent in the design allows the use of very fast scanning mirrors and off-the-shelf, highly sensitive cameras.Download figureOpen in new tabFigure 3: Optical layout using off-the-shelf components.The excitation and emission light paths are shown separately for increased clarity. The two colors for the emission distinguish the light coming from the objective from the light travelling towards the camera. BM: back mirror, DM: dichroic mirror, AD: aperture disk, RAM: right angle mirror, L0-5: lenses, TL: tube lens, CL: cylindrical lens.The use of three passes over a single scanning mirror is a crucial element of our design, because it ensures the synchronization of the scanning process without requiring additional hardware or software. Furthermore, it allows the use of very fast scanning mirrors even if these do not allow full control over the mirror position or other scan parameters. This includes, for example, the inexpensive and reliable resonant scanners that offer scanning rates in the kilohertz regime, with up to 16 kHz scanning frequencies readily available. In principle, this would allow our TriScan to deliver full field-of-view images at a frequency of 32 kHz, though these are unlikely to be realized in practice given the currently-available camera technology as well as the limited brightness of typical samples.In actual measurements the camera therefore integrates the emission originating from many light sweeps into a single image. This is again a feature of our design

A phasor-based approach to improve optical sectioning in any confocal microscope with a tunable pinhole. — Morgana D'Amico et al., 2022

Confocal fluorescence microscopy is a very popular and versatile tool in life sciences (Conchello & Lichtman, 2005 ; Jonkman & Brown, 2015 ). Compared to wide‐field fluorescence microscopy, the main advantage of confocal microscopy is represented by its ability to remove the unwanted blur originating from out‐of‐focus planes and generate thin optical sections of the specimen. By scanning along the optical axis, confocal microscopy can provide three‐dimensional reconstructions of biological specimens. An important feature of point‐scanning confocal microscopy is that it can be easily combined with many modern quantitative fluorescence techniques. Confocal‐based fluorescence lifetime imaging (FLIM) is a powerful technique for the detection of forster resonance energy transfer (FRET) (Broussard et al., 2013 ; Day, 2014 ; Giral et al., 2011 ; Pelicci et al., 2019 ) and the imaging of fluorescent sensors (Ferri et al., 2016 ; Scipioni et al., 2021 ). Confocal‐based spectral imaging can be used to separate multiple spectral components (Fereidouni et al., 2012 ; Shi et al., 2020 ) and to analyze the response of environment‐sensitive fluorescent dyes (Malacrida et al., 2016 ; Sediqi et al., 2018 ). Confocal‐based fluorescence recovery after photobleaching (FRAP) and fluorescence correlation spectroscopy (FCS) are well‐established methods to measure the diffusion of molecules inside live cells (Di Bona et al., 2019 ; Fritzsche & Charras, 2015 ; Scipioni et al., 2018 ). Notably, stimulated emission depletion microscopy (STED), one of the most popular super‐resolution imaging techniques, is also based on a confocal microscope architecture (Vicidomini et al., 2018 ).
In confocal microscopy, optical sectioning is provided by the pinhole, a small aperture placed in front of the point detector, whose position is confocal to the illuminated spot in the specimen (Diaspro, 2019 ; Pawley, 2006 ). The size of the pinhole directly determines the degree of optical sectioning. The use of smaller pinhole sizes improves the discrimination of in‐focus light versus out‐of‐focus light. In addition, imaging with smaller pinholes also improves the lateral resolution with a maximum theoretical improvement of a factor of 1.4 for a fully closed pin

hole (Conchello & Lichtman, 2005 ; Diaspro, 2019 ). Unfortunately, the reduction of the pinhole size drastically reduces light throughput at the detector (reduction of about 95% of the intensity from 1 Airy Unit (AU) to 0.2 AU (Huff, 2015 ) so that one has to find a compromise between optical sectioning and signal‐to‐noise ratio (SNR) (Sheppard et al., 2006 ). The reduction of SNR at smaller pinhole sizes makes difficult, in practice, the use of pinhole sizes significantly smaller than 1 AU.
Here, we propose a simple approach to “virtually” perform confocal imaging at smaller pinhole sizes without the dramatic reduction of SNR. The approach is based on the concept of separation of photons by lifetime tuning (SPLIT) (Lanzano et al., 2015 ), a super‐resolution technique originally introduced in the context of lifetime‐resolved STED microscopy. In SPLIT, the increase in spatial resolution is obtained by decoding the spatial information encoded into the lifetime channel of the microscope. In this additional channel, the signal originating from the center of the excitation spot has a different fingerprint (longer lifetime) compared to the signal coming from the periphery of the excitation spot (shorter lifetime). This fingerprint can be exploited to isolate, via phasor analysis, the fraction of the signal originating from the center of the excitation spot and generate a super‐resolved image (Lanzano et al., 2015 ). We have recently demonstrated that the SPLIT approach to super‐resolution is not limited to lifetime‐resolved images [Sarmento 2018] but has a more general applicability. In other words, the same phasor algorithm can be applied to microscopy images containing an additional channel with encoded spatial information. In STED microscopy, this additional channel can be represented by the fluorescence lifetime variations induced by a STED beam (Coto Hernandez et al., 2019 ; Lanzano et al., 2015 ; Lanzano et al., 2017 ; Tortarolo et al., 2019 ; Wang et al., 2018 ) or a tunable depletion power (Pelicci et al., 2020 ; Sarmento et al., 2018 ). In STED microscopy, the application of SPLIT produces images that have higher resolution and better contrast, compared to their counterpart STED images (Cerutt

i et al., 2021 ). Recently, we have shown that SPLIT can be applied also to structured illumination microscopy (SIM), using as the additional channel the illumination pattern translation step (Cainero et al., 2021 ). In confocal microscopy, the tunable pinhole size encodes spatial information in the axial direction. In fact, for decreasing values of the pinhole size, the percentage contribution of the out‐of‐focus intensity to the total intensity decreases, resulting in a different fingerprint between the out‐of‐focus and in‐focus intensity components. Thus, the tunable pinhole size can be used as the additional channel for application of the phasor‐based SPLIT algorithm.
The SPLIT‐pinhole (SPLIT‐PIN) method is based on the sequential acquisition of multiple confocal images acquired with a different pinhole size. The images are processed to obtain a final image with improved optical sectioning (i.e., virtually smaller pinhole size) but preserved SNR level. Notably, our method can be implemented on any confocal microscope equipped with a tunable pinhole size (to the best of our knowledge, most of the commercially available confocal laser scanning microscopes allow the users to tune the size of the pinhole). To evaluate the quality of the images provided by SPLIT‐PIN, we use the recently introduced QuICS algorithm, a tool based on image correlation spectroscopy (Cerutti et al., 2021 ). QuICS allows extracting three parameters related to the resolution, contrast e SNR of the image. We show with simulations that the SPLIT‐PIN image can provide improved optical sectioning (i.e. virtually smaller pinhole size) with respect to confocal images but better SNR with respect to an image obtained with closed pinhole. For instance, two images acquired at 2 AU and 1 AU can be combined to obtain a SPLIT‐PIN image with a virtual pinhole size of 0.2 AU but with better SNR. As an example of application to biological imaging, we apply SPLIT‐PIN to confocal imaging of the apical membrane in an in vitro model of the intestinal epithelium and we find that the SPLIT‐PIN image has a better contrast compared to conventional confocal imaging. In summary, we demonstrate that SPLIT‐PIN is can be a simple and effective tool to boost the optical sectioning power of any confocal microscope.

Rigor and Reproducibility in Confocal Fluorescence Microscopy — James Jonkman, 2019

Fluorescence microscopy can be an excellent tool for making quantitative measurements in cells. For particularly rigorous applications, many choose to use confocal microscopes, which use pinholes to throw away the out-of-focus signal, producing an image of a thin focal plane as shown in Figure 1 1. Confocals have many more degrees of complexity compared to flow cytometers: they share the spectral dimension, but have in addition potentially three spatial dimensions as well as time for live-cell imaging. Nevertheless, very little progress has been made toward ensuring rigor and reproducibility of confocal images. Unlike their cousins the flow cytometers, on which operators regularly flow intensity calibration beads, confocal microscopes suffer from a lack of standard intensity reference samples. Despite the beautiful images, there are therefore many pitfalls to obtaining truly quantitative data using confocal fluorescence microscopy 2, 3.

              Figure 1Open in figure viewerPowerPoint

              (A) Widefield and (B) confocal fluorescence images of a single focal plane in a 50 μm-thick 3D cell culture labeled with a GFP-tagged nuclear protein and an RFP-tagged membrane protein. The confocal microscope uses pinholes to eliminate the out-of-focus fluorescence, producing a thin optical section. Scalebar = 10 μm. [Color figure can be viewed at wileyonlinelibrary.com]




     Many confocal microscope users know the basic steps required for comparing control and experimental conditions in a quantitative manner. First, the samples should all be prepared the same way (perhaps this is obvious to some readers), and preferably all at the same time

(which is less obvious). There are many steps to preparing fixed samples (fixation, permeabilization, antibody labeling, and mounting) and live samples (fluorescent protein or vital dye labeling, incubation, and plating in coverglass-bottom chambers) and tiny changes in the protocols from 1 day to the next can dramatically affect the results. Second, the samples should all be imaged with the same settings (again, this may be obvious), and preferably all in the same imaging session (which is less obvious). A large experiment may take many hours to image, so more than one session may be required; but care must be taken to ensure that the microscope has not changed from one day to the next. For example, the intervening users might leave the objective lens dirty; or the microscope company might service and realign the confocal, resulting in a completely different laser power reaching the specimen. Furthermore, the samples themselves may change over a multiday experiment, with fixed-cell staining beginning to fade and live cells becoming more confluent. The longer the extent of your experiment, the more likely it is that the images taken along the way will have varying intensities. Consider reimaging control slides at the start of each session to help mitigate these issues. Also, ensure that you use exactly the same imaging conditions by saving and loading the settings, or even loading a previous image and reapplying the settings directly from the image.
Saturation and Cross-Talk

     Well-trained confocal microscope users know that they must choose settings that do not result in saturation of their images or cross-talk between the different fluorescence channels 4. Saturation occurs when the fluorescence intensity is higher than the maximum allowed by the detector or camera: since there is no way to know how much higher the signal really is, saturated images cannot be accurately quantified. Although most confocal users know to avoid saturation, what happens when you get half-way through an experiment and find that some cells are too bright with your current settings? Rather than starting over, consider lowering the laser power (which is relatively linear and can also be measured) and retaking just the saturated images. The laser powers used for the weaker and brighter cell conditions can be measured with a power meter placed on the stage, usually through a 10× dry objective, and the percent difference in intensity can then be used to scale the intensity

results accordingly. Cross-talk (or “bleed-through”) occurs with multiply-labeled specimens when fluorescence from one fluorophore is collected by the detector channel corresponding to a different fluorophore. For example, if the fluorophore DAPI is used to label the nuclei, its broad emission spectrum overlaps the green channel (e.g., Alexa Fluor 488) substantially. This is usually not a problem if each fluorophore is excited sequentially, but since confocal microscopes often have multiple lasers and detectors, cross-talk arises when users try to speed up their acquisition by exciting and detecting them simultaneously. Configuration wizards in the software help you to configure the confocal to increase speed while minimizing bleed-through; but these wizards normalize each fluorophore to the same intensity, which is usually far from reasonable. For example, DAPI has minimal overlap with the red channel (e.g., Cy3) if both fluorophores are the same intensity; but if DAPI is twice as bright as Cy3 then the cross-talk is substantial. To be safe, configure your confocal channels for sequential acquisition.
Photobleaching

     All fluorophores exhibit photobleaching (to various degrees), which means that the act of illuminating them causes the fluorescence to fade. For live-cell imaging, photobleaching also results in phototoxicity 5. Many users are aware of the hazards of photobleaching and are careful to keep the laser powers low for confocal imaging. However, users may be unaware of just how much photobleaching occurs while observing the samples through the binoculars to find an interesting field of view. At their maximum setting, widefield fluorescence lamps can bleach a fluorophore to half of its original intensity in just a few seconds 3! Try setting fluorescence lamps to about 10% of their full power and lowering the room lights: once your eyes adjust to the darker settings, you should find that you can still see well enough to choose and focus on the desired cells or structures.

Selection Bias

     One of the greatest barriers to rigorous and reproducible confocal microscope experiments is encountered before an image is even taken: how do you select the field of view

Confocal Laser-Scanning Fluorescence-Lifetime Single-Molecule Localisation Microscopy — Jan Christoph Thiele et al., 2020

IntroductionConfocal Laser-Scanning Microscopy (CLSM) is one of the most important microscopy techniques for biology and medicine. Its fundamental purpose is to provide so-called optical sectioning and thus to enable the recording of three-dimensional images, which is impossible to achieve with conventional wide-field microscopy. Its disadvantage, when compared to wide-field microscopy, is its inherently slow image acquisition speed, because image formation is realised by sequentially scanning single or multiple foci over a sample. This limits also its overall light throughput (small dwell time per scan position), which is one reason why CLSM was nearly never used for single-molecule localisation based super-resolution microscopy (Single Molecule Localisation Microscopy or SMLM), such as Photo-activation Localisation Microscopy (PALM),(1) (direct) Stochastic Optical Reconstruction Microscopy (dSTORM),(2, 3) or Points Accumulation for Imaging in Nanoscale Topography (PAINT).(4, 5) There are only two exceptions,(6, 7) one where a spinning-disk CLSM was employed for PAINT, exploiting the superior out-of-plane light rejection of a CLSM that is so important for reducing background from freely diffusing dyes in PAINT, and one where a spinning-disk CLSM was employed for STORM with self-blinking dyes, where it was used for reducing excitation intensity. Besides efficient out-of-plane signal rejection, which also enhances contrast and facilitates deep-tissue imaging,(8) CLSM offers several additional advantages that should make it attractive for SMLM. Firstly, single-focus CLSM uses single-point detectors which can be operated in single-photon counting mode (Geiger mode) and thus provide shot-noise limited detection, in contrast to emCCD or sCMOS cameras used in conventional wide-field SMLM, with their read-out, thermal, and electronic noise. Secondly, when using Geiger mode for light detection, CLSM records the positions of single-photon detection events in a quasi-continuous, non-pixelated way, thus avoiding pixel size to affect single-molecule localisation accuracy.(9) Thirdly, and most interestingly, it allows for measuring fluorescence lifetimes, thus allowing to combine Fluorescence Life-time Imaging Microsc

opy (FLIM) with SMLM. This offers, for example, the option to co-localise different molecular species that differ only by their lifetime while having similar excitation and emission spectra,(10) thus efficiently circumventing all problems connected with chromatic aberration that trouble so much multi-colour SMLM.(11, 12) Especially for state-of-the-art SMLM, which now routinely achieves a lateral resolution of only a few nanometers, chromatic aberration is a serious issue,(13) in particular when trying to study biological interactions or the relative arrangement of different cellular structures with respect to each other.Several solutions to the chromatic aberration problem have been proposed in the past. Recently, an aberration-free multi-colour method of SMLM called spectral-demixing dSTORM was presented.(14) This method works well for fluorophores showing good switching performance while utilising the same imaging buffer. The fluorescence signal of the different molecules is separated spectrally, and ratiometric fluorescence measurements are used for spectral demixing and (co-)localising different kinds of molecules. One step further in this direction was the implementation of spectrally-resolved SMLM, where full spectra are measured and used for sorting different molecules and their localisations.(15) A very fascinating approach is multi-colour SMLM that combines PSF engineering with deep learning for identifying and sorting different molecular species without the need of spectrally resolved imaging.(16) In frequency-based multiplexing STORM/DNA-PAINT,(17) one uses frequency-encoded multiplexed excitation and colour-blind detection to circumvent chromatic-aberration problems. Another clever solution is Exchange-PAINT,(18) which sequentially images different targets with the same dye but uses different DNA-tags for directing the dye to different targets decorated with complementary DNA-strands. Similarly, barcoding PAINT(19) exploits the different binding kinetics of imager and docking strands for distinguishing between different target sites. Because one uses the same dye for all the different structures, chromatic aberrations do not impact the SMLM results, but the prize is increasing image acquisition time, ca. linearly increasing with the number of different targets one wants to resolve. Finally, the recently introduced MINFLUX(20) allows for super-resolution imaging with unprecedented

accuracy of only a few nanometers and can be used for chromatic-aberration free multi-colour imaging.(21) Similar to the confocal laser-scanning SMLM that will be presented here, it is also based on scanning, but in an asynchronous manner, so that it can currently localise only one individual molecule at any time.In this work, we present the first (to our knowledge) realisation of SMLM with a time-resolved CLSM using single-photon avalanche-diodes (SPADs) for detection, and a rapid laser-scanning unit for excitation beam scanning. This unit enables us to record images with reasonable acquisition speed as required for efficient SMLM. Our approach combines all the advantages of CLSM with those of SMLM: axial sectioning, shot-noise limited single-photon detection, pixel-free continuous position data, and fluorescence lifetime information of CLSM with the exceptional spatial resolution and single-molecule identification of SMLM. We first demonstrate the feasibility of using CLSM for fluorescence lifetime SMLM (FL-SMLM) by imaging labelled fixed cell samples by combining CLSM with two of the most widely used variants of SMLM, dSTORM (for imaging microtubules and clathrin in human mesenchymal stem cells) and DNA-PAINT (for imaging cellular chromatin in COS-7 cells). To demonstrate the fluorescence lifetime multiplexing capability of FL-SMLM, we record images of polymer beads that are surface-labelled with two different dyes and two cellular targets (microtubules and clathrin in COS7 cells). Our results show that confocal laser-scanning FL-SMLM has great potential for many applications, extending the dimensions of fluorescence super-resolution microscopy by fluorescence lifetime.
Results and DiscussionConfocal laser-scanning SMLM measurements were carried out on a custom-built time-resolved confocal microscope equipped with a fast laser scanner, see Fig.1a (for more details see SI). For dSTORM measurements, a region of interest of 10 µm × 10 µm was scanned with an image scan rate of ∼27 Hz. Excitation was done with a pulsed laser at 640 nm wavelength, a repetition rate of 40 MHz, and a pulse width of ∼50 ps. Single