Light sheet microscopy

Abstract

This chapter introduces the concept of light sheet microscopy along with practical advice on how to design and build such an instrument. Selective plane illumination microscopy is presented as an alternative to confocal microscopy due to several superior features such as high-speed full-frame acquisition, minimal phototoxicity, and multiview sample rotation. Based on our experience over the last 10 years, we summarize the key concepts in light sheet microscopy, typical implementations, and successful applications. In particular, sample mounting for long time-lapse imaging and the resulting challenges in data processing are discussed in detail.

Introduction

Selective plane illumination microscopy (SPIM) was introduced to the life sciences in 2004 (Huisken, Swoger, Del Bene, Wittbrodt, & Stelzer, 2004) although the idea of using a light sheet to achieve optical sectioning had already been around for a century (Siedentopf & Zsigmondy, 1902). SPIM turned out to be a very powerful tool especially for the community of biologists interested in imaging developmental processes in 3D. With the invention of genetically encoded fluorescent proteins in the early 1990s (Prasher, Eckenrode, Ward, Prendergast, & Cormier, 1992), scientist started to label different cell types in a living organism with a variety of colors (Tsien, 2010), but conventional microscopes were not able to provide sufficient penetration and acquisition speed to capture all the details in a live embryo. The size and opacity of whole organisms (often a few millimeters in size) made it difficult to achieve single-cell resolution deep inside the tissue. SPIM has filled this niche and quickly proofed useful for a variety of applications.

Nowadays, the user can choose from a number of different instrumental designs classified as light sheet microscopes, but they all still share the basic features: instantaneous optical sectioning is achieved by illuminating the sample with a sheet of light and generating fluorescence in a thin slice, which is then imaged with a fast camera (Huisken & Stainier, 2007). In SPIM, millimeter-sized specimens can be reconstructed by rotating and imaging them from different sides (multiview imaging) (Preibisch et al., 2010, Swoger et al., 2007). SPIM has been widely used for various biological applications, mainly zebrafish (Keller et al., 2008, Scherz et al., 2008, Swoger et al., 2011) and fly embryos (Huisken et al., 2004, Krzic et al., 2012, Tomer et al., 2012) as well as single cells and spheroids (Lorenzo et al., 2011, Planchon et al., 2011, Ritter et al., 2010, Siedentopf and Zsigmondy, 1902), Caenorhabditis elegans (Fickentscher et al., 2013, Prasher et al., 1992, Wu et al., 2011) and fixed mouse embryos or organs (Ermolayev et al., 2009, Jährling et al., 2009, Silvestri et al., 2012, Tsien, 2010). Recently, also plants have been imaged successfully with SPIM (Huisken et al., 2004, Maizel et al., 2011, Sena et al., 2011).

Phototoxicity in SPIM has been shown to be very low even at high acquisition rates (Preibisch et al., 2010, Reynaud et al., 2008). As a result, the imaging speed is less dictated by how much light the sample tolerates; rather, it is more determined by the speed of the camera. Therefore, SPIM instruments have become the tool of choice for recording fast-changing, weakly expressing tissues in sensitive embryos, where phototoxicity needs to be avoided at all costs (Ahrens et al., 2013, Ichikawa et al., 2013, Jemielita et al., 2012, Keller et al., 2008, Scherz et al., 2008, Swoger et al., 2011, Tomer et al., 2012, Wu et al., 2011). Developmental biologists can now benefit from the ability to watch cellular and morphogenetic events occur in real time in an entire embryo, advancing our understanding on how cells form tissues and organs. At the same time, SPIM challenges existing data and image processing tools, which need to be adapted to extract the desired answers to our biological questions from the large amounts of data (Ahrens et al., 2013, Krzic et al., 2012, Schmid et al., 2013). In SPIM, besides the actual imaging, it is equally important to mount the sample under ideal physiological conditions and have the proper infrastructure to deal with the enormous datasets.

In this chapter, we take a close look at the principles of light sheet microscopy and the possible implementations. We also present mounting strategies for specimens and discuss data acquisition and handling.

Light sheet microscopy combines two distinct optical paths, one for fast wide-field detection and one for illumination with a thin sheet of light, orthogonally to the detection path (Fig. 11.1A; Huisken et al., 2004). The light sheet is aligned with the focal plane of the detection path, and the waist of the sheet is positioned in the center of the field of view (Fig. 11.1B).

SPIM's unique configuration addresses two fundamental limitations of single-lens setups, which are as follows:

• Obtaining thin optical sections is very difficult with low-NA (numerical aperture) objectives.

• The whole sample volume is illuminated when imaging a single section, multiplying the risk of fluorophore bleaching and phototoxicity (Fig. 11.2A). This effect accumulates quickly: when recording a stack of N planes, each plane is exposed N times.

By contrast, in SPIM, only the focal plane of the detection objective is selectively illuminated (Fig. 11.2B), which leads to an efficient decrease in energy input (Reynaud et al., 2008). Each plane is only exposed once during a stack. The thickness of the light sheet—usually a few micrometers—defines the axial extent of the optical section and is much thinner than the depth of field of the detection objective in common microscopy techniques. With light sheet microscopy, one can therefore acquire thin optical sections across large fields of view in big specimens.

The unique optical arrangement in light sheet microscopy provides yet another advantage. While in confocal microscopy, time-consuming scanning and discrimination with pinholes is required (Fig. 11.2C); in SPIM, optical sectioning is achieved directly across the entire plane and the image is recorded in a single exposure (Fig. 11.2D). Each pixel collects photons for the entire duration of the exposure time—usually a few milliseconds. In contrast, in the confocal, the scanner needs to rush from one pixel to the next and can only rest for microseconds. Hence, the parallel recording of all pixels in light sheet microscopy is much more efficient, and the local excitation intensity can be kept very low. In addition, in combination with fast and sensitive cameras, large image datasets are acquired much faster than with any other technique while still offering a superior signal-to-noise ratio and minimal phototoxicity. This makes light sheet microscopy the ideal technique to follow fast and dynamic developmental processes in sensitive living specimens in a minimally invasive manner (Weber & Huisken, 2011).

Light sheet illumination is especially beneficial in microscopes with low-NA, low-magnification, and long-working-distance objectives as used for large samples. While the lateral resolution of a light sheet microscope is the same as in a wide-field microscope, the axial resolution is primarily given by the light sheet thickness. As a result, a light sheet microscope equipped with, for example, a 10 ×/0.3 lens exhibits a twofold better axial resolution than a confocal microscope (Fig. 11.2E; Engelbrecht & Stelzer, 2006).

The optical arrangement of a basic light sheet microscope is quite different from conventional microscopes, yet it is comparatively straightforward. The detection path is generally identical to a wide-field fluorescence microscope: A detection objective collects light from its focal plane and passes it through a fluorescence filter, and a tube lens projects the light onto a camera chip. No dichroic mirror is needed, since no excitation light needs to be introduced through the detection path. The distinct illumination path is oriented orthogonally to the detection path and consists mainly of a coherent light source and optics to form the light sheet and project it through an illumination lens onto the sample.

In the same way that compound microscopes are built in different configurations such as upright or inverted, light sheet microscopy also comes in a variety of implementations. Ideally, the microscope is built around the specimen, providing the best possible image quality and the necessary spatial and temporal resolution while maintaining the sample under ideal conditions for the duration of the experiment. Therefore, depending on the sample of interest, light sheet microscopes may look very different although they still share the same fundamental principle.

Of crucial importance for the performance of the microscope are the properties of the light sheet such as thickness, uniformity, and ability to penetrate scattering tissue. The ideal scenario, a perfectly thin optical section, would be obtained if a sheet of light illuminated only the focal plane of the detection objective. Preferably, this sheet should be as thin as possible and uniform across the field of view. However, the laws of diffraction govern the dimensions of the light sheet and the thickness of the sheet changes across the field of view (Fig. 11.3). The NA of the illumination needs to be chosen carefully, to generate a light sheet that is sufficiently thin across the entire field of view. Due to the diffraction-limited shape of the light sheet, one can generally choose between a thin light sheet (ca. 1 μm) for small fields of view (ca. 60 μm) (Fig. 11.3A) and thicker light sheets (ca. 6 μm) for large fields of view (ca. 600 μm) (Fig. 11.3B; Engelbrecht & Stelzer, 2006).

Fundamentally, we distinguish between two classes of light sheet microscopes (Fig. 11.4C):

• The ones with a static light sheet, usually generated with cylindrical optics (Huisken et al., 2004)

• Microscopes with light sheets generated by rapidly scanning a beam up and down (Keller et al., 2008)

When compared to a cylindrical lens, a scanning mirror offers more flexibility. The height of the light sheet can easily be adapted by changing the scanning amplitude, and changing the diameter of the incoming beam can alter the light sheet thickness. Additionally, the light sheet intensity is homogeneous, and special techniques such as Bessel beams (Planchon et al., 2011), structured illumination (Breuninger et al., 2007, Keller et al., 2010), two-photon excitation (Truong, Supatto, Koos, Choi, & Fraser, 2011), and confocal line detection (Baumgart and Kubitscheck, 2012, Fahrbach et al., 2013) can be integrated. The cylindrical lens on the other hand is easy to integrate and does not require moving parts. The entire field can be illuminated at once, resulting in much less power per line and higher acquisition speeds, which are not limited by the speed of a scanning mirror but only by the speed of the camera.

To accomplish the orthogonal optical arrangement of the two beam paths, light sheet microscopes are often set up in the horizontal plane, with the sample hanging from above at the intersection of illumination and detection paths (Fig. 11.1A). A unique advantage of this vertical sample mounting is the ability to rotate the sample without deformations to quickly and precisely orient it and to image it from multiple angles (multiview acquisition; Swoger et al., 2007). A combination of linear and rotational stages is used to position the specimen and record z-stack, multiposition, and multiview datasets.

While light sheet microscopy can also be used to image fixed, cleared samples faster and more efficiently than with other techniques, light sheet microscopy excels in the imaging of living samples: Dramatically reduced phototoxicity, combined with fast acquisition and flexible sample orientation, is ideal for imaging rapid developmental processes in cell colonies, tissue samples, or entire animals. In order to provide the best environmental conditions for such imaging tasks, many light sheet microscopes are equipped with a medium-filled sample chamber and water-corrected illumination and detection objective lenses (Fig. 11.4A). Combined with water-based sample mounting, a refractive index-matched beam path can be achieved. The medium-filled imaging chambers also provide incubation systems for environmental control or drug supply.

Unfavorable optical properties of the sample can cause absorption, scattering, and refraction, which limit the penetration depth of the light sheet, broaden it, and alter its location (Fig. 11.4F). The resulting images may appear blurry, out of focus, and stripy. In this case, a second illumination arm (three-lens configuration) can improve image quality (Fig. 11.4B; Huisken & Stainier, 2007). Its optical configuration is typically identical to the first illumination arm and illuminates the sample from the opposite direction through an additional illumination lens. The light sheets generated by both illumination arms are aligned to the very same plane, the focal plane of the detection objective. The illumination arms can typically be switched on separately, in an alternating fashion, or both simultaneously. The well-illuminated parts of the images from each illumination side can then be stitched resulting in a final image of evenly good quality (Fig. 11.4G). In addition, both illumination arms may feature additional optics to pivot the light sheet around the center of the field of view to generate a more even illumination and to efficiently eliminate shadows (multidirectional, mSPIM) (Huisken & Stainier, 2007).

Double-sided illumination can also improve overall fluorescence excitation in fixed and cleared specimens (Dodt et al., 2007, Ermolayev et al., 2009, Jährling et al., 2009). For such samples, mostly upright detection paths are chosen with the sample enclosed in a cuvette (ultramicroscopy; Fig. 11.4C). Living embryos, however, are relatively opaque and impossible to penetrate from a single side. The images become deteriorated when imaging deeper into the tissue. One potential way of solving these issues is multiview acquisition, in which the specimen is rotated to acquire datasets from multiple angles and fuse them subsequently (Fig. 11.5A). Multiview acquisition holds two advantages: On the one hand, the overall image quality is improved as additional data are acquired from more favorable angles; on the other hand, overlapping datasets can be processed to yield more isotropic resolution (Swoger et al., 2007).

To facilitate multiview imaging further, a second detection arm can be added (four-lens configuration; Fig. 11.4D and B). Now, the specimen is imaged from two sides simultaneously and both halves—the front and the back of the sample—are imaged in one continuous z-stack. Besides a minor gain in speed, the main advantage is that the views from the two cameras are aligned and no further registration is necessary. The fusion can therefore be performed in real time (Krzic et al., 2012, Schmid et al., 2013). If rotation is not required, or the specimen itself demands a horizontal orientation, light sheet illumination can also be implemented in different arrangements (e.g., iSPIM, Wu et al., 2011; Fig. 11.4E).

Of crucial importance for the optical performance of a light sheet microscope is the choice of objective lenses for illumination and detection. For imaging living specimens, water-corrected objective lenses are the first choice. In case of a horizontal arrangement and the use of a medium-filled chamber, either water-dipping or air objectives can be used. Optically ideal is the use of water-dipping lenses: the minimized amount of interfaces with varying refractive indexes helps to get the best possible light sheet across the entire excitation spectrum. But often, the orthogonal orientation of the illumination and detection paths restricts the selection of suitable objectives. A high-numerical-aperture detection objective is typically too big to be combined with another, orthogonally placed water-dipping objective. Therefore, in many cases, an air illumination objective needs to be used. Fortunately, no high NA is required on the illumination side, and a simple 10 ×/0.3 lens is sufficient to generate a light sheet with a thickness of only a couple of micrometers.

No matter how powerful a certain microscopy technique is, it will only be useful when it is compatible with biological samples; hence, the mounting of the sample is crucial. In light sheet microscopy, sample preparation is radically different from the dish or slide mounting used for conventional microscopy, since usually, the sample is oriented vertically. Further, the mounted sample needs to be cylindrical to observe the specimen from all sides without distortions and exploit the unique sample rotation in light sheet microscopy. Therefore, new mounting strategies have been established dedicated for light sheet microscopy imaging as well as for particular samples and biological questions (Kaufmann, Mickoleit, Weber, & Huisken, 2012). Living samples are most often immersed in an aqueous medium appropriate for the particular organism. For ultramicroscopy, the fixed sample is immersed in a clearing solution and illuminated and imaged from outside the chamber (Dodt et al., 2007). Since light sheet microscopy has a unique potential in imaging living samples, we primarily discuss mounting strategies for in vivo imaging in the succeeding text.

One of the first, broadly used mounting protocols involves low-melting agarose in a glass capillary (Fig. 11.6A; Reynaud et al., 2008). Here, the specimen is embedded in a solid cylinder of agarose, which is extruded into the medium-filled chamber. The transparent agarose matches the refractive index of water (1.33) and biological tissue, and concentrations of 1.0–1.5% provide enough mechanical stability to reproducibly move the sample. The glass capillaries are reusable and easy to handle. Fiducial markers like beads for multiview reconstruction can be added and dispersed easily in liquid agarose yielding an even distribution within the solid agarose cylinder (Fig. 11.6A). Capillaries with different sizes are available, such that the right size can be chosen for the particular model system. This method works well for medium to large specimens like zebrafish and drosophila embryos. Single cells and cysts are generally embedded in hollow agarose cylinders or Matrigel enclosed in a small bag of transparent foil (Keller, Pampaloni, & Stelzer, 2006). Mounting in an agarose cylinder is ideal for snapshots, since the sample is immobilized very well.

If the experimental design requires fast movements or rotations of the sample, the agarose cylinder might shake resulting in a blurred image. Therefore, plastic tubes made of fluorinated ethylene propylene (FEP) can be used to surround the agarose cylinder (Fig. 11.6B). These tubes were chosen for their refractive index (1.34), which is close to the one of water (1.33) allowing to image through the tube without extruding the wobbly agarose cylinder. It was shown that the image quality with tubes is as good as with pure agarose cylinders (Kaufmann et al., 2012). The tubes are commercially available in different sizes, and the appropriate size for the particular experiment can be chosen, for example, zebrafish embryos of various stages inside or outside the chorion. The tubes need to be cleaned prior to mounting by rinsing with NaOH, EtOH, and water and ultrasonication (Kaufmann et al., 2012).

The mounting protocols in FEP tubes can also be adapted for long-term experiments. Imaging developing organisms over many hours or a few days requires careful optimization of the mounting technique, since the organism may grow and change its outer shape significantly over the course of the experiment. For example, the size of zebrafish embryos increases by a factor of 4 within 3 days (Kimmel, Ballard, Kimmel, Ullmann, & Schilling, 1995). Time-lapse imaging of embryos embedded in a rigid scaffold, such as agarose, leads to severe growth restrictions and developmental defects (Kaufmann et al., 2012). The optimal mounting for imaging developmental processes should provide enough space for the sample to grow while keeping it in a fixed position. To ensure the sample's normal growth and development during long-term experiments, the FEP tube can be filled with liquid or viscous media, for example, methylcellulose or low-concentrated agarose (Fig. 11.6C). Sedative drugs can also be used to immobilize the organism but should be kept to a minimum since prolonged exposure might interfere with development (Kaufmann et al., 2012). The tube is closed with a plug made of 1–3% low-melting agarose. This slightly more laborious mounting method ensures normal development of a rapidly growing organism like the zebrafish embryo (Fig. 11.7).

Every microscope is different and so are the user's experiences. SPIM may feel particularly different from a conventional microscope since most setups are fully digital; no eyepiece is available to quickly check the location and orientation of the sample by eye. Moreover, the specimen is often oriented vertically rather than horizontally, and stage controls may be very different from classical microscope interfaces. The ability to quickly scan through the entire depth of the sample without the need to wait for a scan to finish like on a confocal microscope makes the SPIM workflow very fast and straightforward. Similarly, data acquisition is very fast and experiments can be performed in rapid succession. However, a large amount of data can accumulate quickly, and long recordings should be done only under cautious consideration of storage and processing capacity.

In a normal compound fluorescence microscope, the mounted sample is placed on the microscope stage, and both illumination and detection are performed from either above or below (upright or inverted microscope). This constellation fixes the viewing angle and the sample can only be translated in x, y, and z. Light sheet microscopy adds another degree of freedom with sample rotation and thus allows the user to precisely orient the sample. This additional freedom alleviates mounting when it comes to accurately position the sample, since final adjustments in sample orientation can be performed in the microscope while evaluating the resulting images in real time. Additionally, multiple datasets can be acquired from different views and fused to image samples that would otherwise exceed the microscope's penetration depth.

At the same time, in SPIM, the orientation of the sample is also more important than in a classical single-lens microscope. The user needs to ensure that the sample is properly oriented with respect to both the illumination axis and the detection axis: The light sheet needs to reach the region of interest without passing highly refractive or absorbing structures of the sample, and the fluorescence should leave the sample unhindered. The ability to rotate the sample can be crucial for finding the optimal angle.

The first crucial step during the process of image acquisition that differs distinctly from traditional fluorescence microscopy techniques is the alignment of the light sheet. Here, we distinguish between adjusting the basic parameters of the light sheet, namely, height and thickness, and correctly positioning the light sheet in the focal plane of the detection lens. All three steps are essential for achieving the best possible optical sectioning performance as well as reliable, consistent performance. Furthermore, the alignment of the light sheet and the knowledge of its parameters can be a crucial prerequisite for defining further experimental strategies, for example, the spacing of a z-stack and subsequent analysis steps.

Indispensable tools for setting up and aligning the light sheet are three test specimens: a fully reflective mirror, a reflective grid, and (multicolor) beads in agarose. They help to control the thickness of the light sheet at various positions, the homogeneity of illumination, the precise positioning in the focal plane of the detection objective, and (for multicolor experiments) the proper alignment of different excitation sources.

The light sheet needs to cover the entire height of the field of view. In the case of a scanned light sheet, this property can simply be adjusted by setting the scanning amplitude. In the case of a static light sheet, the illumination beam has to be expanded. The intensity profile of the incoming beam typically follows a Gaussian pattern, and only the central part of the beam is considered more or less uniform. A slit aperture limits the height of the light sheet and avoids unnecessary energy input outside the field of view.

As mentioned earlier, the thickness of the light sheet defines the extent of the optical section in axial direction. The thinner the light sheet, the smaller the extent of the waist along the beam propagation axis and, consequently, the smaller the useable field of view (Fig. 11.3A and B). These light sheet properties are generally determined by the optics and are often chosen during the design of the instrument. The user then only needs to confirm the proper alignment of the light sheet before an experiment (Fig. 11.8A).

Once the NA of the illumination has been chosen to give the ideal light sheet for the desired field of view, the light sheet needs to be aligned with respect to the focal plane of the detection arm. The first step is to translate the light sheet along its direction until it is centered in the field of view (Fig. 11.8B). To verify this position, the excitation beam is attenuated and the emission filter removed. A reflective mirror is placed in the focal plane and tilted by 45° into the direction of the light sheet. By moving the mirror along the illumination axis, one can sample the light sheet and find the waist, which needs to be in the center of the field of view. In addition, the homogeneity of the illumination in the center of the light sheet can be checked at this point.

The light sheet needs to illuminate the focal plane of the detection unit (Fig. 11.8C). A semireflective grid is the ideal tool to perform this adjustment. Turned by 45°, the focal plane can be clearly identified in transmission and the light sheet is visible on the reflective gridlines. Using an adjustable mirror in the illumination path (Huisken & Stainier, 2007), the light sheet is moved along the detection axis until it overlaps with the focal plane. Additionally, mounted fluorescent beads can be used to adjust the light sheet until the point-spread function appears symmetric. If the system features multiple excitation laser lines, all light sheets need to overlap, which can be checked easily with multicolor fluorescent beads.

To confirm that the light sheet is overlapping with the focal plane across the entire field of view (Fig. 11.8D), the grid can be moved along the illumination axis. Residual tilt needs to be eliminated.

Light sheet microscopes feature both high speed and low phototoxicity, so that many planes and many time points can be imaged without any impact on the specimen. Although one may aim for highest spatial and temporal resolution, the resulting data volumes are enormous; thus, data storage and strong computation power need to be available to process and archive the data. More than for other microscopy techniques, in SPIM, the user has to think about the necessary amount of information before the experiment.

Acquisition parameters to be considered are

• light sheet thickness,

• z-spacing for stacks,

• exposure time and region of interest of the camera,

• laser illumination power and duration,

• movie frame rate and duration,

• time-lapse interval and duration,

• number of angles for multiview acquisition.

The user needs to find the right balance between image quality and data size. A thinner light sheet gives a better axial resolution, but more planes are required to reconstruct the sample. A shorter exposure and illumination time results in less motion blur in moving specimens and higher temporal resolution, but it requires more energy per frame for the same signal-to-noise ratio. Increasing the frame rate not only gains temporal resolution but also increases the amount of data. Here, the necessary speed depends very much on the actual speed of events in the specimen. The same holds true for the interval for time-lapse recordings.

An exceptional feature of light sheet microscopes is the ability to rotate the specimen. This is advantageous when imaging samples that are too big to be imaged from a single side. Multiview reconstruction also improves the axial resolution by adding high-resolution information from an overlapping acquisition taken under a different angle (Preibisch et al., 2010). Prior to any multiview acquisition, it is crucial to set the right number of angles to achieve full coverage of the sample with sufficient overlap.

One of the striking advantages of SPIM is its ability to image large specimen with high spatial and temporal resolution over several hours or even days (Kaufmann et al., 2012). The resulting amounts of image data are enormous, easily several terabytes. The speed at which the samples can be imaged is not limited anymore by the acquisition but rather by the processing computers, the storage capacity, and data transfer rates.

When imaging at moderate frame rates, the data from the camera can simply be transferred to the storage drive of the acquisition computer by means of standard consumer-level connections such as USB or FireWire or even to a remotely located storage computer using network connections. With faster cameras, data transfer rates become limiting. Some cameras allow streaming image data to internal storage, which needs to be read out after acquisition. Other cameras use faster data connections, which partly require additional hardware (frame grabber) to be installed in the computer. The data are then streamed either to the main memory that needs to be sufficiently large or to a data storage device that must then be equally fast to avoid loss of data.

One way to reduce the data rate is image compression. Image data can be compressed either in the camera, in the frame grabber card, by the processor (CPU), or the graphics card (GPU). This relieves pressure from the data transmission and storage system but increases the demand of processing power for compression and subsequent decompression.

A more powerful option to cope with the amount of data is to predefine regions of interest and store only data from those regions, as it was done with the endoderm during zebrafish gastrulation (Schmid et al., 2013). This kind of real-time processing is custom-made for a certain specimen and a scientific question, but the image data are inherently reduced. At the same time, the data are already nicely visualized to facilitate further analysis. Ideally, the microscope (or the camera) should directly deliver preprocessed and compressed data or, if possible already, the final results (Fig. 11.9).

The third possibility is intelligent microscopy, going one step further and not predefining regions of interest, but rather letting the microscope decide which areas to image at which resolution. This decision could even be based on existing, previously recorded data from similar specimen.

Image data from light sheet microscopes do typically not require excessive enhancement, denoising, or restoration steps, thanks to its superior signal-to-noise ratio and overall image quality. If necessary, various filters used for confocal or wide-field microscopy data can be used for SPIM data as well.

A necessary processing step after acquiring multiview image data (see acquisition) is multiview fusion. The first step is to register the individual views to each other in space, which can be facilitated by fiducial markers (e.g., fluorescent beads, see mounting; Preibisch et al., 2010), structures within the specimen (e.g., fluorescent nuclei), or precise knowledge of the relative motor positions during the acquisition (Krzic et al., 2012, Schmid et al., 2013). Once registered, image data can be fused by averaging the image intensities or in a content-based manner. The result of multiview fusion is a single dataset that inherits the best features of the individual views. Ultimately, one would want to perform this multiview fusion in real time so that only the fusion is saved and none of the raw data.

Methods for image analysis do not differ much between SPIM and conventional, for example, confocal microscopes. Datasets may need to be reconstructed in 3D, objects may need to be segmented and tracked, intensities may need to be measured over space and time, and correlations between multiple channels may need to be analyzed. For colocalization studies and fluorescence quantification, one has to keep in mind that the light sheet dimensions and intensities change towards the edges of the field of view, influencing illumination and fluorescence. Therefore, the user should carefully check the light sheet parameters (see acquisition) and use beads to verify the overlay of individual channels in x, y, and z.

The main difference, however, to common light microscopy data is typically the sheer amount of data. Light sheet microscopy data acquired with high spatial and temporal resolution require significantly more computing power. This especially applies for the analysis of time-lapse data, for which common analysis tools often require the entire dataset to be loaded in the memory, which is often impossible with SPIM data. More than for common light microscopy experiments, the user must think about the required resolution: The highest possible speed, largest field of view, and highest resolution may be nice to have but can render subsequent analysis a lot more difficult.