Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2019 Nov 26;9(1):17625.
doi: 10.1038/s41598-019-53875-y.

Deconvolution of light sheet microscopy recordings

Klaus Becker  1   2 , Saiedeh Saghafi  3 , Marko Pende  3   4 , Inna Sabdyusheva-Litschauer  3   4 , Christian M Hahn  4 , Massih Foroughipour  3   4 , Nina Jährling  3   4 , Hans-Ulrich Dodt  3   4
Affiliations

Deconvolution of light sheet microscopy recordings

Klaus Becker et al. Sci Rep. .

Abstract

We developed a deconvolution software for light sheet microscopy that uses a theoretical point spread function, which we derived from a model of image formation in a light sheet microscope. We show that this approach provides excellent blur reduction and enhancement of fine image details for image stacks recorded with low magnification objectives of relatively high NA and high field numbers as e.g. 2x NA 0.14 FN 22, or 4x NA 0.28 FN 22. For these objectives, which are widely used in light sheet microscopy, sufficiently resolved point spread functions that are suitable for deconvolution are difficult to measure and the results obtained by common deconvolution software developed for confocal microscopy are usually poor. We demonstrate that the deconvolutions computed using our point spread function model are equivalent to those obtained using a measured point spread function for a 10x objective with NA 0.3 and for a 20x objective with NA 0.45.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1
Figure 1
Recording of fluorescent beads. (a) In a light sheet microscope as described in the light sheet is generated by a sole cylinder lens and a slit aperture located directly in front of it. Incident on the slit aperture is a Gaussian distributed laser beam that has been expanded to a diameter of about 20 mm by a Galilean beam expander. Specimen and the tip of the objective are immersed in a water filled glass container. A gelatin block with embedded fluorescent particles of 200 nm diameter is placed in the focal region of the light sheet. If the system is correctly adjusted, a small number of fluorescent particles best possibly fits the focal line of the cylinder lens, as well as the focal plane of the objective. For recording, the glass container is stepwise moved vertically through the slight-sheet, while an image is recorded at each position. Since the objective (or the water-proofed protection cap mounted in front of it, respectively) is submerged in the specimen container, all optical path lengths remain constant during the recording procedure. When entering the specimen chamber, the fan angle α of the light sheet changes to β. However, due to the relation NALs = sin(α) = nsin(β) the numerical aperture NALs of the light sheet remains unchanged. (b) A fluorescence emitting sub-resolution particle that is located in the focus of a correctly adjusted light sheet microscope is subjected to an illumination point spread function PSFIL describing the spatial distribution of the excitation light close to the particle and to a detection point spread function PSFdet describing the near field distribution of the emitted fluorescence light collected by an objective of numerical aperture NAObj. The point spread function PSFLSM of the light sheet microscope can be described as the elementwise product PSFIL x PSFdet.
Figure 2
Figure 2
Comparison between measured and modelled PSFs of a light sheet microscope. (a1) 200 nm sized fluorescent beads were recorded using a 10x objective with NA = 0.3 (UPLFLN 10x, Olympus, Germany). The figure shows a maximum intensity (MIP) projection (xz-direction) obtained from 402 slices. (a2) PSF extracted by registration and averaging of 10 manually selected beads from (a1) (left), compared to a calculated PSF according to Eq. (15) (Model parameters: NA = 0.3, λex = 488 nm, λem = 520 nm, n = 1.561, f = 80 mm, d = 8 mm, no damping). (a3) comparison of lateral (x, y = 0, z = 0) and axial (x = 0, y = 0, z) intensity profiles of the measured and the modelled PSFs depicted in (a2). (b1) 200 nm fluorescent beads were recorded using a 20x objective with NA = 0.45 (LUCPLFLN, Olympus, Germany). The panel shows a maximum intensity projection (MIP, xz-direction) of the 3D reconstructed fluorescent beads obtained from 500 slices. (b2) PSF obtained after registration and averaging of 10 manually selected beads from b1 (left) versus a calculated PSF according to Eq. (15). (Model parameters: NA = 0.6, λex = 488 nm, λem = 520 nm, n = 1.561, f = 80 mm, d = 8 mm, no damping). (b3) comparison of lateral (x, y = 0, z = 0) and axial (x = 0, y = 0, z) intensity profiles of the measured and of the modelled PSFs depicted in (b2). As for the 10x objective visual comparison confirms a good agreement between measurements and theory.
Figure 3
Figure 3
Comparison of deconvolution results either obtained using a measured PSF or a modelled PSF. (a1) 3D-reconstructions of a cleared fruit fly from 353 raw slices (MIP projection). Images were recorded using a 10x objective with NA = 0.3 (UPLFLN 10x, Olympus, Germany). (a2) Data from a1 after deconvolution with the measured PSF depicted in a2. (a3). Data from a1 after deconvolution with the modelled PSF depicted in a2. A comparison of a2 and a3 shows that the deconvolution results are almost identical. (a4) Mean standard deviations (MSD) measured between a1 vs. a2 (4.30 × 105 = 99.8%), a1 vs. a3 (4.31 × 105 = 100%) and a2 vs. a3 (9.33 × 104 = 21.6%). (b1) Detail of a cleared mouse embryo (dorsal root ganglia) obtained from 419 raw slices (MIP projection). Images were recorded using a 20x objective with NA = 0.45 (LUCPLFLN, Olympus, Germany). (b2) Data from b1 after deconvolution using the measured PSF depicted in b2,3). Data from b1 after deconvolution using the modelled PSF depicted in b2 (right). Visual comparison of b2 and b3 confirms that the deconvolution results are virtually identical. (b4) Mean standard deviations (MSD) measured between b1 vs. b2 (1.37 × 108 = 85.5%), b1 vs. b3 (1.61 × 108 = 100%) and b2 vs. b3 (1.52 × 106 = 0.95%).
Figure 4
Figure 4
3D reconstructions from different samples prior and after deconvolution. The deconvolution approach described in this paper provides a significant increase in sharpness and in the level of details of light sheet microscopy recordings for different samples and magnifications ranging from 1x to 20x. (a) 3D reconstructions (MIP projection) of an E12.5 mouse embryo that was immune-stained and chemically cleared according to. Nerve fibers are highlighted by NF-160 fluorescence labelling. The reconstruction was obtained from 667 slices recorded using a 2.5x objective (Zeiss FLUAR 2.5x, Carl Zeiss, Germany) with an NA of 0.12 and a 0.5x post magnification. Recording was performed using a light sheet microscope equipped with a single cylindrical lens of 80 mm focal length and a 6 mm wide slit aperture as described in. For imaging, a Cool Snap Cf CCD camera (Roper Scientific, Germany) with 1392 × 1040 pixel resolution was used. Illumination time: 430 ms. (a1) The left column shows a 3D-reconstruction obtained from the unprocessed raw data. The right side shows a reconstruction obtained from the same data set after deconvolution with a calculated PSF. (a2) Zoomed details from three selected regions before deconvolution (a1–c1) and after deconvolution (a2–c2). Deconvolution parameters were NA = 0.14, λex = 488 nm, λem = 520 nm, n = 1.561, f = 80 mm, d = 8 mm, stop criterion = 0.1%, no damping. (b) 3D reconstructions obtained from an isolated EGFP expressing mouse hippocampus that has been chemically cleared according to. (b1) Reconstruction obtained from 1050 slices recorded using a 5x objective (FLUAR 5x, Carl Zeiss, Germany) with a NA of 0.25 and a 0.5x post magnification (left image) and the same data set after deconvolution (right image). Recording was performed with a light sheet microscope equipped with a 80 mm cylindrical lens and a 6 mm wide slit aperture as described in. For deconvolution, the data set was split into 1 × 1 × 3 equally sized blocks. Deconvolution parameters were NA = 0.25, λex = 488 nm, λem = 520 nm, n = 1.561, f = 80 mm, d = 8 mm, stop criterion = 0.1%, no damping. Imaging was done using a Cool Snap cf camera (Roper Scientific, Germany) with 1392 × 1040 pixel resolution. Illumination time: 2000 ms. (b2) Reconstruction obtained from 221 slices of 1392 × 1040 pixel resolution recorded using a 20x objective (LUCPLFLN 20x, NA 0.45, Olympus, Germany) (left column) and the same data set after deconvolution (right column). Recording was performed with a light sheet microscope equipped with a 80 mm cylindrical lens and a 16 mm wide slit aperture as described in. Deconvolution parameters were NA = 0.45, λex = 488 nm, λem = 520 nm, n = 1.561, f = 80 mm, d = 16 mm, 30 iterations. Imaging was done using a Cool Snap cf (Roper Scientific, Germany) camera with 1392 × 1040 pixel resolution. Illumination time: 10000 ms. c) Part of the head of an entire adult mouse chemically cleared according to. c1) Reconstruction obtained from 1297 slices recorded using a 2x objective (XLFLUOR 2x, Olympus, Germany), with an NA of 0.14 and a 0.63x post magnification (left image) and the same data set after deconvolution (right image). Recording was performed using a light sheet microscope equipped with a modified light sheet generator as described in. For imaging an Andor Neo CMOS camera (Andor, Ireland) with 2560 × 2160 pixel resolution was used. Illumination time: 180 ms. Deconvolution parameters were NA = 0.14, λex = 488 nm, λem = 520 nm, n = 1.561, f = 80 mm, d = 16 mm, 30 iterations, no damping. c2) Zoomed details from regions a-b marked in c1 before deconvolution (left) and after deconvolution (right). d) Cortical neurons recorded in an entire mouse brain that was chemically cleared according to. (d1) Reconstructions obtained from 777 slices recorded using a 4x objective (XLFLUOR 4x, Olympus, Germany) with an NA of 0.28 and a 2x post magnification (left column) and the same data set after deconvolution (right column). Recording was performed using a light sheet microscope with a modified light sheet generator as described in. Imaging was done using an Andor Neo CMOS camera with 2560 × 2160 pixel resolution. Illumination time: 250 ms. Deconvolution parameters were NA = 0.28, λex = 488 nm, λem = 520 nm, n = 1.561, f = 80 mm, d = 16 mm, 30 iterations, stop criterion = 0%, no damping. (d2) Two different zooms of d1 before deconvolution (left) and after deconvolution (right).
Figure 5
Figure 5
Effect of the damping parameter on noise amplification. (a) Single slice out of a light sheet microscopy image stack comprising 51 slices recorded from a GFP-expressing mouse brain. For demonstrating, the effect of the damping parameter, artificial computer generated Gaussian noise was added. (b) Deconvolution of the same data without damping causes an amplification of the noise, masking fine details of the image. The quality after deconvolution is even worse than the quality before deconvolution. (c) Deconvolution of the same data set using 5% damping. The amplification of noise is significantly reduced and a quality improvement compared to the raw data (A) is clearly recognizable now.
Figure 6
Figure 6
Modelling the PSF of a light sheet microscope. Simulations of PSFIL and PSFdet of a light sheet microscope for NALS = 0.1 and NAObj = {0.15, 0.3, 0.45} (left). The according effective point spread function PSFLSM is depicted on the right side. For a 2x objective with NAObj = 0.15 the effect of the detection PSF is small, since most outer parts of PSFdet are multiplied with values that are close to zero. However, the higher NAObj becomes, the more it effects the shape of PSFLSM. To enhance the visibility of the side lobes of the PSFs, the images were gamma corrected by γ = 0.4.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Chatterjee K, Pratiwi FW, Wu FCM, Chen P, Chen BC. Recent Progress in Light Sheet Microscopy for Biological Applications. Appl. Spectrosc. 2018;72:1137–1169. doi: 10.1177/0003702818778851. - DOI - PubMed
    1. Girkin JM, Carvalho MT. The light-sheet microscopy revolution. J. Opt. (United Kingdom) 2018;20:53002.
    1. Dodt H-U, et al. Ultramicroscopy: three-dimensional visualization of neuronal networks in the whole mouse brain. Nat. Methods. 2007;4:331–6. doi: 10.1038/nmeth1036. - DOI - PubMed
    1. Keller PJ, Dodt HU. Light sheet microscopy of living or cleared specimens. Curr.Opin.Neurobiol. 2012;22:138–143. doi: 10.1016/j.conb.2011.08.003. - DOI - PubMed
    1. Huisken, J., Swoger, J., Linkdeck, S. & Stelzer, E. H. K. Selective Plane Illumination Microscopy. In Handbook of Biological Confocal Microscopy (ed. Pawley, J. B.) 672–679 (Springer, 10.1007/978-0-387-45524-2 2006).

Publication types