Highly efficient 3D fluorescence microscopy with a scanning laser optical tomograph

doi:10.1364/OE.19.005419

Highly efficient 3D fluorescence microscopy with a scanning laser optical tomograph

Raoul-Amadeus Lorbeer, Marko Heidrich, Christina Lorbeer, Diego F. Ramírez Ojeda, Gerd Bicker, Heiko Meyer, and Alexander Heisterkamp »View Author Affiliations

Optics Express, Vol. 19, Issue 6, pp. 5419-5430 (2011)
http://dx.doi.org/10.1364/OE.19.005419

View Full Text Article

Acrobat PDF (1928 KB)

Journal Search
Article Lookup

Article Tools

Citations

Alert me when this article is cited
Export Citation/Save

Full Text
Article Info
References (24)
Cited By
Figures (6)
Multimedia (2)
Metrics

Abstract

Optical Projection Tomography (OPT) proved to be useful for the three-dimensional tracking of fluorescence signals in biological model organisms with sizes up to several millimeters. This tomographic technique detects absorption as well as fluorescence to create multimodal three-dimensional data. While the absorption of a specimen is detected very fast usually less than 0.1% of the fluorescence photons are collected. The low efficiency can result in radiation dose dependent artifacts such as photobleaching and phototoxicity. To minimize these effects as well as artifacts introduced due to the use of a CCD- or CMOS- camera-chip, we constructed a Scanning Laser Optical Tomograph (SLOT). Compared to conventional fluorescence OPT our first SLOT enhanced the photon collection efficiency a hundredfold.

1. Introduction

During the last decade, several optical techniques to image specimens of sizes between 1 and 10 millimeters have been developed [1

J. Sharpe, U. Ahlgren, P. Perry, B. Hill, A. Ross, J. Hecksher-Sørensen, R. Baldock, and D. Davidson, “Optical projection tomography as a tool for 3D microscopy and gene expression studies,” Science 296, 541–545 (2002). [CrossRef] [PubMed]

–6

D. Razansky, M. Distel, C. Vinegoni, R. Ma, N. Perrimon, R. W. Köster, and V. Ntziachristos, “Multispectral opto-acoustic tomography of deep-seated fluorescent proteins in vivo,” Nat. Photonics 3, 412–417 (2009). [CrossRef]

]. All these techniques were used to detect the fluorescence after gene expression. This led to advancements in the study of developmental biology on Danio rerio (zebrafish), Drosophila melanogaster, Caenorhabditis elegans and murine embryos. Optical Projection Tomography (OPT), as the only - all optical - technique, has the ability to measure the three-dimensional distribution of optical absorption and fluorescence within a specimen. As a part of the method the numerical aperture (NA) of the detection path is reduced until the focal depth covers the whole sample. Rotating the specimen allows the detection of transmission images (tOPT) from multiple views which then can be reconstructed to a three-dimensional data set similar to x-ray projections in computed tomography (CT) [7

T. M. Buzug, Computed Tomography from Photon Statistics to Modern Cone-beam CT (Springer-Verlag, 2008).

]. Analogous to this, emission OPT (eOPT) uses fluorescent light instead of absorption to create an image contrast [8

J. Sharpe, “Optical projection tomography,” Annu. Rev. Biomed. Eng. 6, 209–228 (2004). [CrossRef] [PubMed]

]. The reduction of the NA causes eOPT to be accompanied by a sincere decrease of the light collection efficiency. Therefore, more excitation light is required resulting in photo-bleaching and phototoxic effects which are significantly higher compared to light sheet based microscopy setups [3

J. Huisken, J. Swoger, F. D. Bene, J. Wittbrodt, and E. H. K. Stelzer, “Optical sectioning deep inside live embryos by selective plane illumination microscopy,” Science 305, 1007–1009 (2004). [CrossRef] [PubMed]

–5

J. Huisken and D. Stainier, “Selective plane illumination microscopy techniques in developmental biology,” Development 136, 1963–1975 (2009). [CrossRef] [PubMed]

To overcome this problem, we introduced a laser scanner in our OPT setup which we will refer to as Scanning Laser Optical Tomograph (SLOT) or Scanning Laser Optical Tomography (SLOTy). The source of fluorescence emission is known due to the scanning position of the focal spot while the fluorescence photons can be collected with simple optics. This decouples the image formation and light collection from each other. That theoretically allows, nearly independent of the sample size, all of the emitted photons to reach the detector. Ring artifacts introduced by pixel to pixel differences of e.g. a CCD- or CMOS-camera-chip [9

J. R. Walls, J. G. Sled, J. Sharpe, and R. M. Henkelman, “Correction of artefacts in optical projection tomography,” Phys. Med. Biol. 50, 4645–4665 (2005). [CrossRef] [PubMed]

] are omitted.

To demonstrate the sensitive imaging of extensively branched cells with SLOTy we visualize here the arborization pattern of serotonergic neurons in the central nervous system of locust and Drosophila larvae, two established model organisms for developmental neurobiology [10

J. Thomas, M. Bastiani, M. B. Bate, and C. Goodman, “From grasshopper to Drosophila: a common plan for neuronal development,” Nature 310, 203–207 (1984). [CrossRef] [PubMed]

]. The biogenic amine serotonin is a neuromodulatory compound found throughout the animal kingdom that can be detected using immunofluorescence staining.

The results show the ability to resolve single neuronal cells including their branching pattern with a high sensitivity covering the whole sample. To demonstrate the suitability of this novel tomographic technique for monitoring, we reduced serotonin levels by reserpine treatment and visualized the corresponding fluorescence changes with SLOTy. Technically the SLOT reaches a hundredfold higher sensitivity than our eOPT setup. The comparison with other microscopy techniques suggests the possibly highest sensitivity of all available optical microscopy techniques for objects larger than 2 mm.

2. Materials and methods

SLOT

The schematic setup of a SLOT is shown in Fig. 1a.

Fig. 1 OPT setups: (a) Schematic drawing of the SLOT. A laser beam is scanned through the sample. The sample is mounted in a rotatable glass capillary and then positioned in a cuvette filled with refractive index matching glycerol. Fluorescent light emitted to both sides is reflected and focused onto a PMT. Transmitted light is captured by a photo diode. (b) Schematic drawing of the used eOPT setup. Again, the sample is mounted in a rotatable glass capillary and then positioned in a cuvette filled with refractive index matching glycerol. Fluorescence is excited by an expanded laser beam perpendicular to a telecentric fluorescence imaging system.

A laser source which is spatially cleaned by a pinhole in a telescope is directed via two silver mirrors onto a x-y-scanning-mirror system. The scanner is placed in the back focal plane of an achromatic asphere, here with a focal length of 30 mm, to focus the laser beam into the sample chamber. In all cases the chamber is a cuvette filled with glycerol containing a sample sustained within a glass capillary. A mechanical stage rotates the capillary to achieve multiple viewing angles. Undeflected laser light is detected by a photo diode - with a maximum sensitivity of

3 \cdot 10^{6} \frac{V}{W}

- behind the chamber. Fluorescent light is collected by a lens system consisting of two plano-convex lenses with a fluorescence filter in between. Both lenses have an aperture and focal length of 25 mm and, therefore, a combined NA of approximately 0.45. They are used to redirect the light onto the single element photon multiplying tube (PMT) (R6357, HAMAMATSU Photonics K.K., Japan) detector with a quantum efficiency (QE) of 20% at 600 nm. Both detectors are read out simultaneously. One projection image is formed by scanning 500x500 points through the sample and integrating the signal generated by the PMT for every point. The scanning time for one image is 2 seconds. This includes the line end turnarounds. The time of the line scans without turnarounds is 1.5 seconds. All acquisitions are taken with a laser beam NA of 0.037 at 500 viewing angles.

eOPT setup

The eOPT system used for comparison is sketched in Fig. 1b. The fluorescence is excited perpendicular to the optical axis of the setup using the same laser light source and filter as in the SLOT above. The laser beam is expanded with a lens of 150 mm focal length and attenuated by a neutral density filter to compensate the power differences compared to the pinhole cleaned SLOT. For a better homogeneity, an aperture blocks the side slopes of the laser beam, which then is adjusted to illuminate only the imaged area. This ensures an illumination comparable with SLOT. The imaging lens of the system is identical to the focusing lens in the SLOT setup. In its back focal plane an aperture reduces the NA to achieve the necessary focal depth and a telecentric beam path. Again, the same filter as in the SLOT is used. To detect the projected fluorescence signal an electron multiplying (em)CCD camera (Andor Luca, Andor Technology plc., Belfast, Ireland) with a 28–300 mm zoom (Sigma 3.5-6.3/28-300 DG Macro NAFD, Sigma Corporation, Japan) objective is used. The zoom objective is adjusted to achieve the same field of view (FOV) as in the SLOT images. The integration time is set to 2.0 seconds - the time the SLOT needs for one image - and the em-gain is adjusted for highest contrast (em-gain = 40). The em-gain raises the CCD signal before the read out to minimize read out noise which therefore can be neglected. The QE of the camera is approximately 60% at 600 nm. The resolution of the images is 501x502 pixels due to the 2x2 binning used and therefore nearly identical with the resolution of the SLOT acquisitions. A gray value offset from the emCCD chip of approximately 488 will be subtracted for further calculations. The dark current at −20 °C of the chip is specified to be 0.17 e⁻/pix/sec and therefore 0.34 e⁻/pix for the full integration time. Here 500 viewing angles are acquired with the aperture reduced to an NA of 0.037.

Fluorescent immunocytochemistry

The specimens are optically cleared larval brains of Locusta migratoria and the central nervous system of the 3rd larval stage of Drosophila melanogaster. Immunocytochemical staining followed essentially the protocol of Stern et al. [11

M. Stern, S. Knipp, and G. Bicker, “Embryonic differentiation of serotonin-containing neurons in the enteric nervous system of the locust (Locusta migratoria),” J. Comp. Neurol. 501, 38–51 (2007). [CrossRef] [PubMed]

]. Briefly, the dissected locust brains were fixed in 4% paraformaldehyde dissolved in phosphate-buffered saline (PBS), permeabilized for 1 hour in 0.3% saponin in PBS (for Drosophila larval brains, permeabilizing time was reduced to ½ hour), rinsed three times in PBS with 0.5% Triton-X (PBST) and blocked overnight at 4°C in 5% normal goat serum (in PBST, Vector Laboratories, Burlingame, CA). After blocking, all samples were incubated at least for 3 days in rabbit-anti-Serotonin antibody (diluted 1:5000 in blocking solution, Sigma-Aldrich) at 4°C, then washed 3 times for several hours with PBST and subsequently incubated in goat-anti-rabbit-Cy3 antibody (diluted 1:250 in blocking solution, Jackson Immunoresearch, West Grove, PA) overnight at 4°C. After washing three times in PBST and PBS, samples were cleared in increasing grades of glycerol/PBS. Brains were mounted in 99.5% glycerol within glass capillaries (Borosilicate glass, Science Products). The imaging is performed with a green laser pointer at 532 nm with less than 1.0 mW output power and a 624/40 OD 6 fluorescence filter (FF01-624/40-25, Semrock Inc., Rochester, USA).

Reserpine depletion of serotonin

4 μg of reserpine, dissolved in 1 μl of DMSO were injected into the haemolymph of cold anesthetized Locusta migratoria third instar larvae through the joint of the hind leg using a Hamilton syringe. After recovery, animals were kept for 20 h and then the brain was dissected, fixed and immunocytochemically stained. The control group was treated equally, except for applying only the vehicle DMSO.

Theoretical estimations

The light collection efficiency which is depending on the NA and the refractive index of the surrounding media can be calculated by the partial integration over the surface of the unity sphere. Assuming an isotropic fluorescence emission and neglecting reflection losses the efficiency is given by

\begin{array}{l} Eff (\frac{NA}{n}) & = & \frac{1}{4 π} \int_{0}^{arcsin (\frac{NA}{n})} 2 π sin (φ) d φ \\ = & - \frac{1}{2} {[cos (φ)]}_{0}^{arcsin (\frac{NA}{n})} \\ = & \frac{1}{2} [1 - cos (arcsin (\frac{NA}{n}))] = \frac{1}{2} [1 - \sqrt{1 - {(\frac{NA}{n})}^{2}}] . \end{array}

(1)

In Fig. 2 Eq. (1) is plotted against

\frac{NA}{n}

in which n equals the refractive index of the sample medium.

Fig. 2 Fluorescent light collection efficiency over the

\frac{NA}{n}

in which n is the refractive index of the sample medium. The shaded areas indicate the practical scope of the technique.

The following examples assume the sample medium to be water (n = 1.33). OPT uses an NA around 0.025 for samples with a size of 2 mm collecting 0.0088% of the photons. In contrast to this, for selective plane illumination microscopy (SPIM) Huisken et al. used NAs between 0.25 and 0.3 accumulating 0.88% to 1.29% of the light [3

]. SLOTy collects fluorescence photons with simple collimation optics easily reaching an NA above 0.45 and, therefore, efficiencies exceeding 2.94%.

Noise calculation

To estimate the signal-to-noise ratio (SNR) the standard deviation of frame to frame differences from a 200 frame sequence are analyzed. To minimize artifacts, the sample was not rotated and the measurements were taken with the eOPT setup first. The readout offset of 488 of the emCCD camera was subtracted. At 500 positions in every image a line of 5 pixels is averaged to receive an intensity value. Thereafter the differences of two consecutive following images are taken for all 200 images. At the same positions as before the remaining standard deviation is calculated. Both results are averaged over 200 images at every position and interpreted as intensity as well as noise. This allows to calculate the SNR at these positions. To make the measurements comparable, first, the laser power is measured and recalculated to the equivalent average intensity. Second, the effective integration times of the setups are taken into account. Both is used to scale the results correctly.

Reconstruction and visualization

For the reconstruction of the projections to a three-dimensional data set, we used a filtered back projection algorithm with a ramp filter. The reconstruction of a 500x500x500 acquisition to a 512x512x500 data set takes 10 minutes and 45 seconds on one single Intel® Core™2 core with 2.4 GHz clock rate and 4 GByte of RAM. With our GPU implementation the same data set is reconstructed in 1.4 seconds on an NVIDIA® GeForce® GTX 260+ graphics card. Additional 1.1 seconds are needed to load the data and initialize the program. Saving on the hard disk takes further 1.3 seconds totaling in 3.8 seconds reconstruction time. Similar work has been published by Vinegoni et al. [12

C. Vinegoni, L. Fexon, P. F. Feruglio, M. Pivovarov, J.-L. Figueiredo, M. Nahrendorf, A. Pozzo, A. Sbarbati, and R. Weissleder, “High throughput transmission optical projection tomography using low cost graphics processing unit,” Opt. Express 17, 22320–22332 (2009). [CrossRef]

]. These data sets then were clipped with ImageJ and visualized using Voreen [13

S. R. M.D. Abramoff and P. J. Magelhaes, “Image processing with imageJ,” Biophotonics Int. 11, 36–42 (2004).

–15

J. Meyer-Spradow, T. Ropinski, J. Mensmann, and K. Hinrichs, “Voreen: A rapid-prototyping environment for ray-casting-based volume visualizations,” IEEE Comput. Graphics Appl. 29, 6–13 (2009). [CrossRef]

3. Results

We took acquisitions of the same sample with both setups which allows to connect the theoretical improvements with our measured data.

The 15 bit gray scale images of the SLOT (Fig. 3a and 3b) and the 14 bit gray scale image of the eOPT setup (Fig. 3c) have been adjusted in contrast with a non linear mapping. The first obvious difference between the two setups lies in the different noise characteristics of the images (Fig. 3a and 3c). These become even more obvious after the reconstruction (Fig. 4a and 4b). So called ring artifacts [9

J. R. Walls, J. G. Sled, J. Sharpe, and R. M. Henkelman, “Correction of artefacts in optical projection tomography,” Phys. Med. Biol. 50, 4645–4665 (2005). [CrossRef] [PubMed]

] only appeared in the eOPT reconstructions. Since absolute intensity measurements are not possible with a PMT, we compared the noise of the acquired images at 500 horizontal lines located at the horizontal center of Fig. 3a and 3c.

Fig. 3 Projection images of the brain of a Locusta migratoria first instar larva. The fluorescence images are inverted and contrast adjusted in a nonlinear manner. (a) Emission projection of the serotonin Cy3 staining with SLOTy. The left box (a’) shows a magnification of the black rectangle resolving serotonergic wide field neurons in the optic lobes. (b) Transmission projection at a laser wavelength of 532 nm acquired with the SLOT. (c) Emission projection of the serotonin Cy3 staining with eOPT. The right box (c’) shows a magnification of the black rectangle. The scale bars represent 300 μm.

Fig. 4 Reconstructed tomographic slice through a Locusta migratoria optical lobe in a side view. (a) Filtered back projected slice of the SLOT acquisition. (b) Filtered back projected slice of the eOPT acquisition. The scale bars represent 300 μm.

To test the power of the SLOT in resolving fluorescence images from a specimen of about one millimeter thickness, we used the larval brain of the locust Locusta migratoria. The maximum intensity projection of Fig. 5a shows intensely fluorescent neuronal somata in the cell body cortex, axonal projections and fine arborizations that cause the diffuse staining of the neuropile. For example in the outwards oriented optic lobes, identified cell body clusters contribute branches to the immunoreactive bands of the lamina and medulla [16

N. Tyrer, J. Turner, and J. Altman, “Identifiable neurons in the locust central nervous system that react with antibodies to serotonin,” J. Comp. Neurol. 227, 313–330 (1984). [CrossRef] [PubMed]

]. The fine grained immunofluorescence in the curved bands represents three-dimensional fiber networks of tangential wide field interneurons. Note that the fluorescent bands are clearly separated by unlabeled projections of columnar interneurons running through the chiasma between lamina and medulla, see Fig. 5a ( Media 1 and Media 2). Depletion of monoamine transmitters by the use of the re-uptake blocker reserpine is a rather straightforward method to manipulate serotonin levels in the nervous system.

Fig. 5 Maximum intensity projections of inverted immunofluorescence reconstructions showing the central nervous systems of the model organisms. (a) Locusta migratoria L1 larva - Chi: chiasma, Me: medulla, La: lamina. ( Media 1 and Media 2) The encircled areas mark the optical lobe neurons described by Tyrer et al. [16

]. (b) 3rd larval stage Drosophila central nervous system. (c) Locusta migratoria L3 larva with reserpine injection. The right rectangle shows a magnification of the indicated area. (d) Control with DMSO injection. The right rectangle shows a magnification of the indicated area. The scale bars represent 300 μm.

A comparison of a reserpined animal with a control larva (Fig. 5c and 5d) shows that SLOTy clearly resolves the decrease in immunofluorescence intensity in single neurons. The reduction in cell body numbers and signal intensity was most obvious in the identified groups of pars intercerebralis neurons (Fig. 5c and 5d). Thus it can be used as a comparatively rapid diagnostic tool for analyzing the effects of pharmacological treatment in experimental animals. To show that SLOTy provides a satisfying spatial resolution of the rather small neurons in Drosophila, we performed serotonin immunocytochemistry on the central nervous system of the 3rd larval stage (Fig. 5b), which ranges between 100–300 μm in thickness. Again, SLOTy resolved the serotonergic neurons in the brain and ventral cord as described by Vallés et al. [17

A. Vallés and K. White, “Serotonin-containing neurons in Drosophila melanogaster: development and distribution,” J. Comp. Neurol. 268, 414–428 (1988). [CrossRef] [PubMed]

To compare the sensitivity of the different setups their signal-to-noise ratios were analyzed. Each dot in Fig. 6 represents a measurement at one of 500 positions and every color indicates a different setup. The top group of the dots belongs to the SLOTy acquisitions with a reflector, below this SLOTy without reflector is shown and the bottom group belongs to the eOPT acquisitions. To account for different intensities the SNR of the SLOT without a reflector and the SNR of the eOPT measurements were scaled respectively. The illumination irradiance for eOPT was

0.110 \pm 0.012 \frac{W}{m m^{2}}

versus SLOT using

0.086 \pm 0.008 \frac{W}{m m^{2}}

while the effective integration time for one image in eOPT was 2 s versus SLOT with 1.5 s. The resulting intensity factor is 1.7. Linear fits through these measurements suggest a shot noise like behavior. They are displayed in Fig. 6 as areas below the scatter plots containing 86% of the measured results. The slopes of these fits considering the irradiance tolarances are:

\begin{array}{r} {SNR}_{SLOTwR} & = & (0.803 \pm 0.038) \sqrt{I}, \\ {SNR}_{SLOTwR} & = & (0.649 \pm 0.032) \sqrt{I}, \\ {SNR}_{eOPT} & = & (0.084 \pm 0.005) \sqrt{I}, \end{array}

(2)

where I was scaled independently for both setups and is considered to be proportional to the illumination irradiance of the laser. Due to the PMT no absolute values can be calculated. The error margins originate from the standard error of the fitted slopes and the uncertainty in the applied laser intensity.

Fig. 6 SNR estimated by the average standard deviation of successive points in a still image series over a horizontal line of 5 pixels. The SNRs with and without reflector are taken with the SLOT (Fig. 3a), while the OPT SNR is taken with the eOPT setup (Fig. 3c). The SNR without reflector is intensity scaled for comparability. The OPT SNR is intensity scaled too, considering a higher illumination irradiance (

eOPT : 0.110 \pm 0.012 \frac{W}{m m^{2}}

versus

SLOT : 0.086 \pm 0.008 \frac{W}{m m^{2}}

) and effective integration times (eOPT: 2 s versus SLOT: 1.5 s) with the eOPT setup. This results in the scaling factor of

\sqrt{1.7}

. Linear fitted areas below the scatter plots in blue, yellow and orange contain 86% of the measured results. Fluorophore: Cy3; NA = 0.0367.

4. Discussion

Benefits compared to eOPT

SLOTy can be seen as a major improvement of eOPT bearing severe advantages. The first advantage is the illumination of the sample. For the compensation of inhomogeneous illumination in eOPT setups additional effort, for example a diffuser, is needed. Nevertheless, while studying our acquisitions we became aware that a SLOT does not show any deviations in its sensitivity within a wide field of view. The technical explanation lies in the large sensitive PMT area combined with a scanned laser beam intensity staying constant over the whole scan range. The second advantage is the avoidance of reconstruction artifacts from eOPT caused by static pixel to pixel response differences. This includes different quantum efficiencies and hot pixels which will be reconstructed as a ring. Therefore, it is necessary to use a cooled camera and apply algorithms to filter these artifacts [9

J. R. Walls, J. G. Sled, J. Sharpe, and R. M. Henkelman, “Correction of artefacts in optical projection tomography,” Phys. Med. Biol. 50, 4645–4665 (2005). [CrossRef] [PubMed]

,18

B. Münch, P. Trtik, F. Marone, and M. Stampanoni, “Stripe and ring artifact removal with combined wavelet - fourier filtering,” Opt. Express 17, 8567–8591 (2009). [CrossRef] [PubMed]

]. The detectable amount of ring artifacts in our comparative eOPT measurements was quite low, therefore, Fig. 4a represents only selected planes. However, with longer integration times, we experienced a significant increase in ring artifacts. The third advantage is the severalfold higher collection efficiency. The SNRs can be used to calculate the efficiency of the SLOT in comparison to the eOPT setup by assuming only shot noise to be the dominant source for noise. In this case the efficiency is proportional to the square of the SNR. The ratio between the SNR of the SLOT without reflector SNR_SLOTwoR and the SNR of the eOPT setup SNR_eOPT is:

\frac{{SNR}_{SLOTwoR}}{{SNR}_{eOPT}} = 7.73 \pm 0.89,

(3)

and, therefore, the ratio for the corresponding efficiencies to each other is equal to the square of this value:

\frac{{Eff}_{SLOTwoR}}{{Eff}_{eOPT}} = 60 \pm 15 .

(4)

Analogous to equation (3) and (4), the ratio between the SNR of the SLOT with reflector, SNR_SLOTwR, and the SNR of the eOPT setup, SNR_eOPT, leads to the efficiency ratio:

\frac{{Eff}_{SLOTwR}}{{Eff}_{eOPT}} = 91 \pm 22 .

(5)

Thus, for similar results eOPT requires a more than 90 times higher light dose than SLOTy does. To compare the raw photon collection efficiency, the quantum efficiency (QE) of the detectors has to be taken into account. The emCCD camera has a threefold higher QE (see methods section). The possible differences in amplification noise of the camera and the PMT [19

J. B. Pawley, Handbook of Biological Confocal Microscopy , 3rd ed. (Springer Science+Business Media, 2006). [CrossRef]

] will not be taken into account. Therefore these ratios rise to:

{(\frac{{Eff}_{SLOTwoR}}{{Eff}_{eOPT}})}_{QEcorr .} = 180 \pm 45,

(6)

and

{(\frac{{Eff}_{SLOTwR}}{{Eff}_{eOPT}})}_{QEcorr .} = 273 \pm 66 .

(7)

The NA of the eOPT setup is known to be 0.037 ±0.003, therefore, using Eq. (1), the collecting NA of the SLOT without reflector is 0.49 ±0.10. This is in accordance with the estimated NA of 0.45 for the two-lens system. Here we used a reflector gaining another 53% of light resulting in an overall 4.4% ±1.1% (

= Eff (\frac{0.49}{1.47}) \times 1.53

) optical collection efficiency. The total acquisition for 500 images at a resolution of 500x500 including line turnarounds and the rotation time between single projection images takes 21 minutes. Following this calculation, it would take 19 hours to achieve a similar quality with the eOPT setup at the same illumination strength.

Similarities with eOPT

Despite these benefits, SLOTy does not outrun eOPT in all points. First, the resolution of both techniques is practically identical. Both setups use the same optics for the later determination of the image points and therefore have the optical resolution of equivalent OPT setups [20

J. R. Walls, J. G. Sled, J. Sharpe, and R. M. Henkelman, “Resolution improvement in emission optical projection tomography,” Phys. Med. Biol. 52, 2775–2790 (2007). [CrossRef] [PubMed]

, 21

R.-A. Lorbeer, H. Meyer, M. Heidrich, H. Lubatschowski, and A. Heisterkamp, “Applying optical Fourier filtering to standard optical projection tomography,” Proc. SPIE 7570, 75700F (2010). [CrossRef]

]. The SLOT has set its laser beam diameter to the aperture of the eOPT setup (see methods section). The only difference arises due to the discrepancy of the excitation and the emission wavelength. At longer wavelengths, being the emission wavelength in the case of eOPT, the resolution decreases slightly - in the special case of 532 nm excitation wavelength and the chosen fluorescence filter - by 17%. Second, the position accuracy of a laser scanner is not as high as of a camera pixel. This leads to offsets, which can reduce the resolution of the reconstructed image. In our measurements we did not observe any resolution losses due to scanner displacement after applying correction algorithms. To verify this impression, a final proof using phantoms as a resolution target has to be performed and is planned within our group. Third, transmission images which do not need a dramatically high dynamic range are acquired faster with a standard OPT setup rather than with a scanning setup. This is caused by the limited scanning speeds of galvanometric scanners. Nevertheless, tOPT with a laser scanning setup is part of current research [22

N. Krstajić and S. Doran, “Initial characterization of fast laser scanning optical CT apparatus for 3-D dosimetry,” Journal of Physics: Conference Series (Institute of Physics Publishing, 2009), vol. 164, page 012022. [CrossRef]

]. As a last point one needs to mention that because of the reconstruction technique the sample has to be immobilized for a full rotation in eOPT as well as SLOTy. Otherwise, there would be motion artifacts. Consequently, other sources of artifacts not concerning the pixel to pixel differences of the camera chip or photobleaching still remain. Brightness fluctuations of the light source or amplification fluctuations of the detector as well as a refractive index mismatch are to be named as possible sources of error. Additionally, for standard galvanometric scanners at large scanning angles, fanning effects for at least the horizontal dimension have to be tolerated.

Light sheet based microscopy techniques

Finally, an important question remains to be answered: Does SLOTy as an improvement of eOPT compete with other techniques as well? OPT itself was already compared to confocal microscopy, μMRI and OCT by Sharpe et al. [1

]. The increase in detection efficiency SLOTy achieves makes it interesting to be compared to optical sectioning techniques which do not require a reconstruction algorithm. This is non trivial since background photons from all layers contribute to the absolute noise in one projection point. In other words a very selective staining can be measured very efficiently with SLOTy while a non selective staining might obscure small details in its noise. The SNR behavior after the reconstruction is investigated for computed tomography already [7

T. M. Buzug, Computed Tomography from Photon Statistics to Modern Cone-beam CT (Springer-Verlag, 2008).

,23

D. W. Wilson and B. M. W. Tsui, “Noise properties of filtered-backprojection and ML-EM reconstructed emission tomographic images,” IRE Trans. Nucl. Sci. 40, 1198–1203 (1993). [CrossRef]

], but still needs to be analyzed for SLOTy in future work. Therefore, we will now compare the collection efficiencies of the raw images which will have to overcompensate these differences.

Light sheet based microscopy (LSM) techniques are among the most efficient three dimensional linear microscopy techniques [5

J. Huisken and D. Stainier, “Selective plane illumination microscopy techniques in developmental biology,” Development 136, 1963–1975 (2009). [CrossRef] [PubMed]

]. While only reaching an isotropic resolution comparable to the depth resolution of one LSM slice, SLOTy has the ability to collect more photons per acquisition. LSM uses available objectives with a large field of view (FOV), which is needed for large objects. These objectives are technically limited in their NA reducing the collection efficiency. Even though a SLOT might look quite similar to LSM setups it is essential to mention the complete loss of spatial information in the detection path. Especially the SLOT detection system is not bound to classical lens optics. The only challenge lies in reaching a high collection efficiency in a large field of view to reduce the acquisition time as well as the light dose. Due to the independence from the FOV, SLOTy surpasses all known linear techniques since the FOV scarcely changes the sensitivity of the collector. Here we will make a theoretical comparison using Eq. (1) and taking our collection efficiency to be 4.4%. For LSM Huisken et al. used a Fluar 5x, NA 0.25 microscope objective which collects 83% less fluorescence photons than our setup [3

]. To image a whole specimen of D. melanogaster, which is roughly the size of a locust L1 brain, Dodt et al. used a Fluar 2.5x presumably with an NA of 0.12 loosing 96% of the photons in contrast to our SLOT [4

H. U. Dodt, U. Leischner, A. Schierloh, N. Jährling, C. P. Mauch, K. Deininger, J. M. Deussing, M. Eder, W. Zieglgänsberger, and K. Becker, “Ultramicroscopy: three-dimensional visualization of neuronal networks in the whole mouse brain,” Nat. Methods 4, 331–336 (2007). [CrossRef] [PubMed]

]. To figure out the most suitable technique for a given problem, experimental comparison of these two techniques definitely will be on the focus of further publications.

Bioimaging of serotonergic neurotransmitter systems

To demonstrate the widespread applicability of SLOTy for visualizing fluorescent structures in cleared specimens, we used immunocytochemical staining for serotonin in the nervous systems of two insect model organisms [10

J. Thomas, M. Bastiani, M. B. Bate, and C. Goodman, “From grasshopper to Drosophila: a common plan for neuronal development,” Nature 310, 203–207 (1984). [CrossRef] [PubMed]

]. Developmental stages of the fruit fly Drosophila and the locust have been extensively investigated to understand the factors that underlie serotonergic neurotransmitter determination of identified neurons. Recently, serotonin has been shown to be an essential mediator for gregarization [24

M. Anstey, S. Rogers, S. Ott, M. Burrows, and S. Simpson, “Serotonin mediates behavioral gregarization underlying swarm formation in desert locusts,” Science 323, 627–630 (2009). [CrossRef] [PubMed]

], a behavior underlying the swarm formation that desert and migratory locusts are notorious for. SLOTy proved to be a sensitive high resolution method for acquiring a comprehensive account of the serotonin immunofluorescence in the nervous systems of the larger locust and smaller Drosophila larvae. As examples, the maximum intensity projections of Fig. 5 display identified groups of serotonin-immunoreactive wide field neurons (circles in Fig. 5a) in the optic ganglia and descending neurons in the pars intercerebralis (rectangles in Fig. 5c and 5d) that have already been reconstructed from serial sections [16

]. SLOTy resolved also the segmentally arranged groups of serotonergic neurons (Fig. 5b) in the brain and ventral nerve cord of larval Drosophila [17

A. Vallés and K. White, “Serotonin-containing neurons in Drosophila melanogaster: development and distribution,” J. Comp. Neurol. 268, 414–428 (1988). [CrossRef] [PubMed]

]. Treatment of locust larvae with the monoamine re-uptake blocker reserpine reduced the number and immunofluorescence intensity of stained neurons. In combination with rapid image acquisition and a quantitative evaluation of the fluorescence, SLOTy might be applied for monitoring single cell responses in whole organisms during pharmacological drug screening.

5. Conclusion

The results show that SLOTy improves eOPT in several points significantly. Despite of a galvanic scanning mirror system and a PMT our SLOT consists of rather simple and inexpensive components. Yet we were able to show the clear impact of this technique onto illumination and detection properties in terms of homogeneity and a hundredfold higher sensitivity compared to eOPT. This sensitivity does not decrease for even larger objects. Other linear techniques would suffer due to the limited field of view of high NA microscope objectives. Additionally ring artifacts caused by long integration times of the camera chip are avoided.

We already started to use the SLOT for other studies reducing our acquisition time by a factor of 20. Vice versa this enables us to raise the number of samples per study by this factor to increase the statistical significance. Our former eOPT setup with LED excitation needed one night per measurement. Now up to 20 measurements per day can be made with the SLOT prototype. In a nutshell: SLOTy is highly suitable for microscopy experiments where photo-bleaching, phototoxicity, large numbers of large samples and three-dimensional information are of importance.

Acknowledgments

We would like to thank our colleagues and coworkers Alexander Krüger and Thomas Stehr for the short term and uncomplicated lend of materials and space to accomplish this work as fast as possible, as well as the program developers of the free software ImageJ (rsbweb.nih.gov/ij/) and Voreen (www.voreen.org). We want to thank the excellence cluster REBIRTH and the Transregio 37 initiative, funded by the German Research Foundation (DFG).

References

1.	J. Sharpe, U. Ahlgren, P. Perry, B. Hill, A. Ross, J. Hecksher-Sørensen, R. Baldock, and D. Davidson, “Optical projection tomography as a tool for 3D microscopy and gene expression studies,” Science 296, 541–545 (2002). [CrossRef] [PubMed]
2.	J. Sharpe, “Optical projection tomography as a new tool for studying embryo anatomy,” J. Anat. 202, 175–181 (2003). [CrossRef] [PubMed]
3.	J. Huisken, J. Swoger, F. D. Bene, J. Wittbrodt, and E. H. K. Stelzer, “Optical sectioning deep inside live embryos by selective plane illumination microscopy,” Science 305, 1007–1009 (2004). [CrossRef] [PubMed]
4.	H. U. Dodt, U. Leischner, A. Schierloh, N. Jährling, C. P. Mauch, K. Deininger, J. M. Deussing, M. Eder, W. Zieglgänsberger, and K. Becker, “Ultramicroscopy: three-dimensional visualization of neuronal networks in the whole mouse brain,” Nat. Methods 4, 331–336 (2007). [CrossRef] [PubMed]
5.	J. Huisken and D. Stainier, “Selective plane illumination microscopy techniques in developmental biology,” Development 136, 1963–1975 (2009). [CrossRef] [PubMed]
6.	D. Razansky, M. Distel, C. Vinegoni, R. Ma, N. Perrimon, R. W. Köster, and V. Ntziachristos, “Multispectral opto-acoustic tomography of deep-seated fluorescent proteins in vivo,” Nat. Photonics 3, 412–417 (2009). [CrossRef]
7.	T. M. Buzug, Computed Tomography from Photon Statistics to Modern Cone-beam CT (Springer-Verlag, 2008).
8.	J. Sharpe, “Optical projection tomography,” Annu. Rev. Biomed. Eng. 6, 209–228 (2004). [CrossRef] [PubMed]
9.	J. R. Walls, J. G. Sled, J. Sharpe, and R. M. Henkelman, “Correction of artefacts in optical projection tomography,” Phys. Med. Biol. 50, 4645–4665 (2005). [CrossRef] [PubMed]
10.	J. Thomas, M. Bastiani, M. B. Bate, and C. Goodman, “From grasshopper to Drosophila: a common plan for neuronal development,” Nature 310, 203–207 (1984). [CrossRef] [PubMed]
11.	M. Stern, S. Knipp, and G. Bicker, “Embryonic differentiation of serotonin-containing neurons in the enteric nervous system of the locust (Locusta migratoria),” J. Comp. Neurol. 501, 38–51 (2007). [CrossRef] [PubMed]
12.	C. Vinegoni, L. Fexon, P. F. Feruglio, M. Pivovarov, J.-L. Figueiredo, M. Nahrendorf, A. Pozzo, A. Sbarbati, and R. Weissleder, “High throughput transmission optical projection tomography using low cost graphics processing unit,” Opt. Express 17, 22320–22332 (2009). [CrossRef]
13.	S. R. M.D. Abramoff and P. J. Magelhaes, “Image processing with imageJ,” Biophotonics Int. 11, 36–42 (2004).
14.	W. Burger and M. J. Burge, Digital Image Processing: An Algorithmic Introduction using Java (Springer Science+Business Media, 2007).
15.	J. Meyer-Spradow, T. Ropinski, J. Mensmann, and K. Hinrichs, “Voreen: A rapid-prototyping environment for ray-casting-based volume visualizations,” IEEE Comput. Graphics Appl. 29, 6–13 (2009). [CrossRef]
16.	N. Tyrer, J. Turner, and J. Altman, “Identifiable neurons in the locust central nervous system that react with antibodies to serotonin,” J. Comp. Neurol. 227, 313–330 (1984). [CrossRef] [PubMed]
17.	A. Vallés and K. White, “Serotonin-containing neurons in Drosophila melanogaster: development and distribution,” J. Comp. Neurol. 268, 414–428 (1988). [CrossRef] [PubMed]
18.	B. Münch, P. Trtik, F. Marone, and M. Stampanoni, “Stripe and ring artifact removal with combined wavelet - fourier filtering,” Opt. Express 17, 8567–8591 (2009). [CrossRef] [PubMed]
19.	J. B. Pawley, Handbook of Biological Confocal Microscopy , 3rd ed. (Springer Science+Business Media, 2006). [CrossRef]
20.	J. R. Walls, J. G. Sled, J. Sharpe, and R. M. Henkelman, “Resolution improvement in emission optical projection tomography,” Phys. Med. Biol. 52, 2775–2790 (2007). [CrossRef] [PubMed]
21.	R.-A. Lorbeer, H. Meyer, M. Heidrich, H. Lubatschowski, and A. Heisterkamp, “Applying optical Fourier filtering to standard optical projection tomography,” Proc. SPIE 7570, 75700F (2010). [CrossRef]
22.	N. Krstajić and S. Doran, “Initial characterization of fast laser scanning optical CT apparatus for 3-D dosimetry,” Journal of Physics: Conference Series (Institute of Physics Publishing, 2009), vol. 164, page 012022. [CrossRef]
23.	D. W. Wilson and B. M. W. Tsui, “Noise properties of filtered-backprojection and ML-EM reconstructed emission tomographic images,” IRE Trans. Nucl. Sci. 40, 1198–1203 (1993). [CrossRef]
24.	M. Anstey, S. Rogers, S. Ott, M. Burrows, and S. Simpson, “Serotonin mediates behavioral gregarization underlying swarm formation in desert locusts,” Science 323, 627–630 (2009). [CrossRef] [PubMed]

OCIS Codes
(110.0110) Imaging systems : Imaging systems
(110.2970) Imaging systems : Image detection systems
(180.0180) Microscopy : Microscopy
(180.2520) Microscopy : Fluorescence microscopy
(180.5810) Microscopy : Scanning microscopy
(180.6900) Microscopy : Three-dimensional microscopy

ToC Category:
Microscopy

History
Original Manuscript: January 4, 2011
Revised Manuscript: February 24, 2011
Manuscript Accepted: February 24, 2011
Published: March 8, 2011

Virtual Issues
Vol. 6, Iss. 4 Virtual Journal for Biomedical Optics

Citation
Raoul-Amadeus Lorbeer, Marko Heidrich, Christina Lorbeer, Diego F. Ramírez Ojeda, Gerd Bicker, Heiko Meyer, and Alexander Heisterkamp, "Highly efficient 3D fluorescence microscopy with a scanning laser optical tomograph," Opt. Express 19, 5419-5430 (2011)
http://www.opticsinfobase.org/vjbo/abstract.cfm?URI=oe-19-6-5419

Sort: Author | Year | Journal | Reset

References

J. Sharpe, U. Ahlgren, P. Perry, B. Hill, A. Ross, J. Hecksher-Sørensen, R. Baldock, and D. Davidson, “Optical projection tomography as a tool for 3D microscopy and gene expression studies,” Science 296, 541–545 (2002). [CrossRef] [PubMed]
J. Sharpe, “Optical projection tomography as a new tool for studying embryo anatomy,” J. Anat. 202, 175–181 (2003). [CrossRef] [PubMed]
J. Huisken, J. Swoger, F. D. Bene, J. Wittbrodt, and E. H. K. Stelzer, “Optical sectioning deep inside live embryos by selective plane illumination microscopy,” Science 305, 1007–1009 (2004). [CrossRef] [PubMed]
H. U. Dodt, U. Leischner, A. Schierloh, N. Jährling, C. P. Mauch, K. Deininger, J. M. Deussing, M. Eder, W. Zieglgänsberger, and K. Becker, “Ultramicroscopy: three-dimensional visualization of neuronal networks in the whole mouse brain,” Nat. Methods 4, 331–336 (2007). [CrossRef] [PubMed]
J. Huisken and D. Stainier, “Selective plane illumination microscopy techniques in developmental biology,” Development 136, 1963–1975 (2009). [CrossRef] [PubMed]
D. Razansky, M. Distel, C. Vinegoni, R. Ma, N. Perrimon, R. W. Köster, and V. Ntziachristos, “Multispectral opto-acoustic tomography of deep-seated fluorescent proteins in vivo,” Nat. Photonics 3, 412–417 (2009). [CrossRef]
T. M. Buzug, Computed Tomography from Photon Statistics to Modern Cone-beam CT (Springer-Verlag, 2008).
J. Sharpe, “Optical projection tomography,” Annu. Rev. Biomed. Eng. 6, 209–228 (2004). [CrossRef] [PubMed]
J. R. Walls, J. G. Sled, J. Sharpe, and R. M. Henkelman, “Correction of artefacts in optical projection tomography,” Phys. Med. Biol. 50, 4645–4665 (2005). [CrossRef] [PubMed]
J. Thomas, M. Bastiani, M. B. Bate, and C. Goodman, “From grasshopper to Drosophila: a common plan for neuronal development,” Nature 310, 203–207 (1984). [CrossRef] [PubMed]
M. Stern, S. Knipp, and G. Bicker, “Embryonic differentiation of serotonin-containing neurons in the enteric nervous system of the locust (Locusta migratoria),” J. Comp. Neurol. 501, 38–51 (2007). [CrossRef] [PubMed]
C. Vinegoni, L. Fexon, P. F. Feruglio, M. Pivovarov, J.-L. Figueiredo, M. Nahrendorf, A. Pozzo, A. Sbarbati, and R. Weissleder, “High throughput transmission optical projection tomography using low cost graphics processing unit,” Opt. Express 17, 22320–22332 (2009). [CrossRef]
S. R. M. D. Abramoff, and P. J. Magelhaes, “Image processing with imageJ,” Biophotonics Int. 11, 36–42 (2004).
W. Burger and M. J. Burge, Digital Image Processing: An Algorithmic Introduction using Java (Springer Science + Business Media, 2007).
J. Meyer-Spradow, T. Ropinski, J. Mensmann, and K. Hinrichs, “Voreen: A rapid-prototyping environment for ray-casting-based volume visualizations,” IEEE Comput. Graph. Appl. 29, 6–13 (2009). [CrossRef]
N. Tyrer, J. Turner, and J. Altman, “Identifiable neurons in the locust central nervous system that react with antibodies to serotonin,” J. Comp. Neurol. 227, 313–330 (1984). [CrossRef] [PubMed]
A. Vallés and K. White, “Serotonin-containing neurons in Drosophila melanogaster: development and distribution,” J. Comp. Neurol. 268, 414–428 (1988). [CrossRef] [PubMed]
B. Münch, P. Trtik, F. Marone, and M. Stampanoni, “Stripe and ring artifact removal with combined wavelet -fourier filtering,” Opt. Express 17, 8567–8591 (2009). [CrossRef] [PubMed]
J. B. Pawley, Handbook of Biological Confocal Microscopy, 3rd ed. (Springer Science + Business Media, 2006). [CrossRef]
J. R. Walls, J. G. Sled, J. Sharpe, and R. M. Henkelman, “Resolution improvement in emission optical projection tomography,” Phys. Med. Biol. 52, 2775–2790 (2007). [CrossRef] [PubMed]
R.-A. Lorbeer, H. Meyer, M. Heidrich, H. Lubatschowski, and A. Heisterkamp, “Applying optical Fourier filtering to standard optical projection tomography,” Proc. SPIE 7570, 75700F (2010). [CrossRef]
N. Krstajíc, and S. Doran, “Initial characterization of fast laser scanning optical CT apparatus for 3-D dosimetry,” Journal of Physics: Conference Series (Institute of Physics Publishing, 2009), vol. 164, page 012022. [CrossRef]
D. W. Wilson and B. M. W. Tsui, “Noise properties of filtered-backprojection andML-EMreconstructed emission tomographic images,” IRE Trans. Nucl. Sci. 40, 1198–1203 (1993). [CrossRef]
M. Anstey, S. Rogers, S. Ott, M. Burrows, and S. Simpson, “Serotonin mediates behavioral gregarization underlying swarm formation in desert locusts,” Science 323, 627–630 (2009). [CrossRef] [PubMed]

Annu. Rev. Biomed. Eng.

J. Sharpe, “Optical projection tomography,” Annu. Rev. Biomed. Eng. 6, 209–228 (2004). [CrossRef] [PubMed]

Biophotonics Int.

S. R. M. D. Abramoff, and P. J. Magelhaes, “Image processing with imageJ,” Biophotonics Int. 11, 36–42 (2004).

Development

J. Huisken and D. Stainier, “Selective plane illumination microscopy techniques in developmental biology,” Development 136, 1963–1975 (2009). [CrossRef] [PubMed]

IEEE Comput. Graph. Appl.

J. Meyer-Spradow, T. Ropinski, J. Mensmann, and K. Hinrichs, “Voreen: A rapid-prototyping environment for ray-casting-based volume visualizations,” IEEE Comput. Graph. Appl. 29, 6–13 (2009). [CrossRef]

IRE Trans. Nucl. Sci.

D. W. Wilson and B. M. W. Tsui, “Noise properties of filtered-backprojection andML-EMreconstructed emission tomographic images,” IRE Trans. Nucl. Sci. 40, 1198–1203 (1993). [CrossRef]

J. Anat.

J. Sharpe, “Optical projection tomography as a new tool for studying embryo anatomy,” J. Anat. 202, 175–181 (2003). [CrossRef] [PubMed]

J. Comp. Neurol.

N. Tyrer, J. Turner, and J. Altman, “Identifiable neurons in the locust central nervous system that react with antibodies to serotonin,” J. Comp. Neurol. 227, 313–330 (1984). [CrossRef] [PubMed]
A. Vallés and K. White, “Serotonin-containing neurons in Drosophila melanogaster: development and distribution,” J. Comp. Neurol. 268, 414–428 (1988). [CrossRef] [PubMed]
M. Stern, S. Knipp, and G. Bicker, “Embryonic differentiation of serotonin-containing neurons in the enteric nervous system of the locust (Locusta migratoria),” J. Comp. Neurol. 501, 38–51 (2007). [CrossRef] [PubMed]

Nat. Methods

H. U. Dodt, U. Leischner, A. Schierloh, N. Jährling, C. P. Mauch, K. Deininger, J. M. Deussing, M. Eder, W. Zieglgänsberger, and K. Becker, “Ultramicroscopy: three-dimensional visualization of neuronal networks in the whole mouse brain,” Nat. Methods 4, 331–336 (2007). [CrossRef] [PubMed]

Nat. Photonics

D. Razansky, M. Distel, C. Vinegoni, R. Ma, N. Perrimon, R. W. Köster, and V. Ntziachristos, “Multispectral opto-acoustic tomography of deep-seated fluorescent proteins in vivo,” Nat. Photonics 3, 412–417 (2009). [CrossRef]

Nature

J. Thomas, M. Bastiani, M. B. Bate, and C. Goodman, “From grasshopper to Drosophila: a common plan for neuronal development,” Nature 310, 203–207 (1984). [CrossRef] [PubMed]

Opt. Express

Phys. Med. Biol.

J. R. Walls, J. G. Sled, J. Sharpe, and R. M. Henkelman, “Correction of artefacts in optical projection tomography,” Phys. Med. Biol. 50, 4645–4665 (2005). [CrossRef] [PubMed]
J. R. Walls, J. G. Sled, J. Sharpe, and R. M. Henkelman, “Resolution improvement in emission optical projection tomography,” Phys. Med. Biol. 52, 2775–2790 (2007). [CrossRef] [PubMed]

Proc. SPIE

R.-A. Lorbeer, H. Meyer, M. Heidrich, H. Lubatschowski, and A. Heisterkamp, “Applying optical Fourier filtering to standard optical projection tomography,” Proc. SPIE 7570, 75700F (2010). [CrossRef]

Science

M. Anstey, S. Rogers, S. Ott, M. Burrows, and S. Simpson, “Serotonin mediates behavioral gregarization underlying swarm formation in desert locusts,” Science 323, 627–630 (2009). [CrossRef] [PubMed]
J. Sharpe, U. Ahlgren, P. Perry, B. Hill, A. Ross, J. Hecksher-Sørensen, R. Baldock, and D. Davidson, “Optical projection tomography as a tool for 3D microscopy and gene expression studies,” Science 296, 541–545 (2002). [CrossRef] [PubMed]
J. Huisken, J. Swoger, F. D. Bene, J. Wittbrodt, and E. H. K. Stelzer, “Optical sectioning deep inside live embryos by selective plane illumination microscopy,” Science 305, 1007–1009 (2004). [CrossRef] [PubMed]

Other

T. M. Buzug, Computed Tomography from Photon Statistics to Modern Cone-beam CT (Springer-Verlag, 2008).
J. B. Pawley, Handbook of Biological Confocal Microscopy, 3rd ed. (Springer Science + Business Media, 2006). [CrossRef]
W. Burger and M. J. Burge, Digital Image Processing: An Algorithmic Introduction using Java (Springer Science + Business Media, 2007).
N. Krstajíc, and S. Doran, “Initial characterization of fast laser scanning optical CT apparatus for 3-D dosimetry,” Journal of Physics: Conference Series (Institute of Physics Publishing, 2009), vol. 164, page 012022. [CrossRef]

Cited By

Alert me when this paper is cited

OSA is able to provide readers links to articles that cite this paper by participating in CrossRef's Cited-By Linking service. CrossRef includes content from more than 3000 publishers and societies. In addition to listing OSA journal articles that cite this paper, citing articles from other participating publishers will also be listed.

Click here to see a list of articles that cite this paper