Label-free second-harmonic phase imaging of biological specimen by digital holographic microscopy.

Abstract

We have previously developed a new way for nonscanning second-harmonic generation (SHG) microscopy [Opt. Lett. 34, 2450 (2009)]. Based on digital holography, this technique captures, in single-shot hologram acquisition, both the amplitude and the phase of a coherent SHG radiation, which makes possible second harmonic phase microscopy. In this work, we present holographic SHG phase microscopy of a label-free biological tissue and discuss its added value to SHG microscopy.

Full text

Page 1

Label-free second-harmonic phase imaging of biological
specimen by digital holographic microscopy
Etienne Shaffer,1,* Corinne Moratal,1Pierre Magistretti,1Pierre Marquet,1,2and Christian Depeursinge1
1École Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland
2Département de Psychiatrie–CHUV, Site de Cery, 1008 Prilly, Lausanne, Switzerland
*Corresponding author: etienne.shaffer@epfl.ch
Received August 27, 2010; revised November 5, 2010; accepted November 8, 2010;
posted November 10, 2010 (Doc. ID 134136); published December 3, 2010
We have previously developed a new way for nonscanning second-harmonic generation (SHG) microscopy [Opt.
Lett. 34, 2450 (2009)]. Based on digital holography, this technique captures, in single-shot hologram acquisition,
both the amplitude and the phase of a coherent SHG radiation, which makes possible second harmonic phase
microscopy.In this work,we presentholographic SHGphasemicroscopyof a label-freebiological tissueanddiscuss
its added value to SHG microscopy.© 2010 Optical Society of America
OCIS codes:090.1995, 180.4315, 110.0180.
Second-harmonic generation (SHG) has emerged as a
promising imaging technique for biological applications
[1–10]. Originally developed
nonlinear microscopy (multiphoton excitation fluores-
cence), it contributed to the emergence of coherent non-
linear microscopy (e.g., higher harmonic generation,
sum- or difference-frequency generation, wave mixing,
coherent anti-Stokes Raman scattering). One relatively
unexploited advantage of SHG—and, more generally,
of coherent nonlinear microscopy—lies in its coherent
nature providing both the amplitude and the phase infor-
mation that can be efficiently retrieved, e.g., by means of
digital holography. Indeed, holographic harmonic micro-
scopy has recently been reported [11] and has led to
quantitative exploitation of the SHG phase [12,13]. The
first holographically recorded SHG intensity images of
label-free biological specimens were reported this very
year [14]. We report in this Letter on the first ever (to
our knowledge) application of SHG phase microscopy,
targeted to a label-free biological specimen. We show
that the SHG phase has the potential to provide key in-
formation for better understanding of material structure,
here collagen, and could lead to quantitative assessments
of refractive index and/or spatial distribution of second-
harmonic scatterers.
Holography consists in using a reference wave to en-
code the complex diffraction pattern of an object into an
interference image called a hologram. Digital holograms
are recorded with a digital camera and numerically re-
constructed with a computer. Images numerically recon-
structed from digital holograms are complex in nature
and contain quantitative information on both the ampli-
tude and the phase of the object wave [15–17]. In this
work, single-shot amplitude and phase image acquisition
is made possible by the off-axis holographic configura-
tion. For details on the hologram reconstruction tech-
nique used and on its application to second-harmonic
fields, see [15,18,12].
Our holographic SHG microscope (Fig. 1) is a Mach–
Zehnder interferometer consisting of an object arm (O)
and a reference arm (R). In O, the laser beam is loosely
(≈30 μm FWHM) focused on the specimen. SHG occurs
in the illuminated region, and the SHG object wave is col-
lected alongside the fundamental illumination by a 63×,
alongsideincoherent
0.75 NA microscope objective and imaged on the detec-
tor. In R, the laser beam is focused on a β-barium borate
frequency doubler crystal (FDC) that generates the sec-
ond-harmonic reference wave, which is then collimated,
expanded, and projected on the detector, where it inter-
feres with the object wave. An optical delay line matches
the optical path length of R to that of O, ensuring that
each pulse interferes with itself at the detector. The light
source is a 800 nm wavelength Ti:sapphire laser, with no
dispersion precompensation module, delivering >150 fs
pulses at a repetition rate of 80 Mhz. The measured en-
ergy in the specimen plane is 40 nJ · pulse−1. For bright-
field images, the femtosecond laser was replaced by a
light-emitting device, whose wavelength matches that
of the SHG signal, so that the optical resolution remains
the same. Polarizers on rotation mounts could be in-
serted before the condenser lens C and in the infinity
space of the microscope objective for cross-polarization
bright-field images. Finally, the detector, designed
for fluorescence imaging, is a 12-bit CCD, with 6:45 μm
pixels, working at 8.3 frames/s, with an exposure time of
120 ms. At all time, bandpass filters ensured that only
light of wavelength 400 ? 20 nm reached the CCD.
Specimens consist of 5-μm-thick coronal sections of
mouse tail, fixed with paraformaldehyde 4% and em-
bedded in paraffin. Some sections were stained with
Fig. 1.
beam expander; DL, optical delay line; C, condenser lens; S,
specimen; MO, microscope objective; M, mirror; TL, tube lens;
F, bandpass filter; L, lens; FDC, β-barium borate frequency dou-
bler crystal.
Experimental setup schematics: BS, beam splitter; BE,
4102 OPTICS LETTERS / Vol. 35, No. 24 / December 15, 2010
0146-9592/10/244102-03$15.00/0© 2010 Optical Society of America

Page 2

histological dyes revealing collagen (Sirius Red, Goldner,
and Masson Trichrome), while others were kept un-
stained. The purpose of stained sections is to identify col-
lagen fibers in the dermis, while the unstained sections
served for label-free SHG measurements. We chose such
specimens, since it is well-known that collagen generates
strong second-harmonic signals [1,5–7,9,10].
In a first step, the stained sections were examined with
a bright-field microscope to identify collagen structures
(e.g., collagen fibers constituting connective tissue) and
then with our holographic SHG microscope to verify that
collagen fibers do generate second-harmonic signals. The
results, not presented in this Letter, confirmed that sec-
ond harmonic was generated by the collagen fibers, as
expected. In a second step, we made similar measure-
ments on an unstained section to validate that the
SHG signal did not come from the histological stains
but was instead intrinsic to the specimen (Fig. 2).
While a bright-field image of the specimen [Fig. 2(a)]
gives a good idea of what the specimen looks like, in
terms of its geometry, it provides no way of differentiat-
ing between the materials. In comparison, a cross-
polarized bright-field image of the same region [Fig. 2(b)]
highlights birefringent materials by mapping the phase
retardation for a given incident polarization. However,
since the specimen is relatively thin (5 μm), birefrin-
gence effects remain rather weak. Finally, holographic
SHG amplitude and phase images [Figs. 2(c) and 2(d)],
reconstructed from the same hologram, reveal the pre-
sence of collagen. All images of Fig. 2 were recorded
at the same imaging wavelength and with the same mi-
croscope objective. Therefore, fine structure seen in
SHG images but not in bright-field images are not a mat-
ter of resolution but of imaging contrast and suggest that
second-harmonic imaging provides specific information,
impossible to retrieve with bright-field microscopy.
Right from the start, it is worth noting that the SHG
amplitude is linked to the Gaussian profile of the illumi-
nation, implying that some normalization would be
required before quantitative pixel-to-pixel comparisons
can be made. While this drawback of nonscanning SHG
microscopy could, of course, be corrected by beam
homogenizing optics, it is interesting that the SHG phase
[Fig. 2(d)] does not depend on the illumination intensity
and, once reconstructed, can be quantitatively analyzed
as is, all over the field of view. Another observation is
that the phase of SHG seems to have a better signal-
to-noise ratio (SNR) than its amplitude, as can be seen
near the edges of the images where almost no second
harmonic is generated but where the phase nevertheless
reveals structures that are impossible to distinguish in
amplitude contrast. Finally, it is easy to distinguish re-
gions where no second harmonic is generated, as they
result in a physically meaningless randomly distributed
SHG phase. In case this disturbs the observer, one could
imagine a filter, based on thresholding of the SHG ampli-
tude, that would remove random phase fluctuations.
Where the SHG phase is defined, i.e., where SHG oc-
curs, it is related to the phase of the fundamental wave at
the location zSHGof nonlinear interaction, as well as on
the optical path length (at SHG wavelength) from that
point to the detector. The SHG phase φ can thus be ex-
pressed in terms of refractive index n, fundamental
wavelength λ, and axial coordinate z as
Z
where zIand zFare the coordinates of the light source
and the detector, respectively. Unfortunately, the SHG
phase itself is not enough to differentiate changes in the
refractive index from changes in axial coordinates of
SHG, for it depends on the coupled effects of both. This
frustrating problem will appear somewhat familiar to the
digital holographists who encountered a similar chal-
lenge in linear holography and came up with methods
for decoupling refractive index and specimen thickness.
Such methods include changing the known value of re-
fractive index of the surrounding medium [19], confining
the specimen in a microchannel of known depth [17], or
using air bubbles in the medium surrounding the speci-
men to determine its refractive index [16]. Readers will
therefore understand that using the SHG phase to quan-
titatively assess physical properties like refractive index
and axial coordinates of SHG is a big challenge, beyond
the scope of this Letter, that will have to be addressed in
future works. It has, however, already been shown that
for SHG-emitting nanoparticles located in a nondisper-
sive medium, the SHG phase can be directly related to
the axial position of the particle at a nanometer-scale
precision [13].
φ ¼
zSHG
zI
2π
λnðz;λÞdz þ
Z
zF
zSHG
2π
λ=2nðz;λ=2Þdz;
ð1Þ
Fig. 2.
phase retardation caused by the birefringence of collagen. (c), (d) SHG amplitude (normalized) and phase (wrapped) reconstructed
from a single hologram. All images present the same region of the specimen and scale bars are 10 μm. Media 1, pseudocolor overlay
of (b) and (c).
Mouse tail dermis and epidermis (bottom left corner). (a) Bright field image. (b) Cross-polarization image showing
December 15, 2010 / Vol. 35, No. 24 / OPTICS LETTERS4103

Page 3

Even if we cannot yet relate the SHG phase to quanti-
tative changes of physical properties, it still has a lot to
offer to SHG microscopy. Reporting this year in Chinese
Optics Letters, Xu et al. [10] propose a quasi-crystal
model of collagen to explain the results of their
polarization-based investigation of its SHG. In their work,
they suggest that single molecule SHG would be too
weak to account for the detected SHG signal and that co-
herent scattering had to play an important role. They add
that a completely turbid medium would not offer the op-
portunity of coherent SHG scattering. The observation of
the phase of the SHG signal we report here tends to sup-
port this theory. Indeed, for coherent scattering to occur,
some phase-matching conditions must be satisfied.
Otherwise, second harmonic generated at different
depths would not add up coherently, and the SHG
phase would be randomly distributed. The well-defined
and continuous SHG phase of Fig. 2(d) is not compatible
with such a model of single-molecule SHG in a comple-
tely turbid collagen medium and instead hints at a coher-
ent scattering process.
Also supporting this coherent scattering process are
the complementary cross-polarization [Fig. 2(b)] and ho-
lographic SHG images [Figs. 2(c) and 2(d)]—see Media 1
for overlay. Where one provides maximal contrast, the
other provides almost none. We believe that this can
be understood by reasoning in terms of phase matching
conditions for coherent scattering. It appears that for the
given incident polarization of the femtosecond laser illu-
mination, phase matching is verified only where no retar-
dation is observed in Fig. 2(b).
In conclusion, we have reported what we believe to be
the first label-free, holographic SHG phase contrast
images of biological specimen. Off-axis digital holo-
graphic microscopy is a nonscanning, single-shot image
acquisition technique, limited only by the camera frame
rate (and available SHG signal). It is therefore especially
suited for real-time imaging and truly exploits the instan-
taneous response of SHG. It should be obvious that, to
this day, nonscanning holographic SHG cannot compete
with scanning SHG microscopy in terms of sensitivity.
Holographic SHG offers coherent amplification, low sen-
sitivity to shot noise and very high phase SNR, but, in the
end, it all comes back to a matter of available light
sources and detectors. Much more powerful ultrafast la-
ser sources are needed to provide peak powers compar-
able to those offered by scanning microscopes but over a
large field of views. In addition, very sensitive array de-
tectors are required to compete with photomultiplier
tubes. It is no wonder that the increasing enthusiasm
around holographic SHG microscopy is fueled by the re-
cent developments of both ultrafast lasers and electron-
multiplying digital cameras.
The authors would like to thank Sandor Kasas for fruit-
ful discussions. This work was financially supported in
part by the Swiss National Competence Center in Biome-
dical Imaging (NCCBI) and by the Swiss National Science
Foundation (SNSF), grant no. 205320-130543.
References
1. I. Freund and M. Deutsch, Opt. Lett. 11, 94 (1986).
2. Y. C. Guo, P. P. Ho, H. Savage, D. Harris, P. Sacks,
S. Schantz, F. Liu, N. Zhadin, and R. R. Alfano, Opt. Lett.
22, 1323 (1997).
3. P. J. Campagnola, M. D. Wei, A. Lewis, and L. M. Loew,
Biophys. J. 77, 3341 (1999).
4. L. Moreaux, O. Sandre, M. Blanchard-Desce, and J. Mertz,
Opt. Lett. 25, 320 (2000).
5. A. Zoumi, A. Yeh, and B. J. Tromberg, Proc. Natl. Acad. Sci.
USA 99, 11014 (2002).
6. W. R. Zipfel, R. M. Williams, R. Christie, A. Y. Nikitin,
B. T. Hyman, and W. W. Webb, Proc. Natl. Acad. Sci.
USA 100, 7075 (2003).
7. S. J. Lin, R. J. Wu, H. Y. Tan, W. Lo, W. C. Lin, T. H. Young,
C. J. Hsu, J. S. Chen, S. H. Jee, and C. Y. Dong, Opt. Lett. 30,
2275 (2005).
8. L. Fu, X. S. Gan, and M. Gu, Opt. Lett. 30, 385
(2005).
9. D. Ait-Belkacem, A. Gasecka, F. Munhoz, S. Brustlein, and
S. Brasselet, Opt. Express 18, 14859 (2010).
10. P. Xu, E. Kable, C. J. R. Sheppard, and G. Cox, Chinese
Optics Letters 8, 213 (2010).
11. Y. Pu, M. Centurion, and D. Psaltis, Appl. Opt. 47,
A103 (2008).
12. E. Shaffer, N. Pavillon, J. Kühn, and C. Depeursinge, Opt.
Lett. 34, 2450 (2009).
13. E. Shaffer, P. Marquet, and C. Depeursinge, Opt. Express
18, 17392 (2010).
14. O. Masihzadeh, P. Schlup, and R. A. Bartels, Opt. Express
18, 9840 (2010).
15. E. Cuche, P. Marquet, and C. Depeursinge, Appl. Opt. 38,
6994 (1999).
16. B. Kemper, D. Carl, J. Schnekenburger, I. Bredebusch, M.
Schäfer, W. Domschke, and G. von Bally, J. Biomed. Opt.
11, 034005 (2006).
17. N. Lue,G. Popescu,T. Ikeda, R.Dasari, K. Badizadegan,and
M. Feld, Opt. Lett. 31, 2759 (2006).
18. U. Schnars and W. P. O. Juptner, Meas. Sci. Technol. 13,
R85 (2002).
19. B. Rappaz, F. Charriere, C. Depeursinge, P. J. Magistretti,
and P. Marquet, Opt. Lett. 33, 744 (2008).
4104OPTICS LETTERS / Vol. 35, No. 24 / December 15, 2010