Biomedical Engineering Reference
In-Depth Information
For laser scanning microscopy, the relative distance between the optical components is strictly regu-
lated. In the ideal case shown in Figure 4.6, they are spaced by the focal length of the lenses except for
the dichroic mirrors and emission filters, which are located in the infinity space. This arrangement is
important to maintain both proper beam path (propagation directionality) and correct beam profile
(size and divergence). The laser beam is expanded by the scan lens and the tube lens to fill the micro-
scope objective back aperture for maximum resolution as the beam can then be focused into the tightest
spot (1′ or 2′) at the sample plane. One crucial aspect is that when the scan mirror moves the laser beam
between 1 and 2, the laser beam actually remains stationary at the position of the scanning mirror,
the back focal plane of the objective and the condenser focal plane, and meanwhile, the emitting rays
remain stationary at the objective (or condenser) focal plane and the detector. Such stationary status
has to be maintained as the objective back aperture, the condenser aperture, and the detector optical
window are beam-restrictive components and violation of this condition will result in the loss of field
of view.
The NLO microscopes all have their own special characteristics, despite the common aspects outlined
earlier, and this is reflected in the apparatus designs. SHG, TPEF, and THG microscopes all use single
laser beam illumination and can be easily implemented as multiple modalities of one setup. However,
they are different in the detected emission wavelength and switching modalities often involve only care-
ful emission filter (EF in Figure 4.6) selection. With a pump laser beam at λ, SHG and THG require a
narrow-band interference filter at λ/2 and λ/3, respectively, while TPEF uses a broad-band filter between
λ/2 and λ or multiple narrower-band filters for the case of multiple fluorophores.
Because THG is a nonlinear process of higher order with a smaller interaction cross section, it often
requires higher pump laser power, which may lead to undesirable sample damage. Additionally, the
third harmonic of the Ti:sapphire output is usually deep in the UV (<333 nm) and can be absorbed and
scattered by the sample, and cannot be efficiently transmitted by conventional glass optical components.
All these make THG signal detection much harder. As a result, an optical oscillator (OPO) is commonly
used to shift the pump laser wavelength toward infrared, which is less invasive [24,32], and this also
moves the third harmonic toward the visible spectrum. The other laser source used for THG has been
the Cr:fosterite laser at ~1240 nm [33].
CARS and SRS imaging are more complex than the previous modalities because in addition to
proper optical filter selection they involve two different laser wavelengths, a pump beam at ω
1
and a
Stokes beam at ω
2
, which need to be combined spatially and before they are introduced to the micro-
scope. Owing to the coherent nature of CARS and SRS, the pulses from the two lasers need to be syn-
chronized so that both wavelengths can be present simultaneously at the same location in the sample
to produce the nonlinear signal. For this purpose, either outputs from two synchronized Ti:sapphire
lasers are used [34] or a single OPO (along with the pump laser) can be used to provide both wave-
lengths with the required timing [27]. Signal detection for CARS can still be accomplished in the same
way as SHG except that now a narrow filter at a shorter wavelength than the input wavelengths is used
(ω
AS
= ω
1
+ ω
vib
> ω
1
> ω
2
).
SRS is very unique among the NLO methods under discussion in terms of signal detection. All these
other NLO methods record the generated nonlinear signal at a new wavelength, and thus the signal
is background free. By comparison, SRS deals with small intensity variation (~10
−6
) of the input laser
beams (Figure 4.7), which means that the resulting SRS signal (Raman gain or loss) is a very weak signal
on top of a large background. In such situations, it is necessary to use a lock-in amplification system
for signal recovery [35]. Here, the intensity of one of the input laser beams is modulated with an optical
chopper, at a reference frequency (~10 kHz) while the intensity of the other beam is monitored using a
photodiode. Because the two laser beams are coupled at the sample via the SRS process, the intensity
variation of the second laser beam also carries the reference frequency and can be retrieved by the
lock-in amplifier, which is only sensitive to the signal at the reference frequency. One should note that
the detector used for SRS imaging is different from those of the other NLO imaging techniques, which
are almost always PMTs detecting low single levels on top of zero background. For SRS, a photodiode
