|
Design issues relating to bioactive devices for regenerative
medicine reflect the spatial and chemical complexity of biological
and materials issues and their interactions. Tools developed for the
purpose of aiding understanding of these systems must have sufficient
discriminatory power to sort out this complexity. We have introduced
a relatively simple broadband spectroscopic microscopy based on CARS
that can be used to rapidly acquire volumetric, chemically-specific
images with submicrometer resolution. |
|
Marcus T. Cicerone |
|
|
Biological research in general could be significantly aided by a
high-resolution, chemically sensitive volumetric imaging method that
allows rapid, non-invasive study of processes on the tissue, cellular
and sub-cellular levels. The need for such a method is particularly
acute in the field of tissue engineering, where cycles of cell-seeding
and analysis can take months. Standard analysis methods are labor
intensive and destructive, so that multiple endpoint studies must
be performed and temporal correlation assumed. Non-invasive micro-spectroscopic
imaging could alleviate many problems associated with the evaluation
of tissue scaffolds both by allowing continuity of analysis on a single
scaffold construct, and by maintaining spatio-temporal information. |
|
Our vision is the creation of a non-invasive microscopic imaging
technique that rapidly discriminate between an arbitrary number of
chemically distinct structures or species in a spatially resolved
way within a biological system. We envision that this would be done
without staining or otherwise intrusively manipulating the system.
We expect that such a microscopy tool would have a significant impact
on the way in-vitro tissue-engineering studies are carried out. |
|
While our vision may sound fanciful, the basis for such powerful
and generally applicable chemical discrimination was established many
decades ago via vibrational spectroscopy. Our contribution has been
to marry broadband vibrational spectroscopy, with its inherent chemical
resolving power, to a nonlinear, multiphoton microscopy (CARS microscopy),
which has already been demonstrated to have significant utility for
non-invasively imaging biological systems, even in the absence of
significant chemical resolving power. |
|
Microscopic coherence anti-Stokes Raman scattering (µCARS)
can answer important imaging needs of the biological research community,
including relative non-invasiveness, rapidity, and chemical specificity.
CARS provides chemical specificity through its intrinsic sensitivity
to molecular vibrational transitions. CARS is sensitive to the third
order polarizability, ?(3), which has nonresonant and resonant components,
the latter being related to the Raman scattering cross section. mCARS
therefore uses Raman (vibrational) susceptibility as a contrast mechanism,
but is approximately 103 more efficient than spontaneous Raman scattering.
|
|
|
Figure 1: Energy level diagram for single frequency CARS
process (solid vertical arrows) and broadband CARS (solid and dashed
vertical arrows). |
|
The solid vertical lines of Figure 1 represent transitions made
by molecules in the "classic" CARS scheme. Molecules in
a sample of interest are promoted to a specific energy E = h (wp -
wS), where wp and wS are the frequencies of the near infrared pump
and Stokes light pulses, respectively, and h is h/2p where h is Planck's
constant. If that energy corresponds to a particular excited vibrational
state in the molecule, that state is populated. A subsequent probe
photon (also wp) promotes the vibrationally exited molecule to another
(higher) virtual state, from which it relaxes and emits an anti-Stokes
photon at a specific frequency, waS. The flux of anti-Stokes photons
detected is proportional to the number of molecules initially prepared
in the targeted vibrationally excited state. To date, this "classic"
mCARS scheme has been used to acquire only single-frequency or narrowband
(200 cm-1) vibrational information, and thus, chemical specificity
is limited to only one or two species simultaneously. |
|
We have developed, and demonstrated for the first time, a CARS microscope
that yields vibrational spectra over a 2500 cm-1 range. This method
can give chemically specific information for multiple species simultaneously.
Furthermore, information about subtle changes in such things as metabolic
state or phenotype of cells, which are of interest when considering
cell-scaffold interactions, can be extracted from these broadband
vibrational spectra. |
|
Figure 1 illustrates that the key to increased bandwidth of vibrational
sensitivity in CARS is an increased bandwidth of the Stokes light.
As our Stokes light source, we have used continuum light generated
from ultrafast light pulses passed through a tapered optical fiber.
Our apparatus uses the 785 nm output of a Ti:Sapphire oscillator (200
mW, 76MHz, 150 fs pulses) fed into a tapered nonlinear fiber, 9 mm
long with a 2 mm diameter waist to create a continuum that spans the
wavelength range (500 nm to 1100 nm). A spectrally narrow portion
of the same oscillator output is used for the pump and probe light. |
|
Figure 2a is a first demonstration of this method; it is an image
of a tertiary polymer containing equal parts of polystyrene (PS),
poly(methyl methacrylate) (PMMA) and poly(ethylene terephthalate)
(PET); the polymer blend was annealed at 250 °C for 120 s, then
melt-pressed at this temperature. Spatially-resolved spectra were
obtained by broadband mCARS, and pseudo-colors were assigned to each
pixel based on the primary polymer component - red, green, and yellow
for PMMA, PS, and PET respectively. The reference spectra used for
identification are plotted in Figure 2b, with an arbitrary vertical
shift applied for clarity. |
|
The primary polymer component within an image pixel (i.e., the pixel
"identity") was determined as follows. First, the CARS spectra
from the image and the reference spectra were systematically normalized
by the nonresonant background. Next, a dot product was calculated
between the spectrum belonging to the pixel in question and the sections
of each reference spectrum enclosed by the hatched boxes in Figure
2b. The dot product yielding the greatest value of the three determined
the assigned polymer identity for that pixel. The reference spectra
sections were amplitude-scaled such that a dot product of each with
itself gave the same value. Using this procedure, we could assign
a polymer identity to each pixel with a standard uncertainty of 0.01. |
|
|
Figure 2: Panel a: Pseudo-color image, generated
with broadband CARS microscopy, of phase-separated polymer blend including
equal parts of PMMA, PS, and PET. The colors red, green and yellow
correspond to PMMA, PS, and PET respectively. Panel b: Reference spectra
used for computing dot products with spectra from each pixel in panel
a. The hatched boxes indicate spectral regions that were used for
computing dot products (see text for discussion). |
|
An account of this work has been accepted for publication in Optics
Letters. We will also report on the new method at the 2004 "Vibrational
Spectroscopy" and "Biomedical Imaging" Gordon Research
Conferences (2004), and the August 2004 American Chemical Society
National Meeting. |
|
We gratefully acknowledge Dr. Lee Richter for his interest and input.
We also thank Dr. John Lawall and Dr. Tsvetelina Petrova for helpful
guidance with regards to continuum generation. This work was funded
in part by NIH Grant # 1 R21 EB002468-01. |
|
|
For More Information on this Topic |
|
Tak W. Kee (Polymers Division,
NIST) |
|
|