CARS Microscope - Label Free Imaging: The Best of Both Worlds - Researchers at King’s College London have developed a simple and robust optical way of acquiring spontaneous Raman spectra with the speed and sensitivity of CARS.
The instrument gives linear, quantitative, NRB-free signals for any laser system capable of generating CARS, including both narrowband and broadband systems. Broadband micro-Raman spectroscopy using supercontinuum laser sources can therefore give spectra equivalent to conventional linear Raman spectroscopy, but with significantly enhanced signal levels, and existing narrowband microscopes with picosecond OPO’s can give quantitative measurements of chemical concentrations along with their spatial distribution.
Advantages:
- NRB-free spectrum in a single exposure
- Simple optical setup, requires only passive polarisation optics
- Inherently robust: common-path interferometry, with balanced homodyne detection
- Suitable for broadband spectral measurements, and high-speed, narrowband measurements
- Ability to image without fluorescent labelling/ staining
- Excellent 3D resolution
- Chemical selectivity
Technical development:
A proof of principle prototype has been built and tested
Looking to license the technology to develop and launch a commercial product
Patent status:
PCT Application: WO 2012/017201
Granted in Europe: EP 2601514
Pending in the US and Japan
Examples of Images
Figure 2 Hyperspectral imaging of a sample containing both polystyrene (PS) and polymethylmethacrylate (PMMA) beads of 1 micrometre diameter. Scale bar: 5 micrometre; pixel size: 300 nm, 64×64 pixels, 100 ms/pixel. Multivariate analysis was performed (Pearson clustering), and the component spectra are shown next to their corresponding intensity maps. The two bead populations can be clearly distinguished; left panel: PMMA beads; right panel: PS. Also shown for PS is the corresponding equivalent polarised spontaneous Raman spectrum (green).
Figure 3 Hyperspectral image of a live HeLa cell nucleus. 100×100 pixels, 100 ms/pixel; scale bar 5 micrometre; pixel size 300nm. (a) NRB image. (b) Primary principal component image. (c) Map of peak shift of 1450 cm-1 vibrational band. Higher values (red) correspond to higher CH3 to CH2 bond fraction. Nucleoli and lipid droplets can be clearly distinguished.
Faster vibrational spectroscopy
Raman scattering provides a powerful optical route to obtain chemically specific information and is widely used in biology, chemistry and materials science. Molecular vibrations are probed using optical frequencies, allowing chemical information to be acquired with high spatial resolution. However, the scattering process is extremely weak, leading to long acquisition times and often requiring severe trade-offs to be made between spectral, temporal, and spatial resolution.
CARS (Coherent anti-Stokes Raman Scattering) microscopy holds the promise of much higher sensitivity but its widespread application and commercialisation has been prevented by the need to deal with the concomitant non-resonant background (NRB). This background signal coherently interferes with the vibrational response, distorting and shifted spectral lines, and is non-linearly mixed with the resonant signal such that it cannot simply be subtracted or normalised away.
The ability to acquire images with intrinsic chemical contrast – particularly of lipids – is of such utility in microscopy that commercial systems have been developed, even with the limitations caused by the NRB. These microscopes probe only single vibrational resonances, and only give quantitative spatial information: due to the NRB, and the non-linear nature of CARS, signal intensities cannot be related to chemical concentration.
Removing the non-resonant background
Many experimental methods have been developed to deal with the NRB, but these either sacrifice the majority of the signal, are experimentally complex, slow to implement (as they require scanning of wavelength or phase), or are applicable only to specific, well controlled laser pulses. Computational methods have also been developed, but these require considerable art to use. Careful calibration measurements are needed, and retrieved spectra have an intrinsic, spectrally varying error. Also, measurements are limited to spectra with specific properties (sufficient spectral width, bounded by non-resonant regions).
Academics at Kings have overcome these issues by creating an experimentally simple, robust, all-optical scheme to acquire NRB-free CARS measurements. The instrument gives linear, quantitative, NRB-free signals for any laser system capable of generating CARS, including both narrowband and broadband systems. Broadband micro-Raman spectroscopy using supercontinuum laser sources can therefore give spectra equivalent to conventional linear Raman spectroscopy, but with significantly enhanced signal levels, and existing narrowband microscopes with picosecond OPO’s can give quantitative measurements of chemical concentrations along with their spatial distribution.
The method works by exploiting symmetries in the CARS polarisation response, combined with a balanced homodyne detection scheme to cancel the non-resonant background (NRB) (and the real components of the CARS signal). The resulting signal is equivalent to the spontaneous Raman response. The technique requires only passive polarisation optics, and the interfering fields are all common-path, hence there is no requirement for the optical system to be interferometrically stable. Intensity fluctuations in the laser sources do not affect the removal of the NRB and real CARS components, and there are no extra requirements on the coherence of the laser sources beyond that necessary to generate CARS. As with other multiphoton techniques, such as 2-photon fluorescence, signal arises only from the centre of the laser focus, giving intrinsic optical sectioning.