Slepkov, Aaron D

Coherent anti-Stokes Raman scattering (CARS) microscopy is a label-free chemical imaging modality that uses CARS as a contrast mechanism to spatially resolve materials based on their molecular vibrational spectra. Due to the presence of a non resonant background that obfuscates the chemical information contained in CARS spectra, CARS images suffer from poor contrast and cannot be readily used for quantitative chemical analysis. Over the past two decades, spectral retrieval methods have been developed to obtain Raman-like spectra from CARS spectra. These methods promise to improve image contrast and enable reliable quantitative analysis. In this work I systematically evaluate a selection of the forefront spectral retrieval methods, including both analytical and machine learning approaches, to determine how well they perform at the task of non resonant background removal. The more recent machine learning methods demonstrate remarkable performance on spectra resembling the training dataset but are not as suitable as the analytical methods in general. The analytical methods based on the discrete Hilbert transform thus remain preferable due to their ease-of-use and general applicability.

Author Keywords: chemical imaging, coherent anti-stokes raman scattering, kramers-kronig analysis, machine learning, non-resonant background, spectral phase retrieval

In this work, a deep learning approach proposed by Valensise et al. [3] for extracting Raman resonant spectra from measured broadband CARS spectra was explored to see how effective it is at removing NRB from our experimentally measured "spectral-focusing"-based approach to CARS. A large dataset of realistic simulated CARS spectra was used to train a model capable of performing this spectral retrieval task. The non-resonant background shape used in creating the simulated CARS spectra was altered, to mimic our experimentally measured NRB response. Two models were trained: one using the original approach (Specnet) and one using the updated NRB "Specnet Plus", and then tested their ability to retrieve the vibrationally resonant spectrum from simulated and measured CARS spectra. An error analysis was performed to compare the model's retrieval performance on two simulated CARS spectra. The modified model's mean squared error value was five and two times lower for the first and second simulated CARS spectra, respectively. Specnet Plus was found to be more effective at extracting the resonant signals. Finally, the NRB extraction abilities of both models are tested on two experimentally measured CARS hyperspectroscopy samples (starch and chitin), with the updated NRB model (Specnet Plus) outperforming the original Specnet model. These results suggest that tailoring the approach to reflect what we observe experimentally will improve our spectral analysis workflow and increase our imaging potential.

The typical contrast mechanism employed in multimodal CARS microscopy involves the use of other nonlinear imaging modalities such as two-photon excitation fluorescence (TPEF) microscopy and second harmonic generation (SHG) microscopy to produce a molecule-specific pseudocolor image. In this work, I explore the use of unsupervised multivariate statistical analysis tools such as Principal Component Analysis (PCA) and Vertex Component Analysis (VCA) to provide better contrast using the hyperspectral CARS data alone. Using simulated CARS images, I investigate the effects of the quadratic dependence of CARS signal on concentration on the pixel clustering and classification and I find that a normalization step is necessary to improve pixel color assignment. Using an atherosclerotic rabbit aorta test image, I show that the VCA algorithm provides pseudocolor contrast that is comparable to multimodal imaging, thus showing that much of the information gleaned from a multimodal approach can be sufficiently extracted from the CARS hyperspectral stack itself.

Author Keywords: Coherent Anti-Stokes Raman Scattering Microscopy, Hyperspectral Imaging, Multimodal Imaging, Multivariate Analysis, Principal Component Analysis, Vertex Component Analysis

Spectral-focusing-based coherent anti-Stokes Raman scattering (SF-CARS) microscopy is a powerful imaging technique that involves temporally and spectrally stretching ultrashort laser pulses and controlling their frequency-time characteristics. However, a broader and more detailed understanding of the frequency-time characteristics of the laser pulses and signals involved, how they are related, and how they influence important aspects such as the spectral resolution is needed to understand the full potential of SF-CARS systems. In this work, I elucidate these relationships and discuss how they can be exploited to optimize SF-CARS microscopy setups. I present a theoretical analysis of the relationships between the spectral resolution, the degree of chirp-matching, and pulse bandwidth in SF-CARS. I find that, despite allowing better ultimate spectral resolution when chirp-matching is attained, the use of the broadest bandwidth pulses can significantly worsen the spectral resolution if the pulses are not chirp-matched. I demonstrate that the bandwidth of the detected anti-Stokes signal is minimized when the pump is twice as chirped as the Stokes, meaning that (perhaps counter-intuitively) a narrow anti-Stokes bandwidth does not imply good spectral resolution. I present approximate expressions that relate the bandwidths of the pump, Stokes, and anti-Stokes pulses to the degree of chirp-matching and outline how these could be used to estimate the amount of glass needed to attain chirp-matching.

I develop a spectral-focusing-based polarization-resolved (SFP-CARS) setup, by modifying our existing system, to explore the merits of integrating polarization-dependent detection as an add-on to existing SF-CARS setups. By using the system to study polarization-dependent features in the CARS spectrum of benzonitrile, I assess its capabilities and demonstrate its ability to accurately determine Raman depolarization ratios. Ultimately, the detected anti-Stokes signals are more elliptically polarized than desired, hindering a complete suppression of the non-resonant background. Nevertheless, I find that the SFP-CARS setup is a useful tool for studying polarization-dependent features in the CARS spectra of various samples and is worthy of further investigation. This work clarifies several technical aspects of SF-CARS microscopy and provides researchers with valuable information to consider when working with SF-CARS microscopy systems.

Author Keywords: coherent anti-Stokes Raman scattering, nonlinear microscopy, polarization, spectral focusing, spectroscopy

I have developed and improved a coherent anti-Stokes Raman scattering (CARS) microscope based on the spectral focusing (SF) technique. The CARS microscope uses an 800 nm oscillator and a photonic crystal fibre module to generate the supercontinuum Stokes. The photonic crystal fibre was originally designed to generate light beyond 945 nm which is useful for CARS microscopy in the CH/OH frequencies but essentially prevents access to the important fingerprint region at lower frequencies. With expert and nontraditional approaches to generating supercontinuum with sufficient power at wavelengths below 945 nm, I substantially extend the usefulness of the module for SF-CARS microscopy deep into the fingerprint region. Moreover, with the invention of a dynamic supercontinuum generation scheme we call "spectral surfing," I improve both the brightness of the CARS signal and extend the accessible CARS frequency range to frequencies as low as 350 cm$^{-1}$ and as high as 3500 cm$^{-1}$---all in a single scan-window. I demonstrate the capabilities of our broadband SF-CARS system through CARS and four-wave mixing hyperspectroscopy on samples such as astaxanthin, lily pollen and glass; liquid chemicals such as benzonitrile, nitrobenzene and dimethyl sulfoxide; and on pharmaceutical samples such as acetaminophen, ibuprofen, and cetirizine. Furthermore, In search of more useful Stokes supercontinuum sources, I compare the performance of two commercial photonic crystal fibre modules for use in SF-CARS applications, ultimately finding that one module provides better spectral characteristics for static supercontinuum use, while the other provides improved characteristics when spectral surfing is implemented.

Author Keywords: coherent anti-Stokes Raman scattering, nonlinear microscopy, scanning microscopy, spectroscopy, supercontinuum generation, vibrational spectroscopy