Coherent Raman Microscopy: From Instrumentation To Applications

Over the last decade, there is a surge of interest in applying coherent Raman microscopy (CRM) to cell and tissue imaging^1,2. CRM is a powerful chemical imaging technique with two popular variants: coherent anti-Stokes Raman scattering (CARS) microscopy and stimulated Raman scattering (SRS) microscopy. SRS is increasingly popular compared to CARS because it is less affected by non-Raman background interference and has a linear dependence on concentration³. CRM acquires molecular vibration information similar to Raman spectroscopy, but offers significantly improved acquisition speed⁴. This is achieved by coherent vibrational excitation through a pair of pulsed lasers with high peak intensity. Because CRM provides spatially resolved chemical mapping at high spatial and temporal resolution, it has found important applications in cancer diagnosis, lipid metabolism, cell metabolism, drug transport, and many others^5,6,7,8. However, one major bottleneck that hinders widespread adoption of the technique is its technical complexity and lack of standardized protocols. Although commercial lasers and microscopes are already available, almost all systems that are currently used in the field involve various levels of customization of lasers, optics, microscopes, and detectors, which create a significant barrier for researchers interested in entering the field or adopting this technology. The lack of access further impedes biological applications. The aim of this methods collection is to provide detailed examples and guidance on instrumentation, new imaging approaches, and in vitro/in vivo cell or tissue imaging applications. We included seven publications from experts in the field that are summarized here.

Clark et al. present a side-by-side comparison of hyperspectral SRS and CARS in terms of sensitivity, spatial resolution, and spectral resolution⁹. Both techniques are implemented with spectral focusing of femtosecond lasers, a commonly employed approach that uses the spectrally chirped broadband femtosecond lasers to achieve narrowband Raman resolution. A direct comparison allows the assessment of detection differences and optimal selection of the modality for different chemical imaging applications.

De la Cadena et al. publish a new multiplex SRS method that can acquire broadband SRS images rapidly¹⁰. The technique is based on a custom differential multichannel lock-in amplifier detection, which allows for detection of up to 32 channels simultaneously. In addition, detailed noise characterization is shown, and a balanced detection method is demonstrated to eliminate laser noise. This protocol provides practical guidance for users interested in building multiplex SRS systems.

Shi et al. demonstrate the use of highly multiplexed Raman dyes in SRS imaging¹¹. With pre-resonance enhancement, they demonstrate that SRS could provide sensitivity that is comparable to standard immunofluorescence. This method circumvents the limit of spectrally resolvable channels in conventional immunofluorescence and allows monitoring of many proteins simultaneously and studying their interactions. Detailed guidance on the sample processing and Raman dye staining is provided to facilitate adoption of this powerful method.

Miao et al. present vibrational imaging of swelled tissue and analysis (VISTA), a super-resolution imaging technique that combines SRS with expansion microscopy¹². VISTA circumvents obstacles associated with fluorescent imaging and achieves label-free super-resolution volumetric imaging in biological samples with spatial resolution down to 78 nm. They further exploit a deep learning algorithm to achieve multi-component segmentation, including protein aggregates in cells and brain tissues.

Espinoza et al. present the use of real-time two-color SRS imaging of proteins and lipids for tissue diagnostic applications¹³. Two-color SRS imaging can generate H&E-equivalent images for pathology applications. This protocol describes the details of implementing an orthogonal modulation technique and spectral focusing for simultaneous two-color SRS imaging. Conversion of two-color SRS images to pseudo-H&E images is also demonstrated with mouse brain tissue to illustrate tissue diagnosis applications.

Yuan et al. publish a flexible chamber system for long-term SRS imaging of live cells¹⁴. Most of the existing live-cell observation chambers are designed for bright field and fluorescence imaging. They are incompatible with SRS detection because of the requirement of a high NA objective and condenser. The authors demonstrate 24 h time-lapse SRS imaging with the flexible chamber, which maintained constant temperature, CO₂, and humidity.

Lastly, Wu et al. demonstrate simultaneous in vivo two-photon fluorescence (TPEF) and SRS imaging¹⁵. Two separate laser systems are combined to achieve optimal sensitivity for both imaging modalities to provide complementary information. In vivo SRS imaging of tissue has unique challenges because of the need for epi-detection of depolarized, backscattered light. This protocol discusses the details of the optimal setup of a TPEF-SRS microscope and how it can be used for in vivo imaging of the mouse’s spinal cord.

This collection of publications represents the current state-of-the-art in CRM instrumentation and technical advances. Researchers interested in both method development and applications will find this collection useful to compare setups and establish benchmarks (sensitivity, resolution, etc.) for constructing or optimizing their own systems. The guidelines and best practices provided in these protocols will inform the existing and future in vitro and in vivo applications. The large-scale adoption of the CRM technology still awaits key and innovative applications and the commercialization of complete CRM solutions. Nonetheless, standardization of imaging approaches and benchmarks is essential to ensure reproducibility and reliability and facilitate broader implementation of CRM¹⁶.

The field of CRM is growing at a rapid speed. The technical developments pushing the sensitivity, specificity, and resolution limit are continuing to drive the technology into new territories. For example, the spatial resolution of SRS is beginning to surpass the diffraction limit^17,18. The sensitivity of electronic pre-resonance SRS is now approaching that of fluorescence¹⁹. As compared to fluorescence, CRM has distinct advantages. For example, the use of deuterium tracing allowed in vivo study of metabolic dynamics of small molecules at a subcellular resolution for the first time²⁰. With small tags, it is possible to perform highly multiplexed Raman detection of 24 or more labeled targets²¹. The advances in artificial intelligence enable new capabilities for CRM by leveraging its high information content and molecular-specific features to segment and classify objects with subtle spatial and spectral feature differences²². Together, these technical advances in CRM are poised to accelerate and broaden applications in both biological and biomedical imaging.

The author has nothing to disclose.

The author acknowledges financial support from the National Institute of Health (NIH R35 GM133435) and the Eli Lilly Young Investigator Award.