Advance bioimaging helps create improved clinical devices

A new, label-free, noninvasive method—coherent anti-Stokes Raman microscopy (CARS)—acquires images of biological samples and the chemical constituents of human tissues and cells faster and more efficiently than is possible with existing techniques. Potential? More effective clinical devices for disease diagnosis.

The team led by researcher Dario Polli, Associate Professor of Physics at the Politecnico di Milano (Milan, Italy), in collaboration with researchers from the Humanitas Research Hospital (Milan, Italy) and the Institute for Genetic and Biomedical Research and the Italian National Research Council Institute for Photonics and Nanotechnology , developed the CARS microscope and technique. The technology leverages Raman-based advantages, including label-free modality and chemical specificity, at higher speeds.

Beyond standard methods

Raman spectroscopy, a standard label-free and non-invasive chemical analysis technique, provides vibrational spectra of biological samples—namely cells and tissues—creating a unique signature to identify their chemical components.

The simplest approach for Raman imaging of biological samples uses quasi-monochromatic visible or near-infrared laser light to illuminate the samples. It also measures spontaneously emitted, inelastically scattered spectra that carry vibrational information. “However, this technique suffers from very low scattering cross sections, which requires long acquisition times on the order of ~1 s per pixel, preventing high-speed imaging,” says Polli.

The CARS technique overcomes this limitation and provides higher velocities by several orders of magnitude, thanks to the coherent excitation of molecules in the focal plane. Polli’s method uses the interaction between two ultrashort laser pulses – a pump and a Stokes – as well as biological samples to find information about how molecules vibrate when “tickled” by laser beams.

Fluorescence microscopy and spontaneous Raman (SR) microscopy are other standard imaging methods used to image biological samples.

Fluorescence microscopy, a rapid imaging technique that offers increased sensitivity, requires chemically specific fluorescent markers. The addition of markers can cause severe stress in the cells or tissues being tested and potentially interfere with their biological function. Such work benefits greatly from SR microscopy, a label-free technique that does not require fluorescent markers.

“SR microscopy allows the user to selectively distinguish many biomolecules in biological tissues,” says Politecnico di Milano PhD student Federico Vernuccio, citing a drawback: “It is quite slow and essentially does not provide 3D sectioning capabilities.”

Overcoming obstacles

As a third-order nonlinear optical process, CARS overcomes the limitations of fluorescence and SR techniques and enables label-free 3D sectioning without the need for a confocal aperture. It also enables imaging of samples at a higher speed than SR microscopy, thanks to the coherent excitation of molecules in the plane of the sample.

The CARS system operates at 2 MHz, which is a much lower repetition rate than standard systems. This allows for a time delay of 0.5 µs between two consecutive pulses, leaving more time in the system for thermal energy dissipation and ultimately reduced photothermal damage.

The ability to generate broadband, red-shifted Stokes pulses covering the entire vibrational region of the fingerprint is also an advantage. It uses white light supercontinuum (WLC) generation in a bulk crystal rather than a photonic crystal fiber (see Figure 1). WLC in bulk media is a more compact, robust, simple and alignment-insensitive technique that provides a much simpler technical solution, Polli explains.

“The WLC exhibits high cross-correlations between the intensities of its spectral components, low pulse-to-pulse fluctuations, and excellent long-term stability, which is comparable to that of a pump laser source itself,” he says. “Additionally, given the average power at the focus, limited by sample degradation, a low repetition rate implies higher pulse energy and higher peak intensity, which generates a stronger CARS signal due to the nonlinear nature of the optical effect.”

The performance is further improved by its tuning (see Fig. 2), which operates in the red-shifted spectral region (1035 nm for the pump and 1050–1300 nm for the Stokes beam). The researchers attribute the higher laser intensities on the sample before the onset of photodamage to reduced multiphoton absorption from cellular/tissue pigments and DNA.

“We use a data processing pipeline that combines artificial intelligence methods and numerical algorithms, extracting the maximum amount of information from captured CARS images,” says Polli. “Our microscope delivers high-quality images at state-of-the-art acquisition speed, with a pixel dwell time of <1 ms, limited spectrometer refresh rate, and without compromising sample integrity."

CARS technology further provides access to the fingerprint region – a specific part of the vibrational spectrum of a molecule. This region is difficult to detect, says nonlinear optics researcher Giulio Cerullo, professor of physics at the Politecnico di Milano, because it has weak signals, but “carries the unique signature of each molecule because different compounds produce different peak patterns in this spectral region.”

The broadband approach of this technique enables the collection of information in a single vibrational mode, covering the entire key region of the fingerprint in a single exposure time. To do this, the researchers generated a narrowband 10 cm pump beam^-1 full width at half maximum intensity. This explains the spectral resolution, according to their study –published in Optics Express.

“These features of the experimental setup allow us to reduce the pixel dwell time to less than 1 ms to collect CARS spectra, thus increasing the speed of vibrational imaging techniques and enabling high-speed CARS microscopy,” says Polli.

The technique also collects hyperspectral data. This, combined with deep learning-based and numerical algorithms, can yield chemical maps that distinguish between different chemical species in heterogeneous biological samples (see Figure 3).

CRIMSON project

This paper is part CRIMSON projecta European Commission-funded initiative that seeks to provide a next-generation biophotonic imaging device based on vibrational spectroscopy, with the potential to revolutionize the study of the cellular origins of disease, enabling new approaches to personalized therapy.

According to Polli, who also serves as the CRIMSON coordinator, the project focuses on developing label-free broadband coherent Raman scattering schemes in the fingerprint spectral range with the highest sensitivity and imaging speed.

“Combined with AI spectroscopic data analysis, we aim to provide a turnkey instrument for rapid cell/tissue classification with unprecedented biochemical sensitivity,” he says.

The development and work of his CARS microscopy research team are among the greatest achievements at the end of CRIMSON’s first year. “Our system can answer urgent and timely biomedical questions, especially in the field of cancer research,” says Polli. “For example, analyzing the interaction of cancer cells with immune cells in head and neck cancers, and characterizing and detecting early-stage chemotherapy-induced cells.”

Histopathology can also benefit from the advantages provided by the new CARS system since relatively large areas of the specimen must be visualized and characterized to provide an accurate diagnosis. “Detection of complex vibrational features across the entire fingerprint region, with high spatial resolution and coverage of relevant tissue regions, would help introduce Raman-based spectral histopathology in clinical settings,” he says.