CARS Is Finding Its Niche

Gary Boas, Contributing Editor, gary.boas@photonics.com

Overcoming challenges on the path to commercialization

No matter how useful a technology might prove in the research lab, its developers sometimes still have trouble finding a commercial place for it. The technology might be too expensive or too complicated for end users, or the list of applications it serves might simply be too short. So, on the path to commercialization, developers must address a range of issues beyond the question, “Does the technology work?” And, every so often, they must face the possibility that they have gone as far as they can with it.

Coherent anti-Stokes Raman spectroscopy has found itself at just such a crossroads. But instead of lamenting the death of the technique, developers have come to view this as an opportunity to find new ways of meeting the end users’ needs and addressing obstacles.

A new technique based on stimulated Raman spectroscopy helps to address the sensitivity issue with CARS. Shown are epi-SRS images of fresh normal brain slices. These include a full fresh 2-mm-thick coronal section (a) and structural features such as cortex (b), hippocampus (c), corpus callosum (d), choroid plexus (e), hypothalamic nuclei (f), habenular nucleus (g) and caudoputamen (h), all of which demonstrate the expected histologic architecture. Note that fresh-tissue images are free of the artifacts associated with fixation and freezing – techniques typically used for histologic imaging. Courtesy of the Xie Group, Harvard University.

In the case of CARS, early successes with lipid imaging readily demonstrated the commercial viability of the technique. But development for further applications has been complicated by two important factors: cost and sensitivity.

“Sensitivity is still a major issue moving forward,” said Yiwei “Kevin” Jia of Olympus America, who played a role in developing early CARS systems while at Harvard University in the early part of the last decade. With its C-H molecular vibrational mode, CARS has had great success in imaging lipid-rich samples, including in vivo axonal myelin (for studies of multiple sclerosis, stroke and injury), lipid trafficking or lipid bodies in a C. elegans (used in obesity studies, for instance). However, the technique has not yet reached its goal of providing a general nonstaining chemical imaging method and still might not be sufficient to answer some of the biological questions researchers are asking; e.g., those requiring detection of extremely small lipids or low-concentration cellular components.

As a result, the applications afforded by the technology may not be broad enough to justify a completely independent imaging system – or at least not enough to make it commercially viable.

“We could one day reach a point,” Jia continued, “where we have to ask ourselves, ‘Can we get the sensitivity to where it needs to be for this to become a general chemical imaging method, or do we just say, “That’s it”?’ Might the best choice be to focus on the CARS method as add-on modality of a multiphoton system?”

A number of researchers are still working to push the sensitivity of the technique – or even to evolve the technique into something new. “We have something better than CARS,” said Sunney Xie, the Harvard researcher who pioneered the method. Xie cites sensitivity and specificity as the primary limiting factors with CARS. “There was an intense effort in my group, trying to increase the sensitivity,” but ultimately the challenges were too great because of the nonresonant (spectral) background.

In a 2008 Science paper, Xie and colleagues reported a related technique based on stimulated Raman scattering (SRS). Here, the Raman signal is amplified when the difference in laser frequencies matches a given molecular frequency; there won’t be any signal when it does not. Thus, the spectral background that has contributed to the limited sensitivity of CARS is eliminated, enabling higher sensitivity with the new technique.

The SRS technique could offer an alternative to H&E staining for histopathology. Shown are SRS and H&E images of human glioblastoma multiforme xenografts. Thin sections of snap-frozen brain from an implanted human glioblastoma multiforme xenograft mouse model were imaged with both SRS and H&E microscopy (a). High-magnification individual fields of view demonstrating normal to minimally hypercellular cortex (<25% tumor infiltration) (b), infiltrating glioma (25% to 74% tumor infiltration) (c), and high-density glioma (>75% tumor infiltration) (d). Courtesy of the Xie Group, Harvard University.

Xie and his team have focused on applications since that first paper in Science, and in 2010 they demonstrated video-rate in vivo imaging on humans. In a recent Laboratory Investigation paper, they describe multicolored, stain-free histopathology using the technique and have demonstrated in vivo imaging in brain cancer mouse models in collaboration with neurosurgeon Dr. Daniel A. Orringer. They showed they could identify tumor margins with a contrast almost identical to that of hematoxylin and eosin (H&E), which has been the gold standard for histopathological diagnosis.

“At this point, we say it’s identical,” Xie said. “If we say it’s better than H&E, then no one will believe us. But, really, we know it’s better because you don’t have to fix the cell.”

Indeed, the researchers are already looking to commercialize the technology. After completing his PhD in Xie’s lab, Christian Freudiger decided to continue his work developing SRS technology by co-founding a company with professor Xie and biotechnology veteran Jay Trautman. Last summer, they were awarded a Phase I Small Business Innovation Research (SBIR) grant by the National Science Foundation and immediately set to work.

The SBIR grants, administered by all major funding agencies, serve to support early-stage research, which can be especially important in bridging the gap from academia to industry in today’s environment. “The funding climate in the US is conservative these days,” Freudiger said. “There’s a bit of risk aversion among the investors, and the grant support can help eliminate the technical risk.”

One of the most significant hurdles in commercializing Raman technology is the cost of the instruments; the lasers alone can cost upward of $300,000. To address this, Freudiger and colleagues have been seeking to develop a fiber laser system using components from the telecommunications industry. This will provide essentially the same functionality as the solid-state laser systems currently used, he said, but at considerably lower cost. The researchers have been working with professor Khanh Kieu of the University of Arizona to develop the fiber laser technology.

As for potential markets, the new technology could be of interest to the academic research community. “If you look at the existing Raman instrumentation market, researchers account for a significant portion of the microscopes being sold in the US,” he said. But, really, the holy grail for the burgeoning company is the medical device market, which greatly benefits from the much-improved imaging speed of SRS.

So far, the researchers have been looking mostly at the possibility of developing handheld scanners to help surgeons determine whether a cut is sufficient during tumor resection. But they envision a host of other clinical applications, such as breast cancer surgery, tissue conservation and more.

CARS holography for faster imaging

Imaging speed has been one of the major limitations of CARS microscopy techniques. In many cases, investigators perform layer-by-layer 2-D scanning to obtain a full 3-D image of a sample. Now, however, a team is seeking to address the need for scanning – and thereby improve the imaging speed – through implementation of CARS holography.

“We combine CARS with holography by recording a hologram of the complex anti-Stokes field,” said team member Perry Edwards, a researcher with Pennsylvania State University in University Park. “A single CARS hologram contains both the amplitude and phase information necessary to reconstruct the CARS image over a large depth of field, alleviating the need for scanning.”

Researchers have reported a CARS holography technique that offers improved imaging speed through digital back-propagation, a process analogous to “focus adjusting” with a traditional microscope. Shown are images of HeLa cells at different-depth positions obtained with CARS holography (upper) and wide-field CARS imaging (bottom, with manual focusing). A comparison of the images confirms the equivalence of digital and analog focusing. Reprinted with permission from Kebin Shi, Perry S. Edwards, Jing Hu, Qian Xu, Yanming Wang, Demetri Psaltis and Zhiwen Liu (2012). Holographic coherent anti-Stokes Raman scattering bioimaging. Biomed Opt Exp 3, pp. 1744-1749. ©2012 OSA.

Here, the 3-D imaging capability is made possible by digital back-propagation. “The digital back-propagation process is analogous to the ‘focus adjusting’ process of a traditional microscope,” Edwards said, “and, as a result, a 3-D scene can be captured in a single exposure and later reconstructed by ‘digital focusing.’ ”

This is different from optical sectioned scanning CARS microscopy. The latter obtains the complete 3-D volume distribution of the sample, but CARS holography digitally focuses at different depths. As a result, images will usually include an out-of-focus background. If the sample is sparse, though – with small, localized portions of the sample generating signal – researchers can incorporate compressive sensing to perform 3-D tomographic reconstruction and, thus, significantly improve the quality of CARS images.

CARS holography was described in the October 2011 issue of BioPhotonics. Since then, Edwards and colleagues – Qian Xu, Jing Hu, Yanming Wang and Zhiwen Liu of Pennsylvania State University, Kebin Shi of Peking University and Demetri Psaltis of École Polytechnique Fédérale de Lausanne in Switzerland – have made important strides, which they reported in a recent issue of Biomedical Optics Express.

“Prior to the Biomedical Optics Express paper, we had only demonstrated the recording of CARS holograms of polymer microsphere samples,” Edwards said.

“The recent paper demonstrates imaging of biological species (HeLa cells) and suggests CARS holography is a capable technique for performing noninvasive 3-D bioimaging.

“CARS holography can potentially open the door for the study of fast biological processes and provide intracellular dynamic information, as measurements can be performed ‘instantaneously’ at laser-pulse-width limited speeds.”