Elizabeth Vargis, Ph.D. and Anita Mahadevan-Jansen, Ph.D.
Department of Biomedical Engineering
Vanderbilt University

Introduction

Accurate, rapid and non-invasive detection and diagnosis of malignant disease in tissues is an important goal of biomedical research. Optical methods, such as diffuse reflectance, fluorescence spectroscopy, and Raman spectroscopy, have all been investigated as ways to attain this goal. Diffuse reflectance utilizes the absorption and scattering properties of tissues, particularly from cell nuclei and stroma. Changes in the scattering properties of tissues arise as the tissue becomes more dysplastic^{1, 2} due to variations in hemoglobin content³ and neovascularization⁴. Fluorescence spectroscopy is also influenced by the changes in the optical properties of tissues and has been used to diagnose dysplasia.^5,6.7,8 However, there are a number of disadvantages to these techniques, including the need for extensive sample preparation or excision, as well as low sensitivity and specificity rates.^6,9

Many research groups have instead used Raman spectroscopy to detect and diagnose disease in vivo without the need for tissue removal or the addition of exogenous agents. Raman spectroscopy, a method based on Raman scattering, is a powerful technique that can be applied to many tissue sites. Raman spectroscopy is a molecular-specific technique that probes the vibrational or rotational transitions in chemical bonds and provides detailed information about the biochemical composition of a sample.¹⁰ The sensitivity of this technique is so high that a Raman spectrum is effectively a precise fingerprint of the biochemical makeup of the tissue.

A probe-based Raman spectroscopy system has been developed to non-invasively obtain Raman spectra in vivo for our research. The overarching goals of our group are to use Raman spectroscopy to successfully detect and diagnose abnormal tissues regardless of a patient’s age, race, body mass index (BMI), or medical history. Nearly identical systems have been set up to acquire Raman data to study a variety of malignancies, such as cervical dysplasia, changes indicative of preterm labor in the cervix, skin cancer, colon cancer, and breast tumor margins. After acquisition, Raman spectra are calibrated to account for day to day variations and processed to subtract background fluorescence and smooth noise. Lastly, statistical analyses are performed to determine if Raman spectroscopy is capable of diagnosing malignant areas.

Setup and Methods

A schematic and picture of one experimental setup is shown in Figure 1. It consists of an EMVision fiber optic probe connected to a 785 nm diode laser (from Process Instruments, Inc. or Innovative Photonics Solutions), a Kaiser Optical Systems imaging spectrograph (Holospec, f/1.8i-NIR) and a back-illuminated, deep-depletion, thermo-electrically cooled Princeton Instruments CCD camera (PIXIS 256BR). These systems are all controlled with a laptop computer using software provided by Princeton Instruments (Winspec). In most experimental protocols, the fiber optic probe delivers between 80 and 100 mW of light onto the tissue with an integration time between 2-5 seconds. During the measurements, all room lights and the computer monitor are turned off. A spectral resolution of 8 wave numbers (cm^-1) is achieved.

For these studies, 7-around-1 fiber optic probes have been developed. The excitation fiber is 400 μm in diameter and each collection fiber is 300 μm in diameter. The following is a typical protocol for acquiring Raman data. First, the area of interrogation is cleaned with a dry swab and then with saline. After the lights are turned off, the Raman probe is placed on the area and a measurement is taken. When all of the measurements are complete, the probe is wiped off, placed in a 10% bleach solution for at least 10 minutes and then placed in deionized water for 5-10 minutes.

In general, the protocol for acquiring Raman spectra is determined by the goals of the study. For our cervical dysplasia study, data are acquired from two sets of patient groups: 1) a mostly normal population who come to the clinic for screening and 2) a diseased population who come to the clinic for disease diagnosis. In the first patient group, the cervix is cleaned and Raman spectra are then acquired from three areas of the cervix. The spectra are correlated with the resulting pathology report, which can be normal, atypical cells, low-grade dysplasia, or highgrade dysplasia. In the second patient group, the patient has already been screened for cervical dysplasia and has either atypical cells, low-grade or high-grade dysplasia.

Figure 1: a) Illustration of in vivo Raman spectroscopy setup, including the diode laser, probe (with details of probe tip), spectrograph and CCD camera. b) Photo of system on cart, ready for use in the clinic. Note the yellow arrow pointing to the probe.

During this visit, acetic acid is placed on the cervix, turning any abnormal areas white. The Raman spectra are then acquired from any area the medical provider decides to biopsy as well as one visually normal area. The biopsy is then performed. The pathology results from this procedure are negative, inflammation/metaplasia, cervical intraepithelial neoplasia grades I-III (CIN I-III), or carcinoma-in-situ (CIS).

The Raman data are then calibrated and processed to subtract background fluorescence and smooth noise. These methods have been described previously.¹¹ In the case of cervical dysplasia detection, sparse multinomial logistic regression (SMLR) is used to classify spectra from each patient as normal, benign, low-grade (CIN I) or high-grade (CIN II or III).¹² Briefly, SMLR is a Bayesian machine-learning framework that computes the posterior probability of a spectrum belonging to each tissue class based on a labeled training set. A composite spectrum averaging Raman measurements from each patient are used for group 1. For the second group, each individual spectrum was used, as they are acquired from distinct sites that can be matched to a specific pathology report.

Results and Discussion

Detecting cervical dysplasia using Raman spectroscopy has been an ongoing study of our lab. In our initial studies, spectra were acquired from the cervix of patients undergoing a total hysterectomy for either benign or malignant conditions. In those studies, Raman spectra were classified as normal, benign or malignant with a sensitivity of 89% and a specificity of 81%. Although these results are similar to those found when using standard clinical techniques, they did not provide enough proof that Raman spectroscopy can successfully diagnose cervical dysplasia and be easily implemented within the clinic.

More recently, we have been collecting data as described above in the Setup and Methods section, which had led to measurements from over 500 patients. In the first set of studies, spectra were acquired from the two patient groups and classified with over 88% accuracy.¹³ While these results were still not clinically applicable, the methods used to classify the data (SMLR) showed that there was a large amount of variance among the normal data.

Further studies have focused on finding sources of this variance and examining whether they affect the classification of disease spectra. For example, our previous work has shown that differences may be due to variations resulting from hormonal levels present during a normal menstrual cycle and before and after menopause.¹⁴ Also, the presence of disease permanently alters the cervix and thus the Raman spectra acquired from it (Figure 2). Incorporating these results into the SMLR algorithm prior to disease classification increased the classification accuracy to 94% and 97% in each study.

Currently, we are looking at variations that may be caused by ethnicity, parity, socioeconomic status and BMI. Our initial results have demonstrated that parity, i.e. comparing Raman measurements from women who have given birth to those who have never been pregnant, as well as variations in BMI have a significant effect on the dispersion of normal Raman spectra (in press). Studies are underway to see if accounting for these variations also leads to an increased rate of disease classification, similar to the results found with hormonal status and previous disease.

These slight variations have only been observed, accounted for, and incorporated into disease classification algorithms to increase sensitivity and specificity because of the increasingly sophisticated technology used in our system. Specifically, the fiber optic probe and the CCD camera are irreplaceable components that have dramatically increased the resolution of the system, enabling us to detect the small changes that are due to normal patient variations. Accounting for those changes allows the algorithm to focus on detecting the changes due to disease, resulting in higher disease classification rates.

Future Directions

While this note has focused on using Raman spectroscopy for detecting cervical dysplasia, Raman spectroscopy has also been utilized for detecting malignancies in other tissue sites, such as the breast, skin, colon and prostate, both by our research group and others. Also, Raman has been explored as a method for detecting endocrine glands, changes in the cervix indicative of labor, and inflammatory bowel disease. The fiber probe can be used alone, as described, or it can be inserted into clinical tools, such as an endoscope, to collect Raman data.

There are also many uses of Raman spectroscopy that move away from the conventional fiber optic probe. Spatially offset Raman spectroscopy (SORS) has been developed to interrogate deeper layers of tissue by separating the fibers within the probe. SORS has been used for bone applications and assessing tumor margins. Surface enhanced Raman spectroscopy (SERS) relies on an increase in Raman signal after a metal surface has been added, such as after gold nanoparticles are applied to a sample. This technique has been developed for in vivo use, as well as with Raman microscopy. Finally, Raman microscopy provides even higher resolution than what can be achieved with a probe-based system. Most of these systems cannot be used for in vivo diagnosis, however research into ways of combining the resolution of a microscope with the accuracy of Raman, such as confocal Raman spectroscopy, is currently underway.

Raman spectroscopy has the potential to assist in solving many problems facing the medical community. Using optimized equipment is essential to maximize the success of applying this technique for disease diagnosis. We believe that, as advances in technology continue, medical providers will move away from conventional techniques to optical methods such as Raman spectroscopy to deliver fast, accurate, non-invasive diagnoses.

About the Authors

Elizabeth Vargis, Ph.D.is a Biomedical Engineering Associate Professor at Utah State University (however contributed to this article during her Ph.D. at Vanderbilt University, Nashville, TN). She earned her B.S. in Bioengineering from the University of California, Berkeley and her M.S. and Ph.D. in Biomedical Engineering from Vanderbilt.

Anita Mahadevan-Jansen, Ph.D., is the Orrin H. Ingram Professor of Biomedical Engineering and Professor of Neurological Surgery at Vanderbilt University, Nashville, TN. She earned her B.S. and M.S. in Physics from the University of Bombay and her M.S. and Ph.D. in Biomedical Engineering from the University of Texas @ Austin.

Literature Cited

¹ Pavlova, I. et al. Microanatomical and biochemical origins of normal and precancerous cervical autofluorescence using laser-scanning fluorescence confocal microscopy. Photochem Photobiol 77, 550-5 (2003).

² Arifler, D., MacAulay, C., Follen, M. & Richards-Kortum, R. Spatially resolved reflectance spectroscopy for diagnosis of cervical precancer: Monte Carlo modeling and comparison to clinical measurements. J Biomed Opt 11, 064027 (2006).

³ Chang, V.T. et al. Quantitative physiology of the precancerous cervix in vivo through optical spectroscopy. Neoplasia 11, 325-32 (2009).

⁴ Chang, V.T., Bean, S.M., Cartwright, P.S. & Ramanujam, N. Visible light optical spectroscopy is sensitive to neovascularization in the dysplastic cervix. J Biomed Opt 15, 057006.

⁵ Martin, S.F. et al. Fluorescence spectroscopy of an in vitro model of human cervical precancer identifies neoplastic phenotype. Int J Cancer 120, 1964-70 (2007).

⁶ Rodero, A.B., Silveira, L., Jr., Rodero, D.A., Racanicchi, R. & Pacheco, M.T. Fluorescence spectroscopy for diagnostic differentiation in uteri’s cervix biopsies with cervical/vaginal atypical cytology. J Fluoresc 18, 979-85 (2008).

⁷ Macaulay, C. et al. Variation of fluorescence spectroscopy during the menstrual cycle. Opt Express 10, 493-504 (2002).

⁸ Sung, K.C. et al. Fluorescence spectroscopy for cervical precancer detection: Is there variance across the menstrual cycle? Journal of Biomedical Optics 7, 595-602 (2002).

⁹ Mourant, J.R. et al. In vivo light scattering for the detection of cancerous and precancerous lesions of the cervix. Appl Opt 48, D26-35 (2009).

¹⁰ Raman, C. & Krishnan, K. A new type of secondary radiation. Nature 121, 501-502 (1928).

¹¹ Lieber, C.A. & Mahadevan-Jansen, A. Automated method for subtraction of fluorescence from biological Raman spectra. Appl Spectrosc 57, 1363-7 (2003).

¹² Krishnapuram, B., Carin, L., Figueiredo, M.A. & Hartemink, A.J. Sparse multinomial logistic regression: fast algorithms and generalization bounds. IEEE Trans Pattern Anal Mach Intell 27, 957-68 (2005).

¹³ Kanter, E.M. et al. Application of Raman spectroscopy for cervical dysplasia diagnosis. J Biophotonics 2, 81-90 (2009).

¹⁴ Kanter, E.M., Majumder, S., Kanter, G.J., Woeste, E.M. & Mahadevan-Jansen, A. Effect of hormonal variation on Raman spectra for cervical disease detection. Am J Obstet Gynecol 200, 512 e1-5 (2009).

¹⁵ Vargis, E. et al. Effect of normal variations on disease classification of Raman spectra from cervical tissue. Analyst 139, 2981-2987 (2011).

Abstract

A novel astigmatism-free spectrograph design, the Princeton Instruments SCT 320 IsoPlane Schmidt-Czerny-Turner (SCT) spectrograph, is shown to give Raman spectra with better resolution and signal-to-noise ratios than traditional Czerny-Turner (CT) spectrographs. A single-stage SCT spectrograph has been interfaced to a new low-frequency Raman spectroscopy module that uses volume phase holographic gratings as Rayleigh line filters. The system has been used to measure the low-frequency Raman spectra of several crystalline amino acids. A spectrum of l-cystine shows that peaks as close as 10 cm^-1 to the Rayleigh line are measureable. The spectra show complex patterns of peaks that can be used to easily distinguish the acids from each other, resulting in what may best be called a “second fingerprint region.” While useful for low-frequency Raman work, the system described here is useful for examining higher-frequency modes of samples as well.

Introduction

The Czerny-Turner (CT) spectrograph has been used to measure Raman spectra of samples for decades.^1,2 Inherent in the design of CT spectrographs are optical aberrations, including astigmatism, coma, and spherical aberration.³ Astigmatism occurs when mirrors are used to focus a source off axis, giving distorted images and spectral peaks that are broadened, thus degrading spectral resolution. Coma occurs when mirrors are used to image a source off axis, and gives rise to images with a comet-like tail and spectra with asymmetrically broadened peaks. Spherical aberration is caused by using spherical mirrors to focus an image and causes a symmetrical blur in images and broadening in spectra. These optical aberrations are a result of the laws of physics and are present in CT spectrographs regardless of their manufacturer.

The net result of the astigmatism, coma, and spherical aberration inherent in the CT design are Raman peaks of poor quality. The 802 cm^-1 Raman shift peak of cyclohexane measured with a CT spectrograph seen in Figure 1 has an asymmetric lineshape, is broadened so it has degraded spectral resolution, and is short, thus lowering the signal-to-noise ratio (SNR).

In response to the problems with the CT spectrograph, Princeton Instruments has developed the Schmidt-Czerny-Turner (SCT) spectrograph, or IsoPlane® SCT 320 spectrograph. Compared to the CT design, the SCT spectrograph has zero astigmatism at all wavelengths across the entire focal plane, as well as reduced levels of coma and spherical aberration. Figure 1 shows a comparison of the 802 cm^-1 Raman shift peak of cyclohexane measured with CT and SCT spectrographs. The SCT peak has a symmetric shape, is narrower than the CT peak and thus has better spectral resolution, and is taller, resulting in better SNR. By reducing optical aberrations, the SCT spectrograph gives Raman spectra with better resolution and SNR than CT spectrographs.

Low-frequency (10–200 cm^-1) Raman bands provide information on lattice modes in solids, can be used to distinguish active pharmaceutical ingredient polymorphs,⁴ provide information about the metal ions in inorganic and organometallic compounds, and are used to determine the diameter of carbon nanotubes from the peak position of the radial breathing mode.⁵

Traditionally, triple monochromators were needed to reject the Rayleigh line allowing low frequency Raman spectra to be measured. A new low-frequency Raman analysis system, the XLF-CLM from Ondax, uses volume phase holographic gratings acting as ultra-narrow notch Rayleigh line filters. This system rejects enough elastically scattered photons to allow bands as close as 10 cm^-1 to the Rayleigh line to be observed using only a single-stage spectrograph. This gives a low frequency Raman system that is light, easy to use, and reasonably priced.

The limitations of this system are that it can only work with one excitation wavelength, in this case 785 nm. Also, because of the volume phase holographic gratings used, this system will not work in the ultraviolet. Thus triple monochromators are still useful for researchers using multiple laser excitation lines, who have tunable laser systems, or who work in the ultraviolet. Also, since a triple monochromator contains three spectrographs it can produce spectra at higher resolution than the system described herein whose resolution is limited by using only one spectrograph.

The XLF-CLM low-frequency Raman front end has been interfaced with a Schmidt-Czerny-Turner spectrograph to produce quality low frequency Raman spectra. The low-frequency Raman spectra of several amino acids have been measured with this system. Surprisingly unique and complex patterns of peaks were observed, making the 10–200 cm^-1 region of these spectra a “second fingerprint region.” It is possible that this wavenumber region could be very useful in distinguishing compounds of similar chemical structure.

Experimental

The amino acids analyzed were cystine, glutamic acid, cysteine, histidine, methionine, tryptophan, phenylalanine, and tyrosine. The amino acids were purchased as crystalline solids from Sigma-Aldrich (St. Louis, MO) and used as is. In all cases the L-isomers were analyzed. The samples were contained in glass vials with screw-on caps. The vials had a reproducible but small fluorescence spectrum that was subtracted from sample spectra prior to data workup.

The XLF-CLM low-frequency Raman analysis system from Ondax (Monrovia, CA) was used to measure low-frequency Raman spectra. This module contains a sample holder, focusing objective, 785 nm Ondax SureLock™ laser, amplified spontaneous emission filters, a neutral density filter, and Ondax SureBlock™ proprietary, solid-glass, volume holographic gratings with an optical density >8. Light was collected in a 180° back-scattering mode. Laser powers of ~50 mW were used. Light from the Ondax system was coupled to the spectrograph via a single-core 25 μm diameter optical fiber.

The spectrograph used was a Princeton Instruments (Trenton, NJ) IsoPlane SCT 320 Schmidt-Czerny-Turner spectrograph. This spectrograph has a focal length of 320 mm and was equipped with a 600 groove/mm grating blazed at 500 nm. The wavelength axis of the IntelliCal® wavelength calibration system. The accuracy of the wavelength calibration was 0.01 nm. The calibration was checked by taking the low-frequency Raman spectrum of sulfur and comparing it to the literature.⁸ The measured and literature peak positions for eight sulfur peaks between 20 and 220 cm^-1 agreed with a standard deviation of 0.57 cm^-1. Comparison of five peaks measured in the spectrum of L-cystine to literature values⁴ produced a standard deviation of 0.29 cm^-1. Stokes and anti-Stokes spectra were measured.

The camera used was a Princeton Instruments PIXIS:400BRX back-illuminated deep-depletion sensor with eXcelon®. The CCD was cooled to -70°C. A region of interest 15 rows tall in the center of the chip was defined and all rows in this region were vertically binned. Sensor rows outside of this region were found to contain negligible signal counts. Equipment was controlled and data collected using Princeton Instruments LightField® software v4.5. Data were processed and displayed using GRAMS v9.1 software from Thermo Fisher Scientific (Waltham, MA).

Results and Discussion

Frequently Raman spectra are measured with shifts from 4000 to 200 cm^-1. Peaks closer to the Rayleigh line are difficult to see because of the large size and width of this peak. Triple monochromators⁶ can reject most Rayleigh scattering photons and allow low-frequency Raman shifts to be seen. However, these systems are large and expensive. As a result, Raman shifts from 10 to 200 cm^-1 are not as routinely measured as higher-frequency Raman peaks.

The spectrum of L-cystine is an excellent test of the ability of a Raman system to get close to the Rayleigh line since it has known peaks at 5 and 10 cm^-1.^4,6 Figure 2 shows the spectrum of cystine close to the Rayleigh line measured with the Ondax low-frequency Raman front end and the Princeton Instruments IsoPlane spectrograph. Note that there are Stokes and anti-Stokes peaks at 9.7 cm^-1, confirming that this system can get as close as 10 cm^-1 away from the Rayleigh line. Figure 2 shows that the volume holographic gratings in the present system do an excellent job of rejecting the Rayleigh line, meaning now only a single-stage spectrograph is needed to obtain low-frequency Raman spectra.

Although this system provides low frequency Raman spectra, the use of coarse gratings and adjusting the center wavelength allows low-frequency and higher-frequency peaks to be observed at the same time, as seen in Figure 3, which shows the Raman spectrum of cystine from 15 to 800 cm^-1.

The vibrations sampled at Raman shifts greater than 200 cm^-1 typically involve individual functional groups or single molecules that are best characterized as internal or intramolecular modes. Below 200 cm^-1, modes involving groups of molecules are seen. Frequently these vibrations involve frustrated translations or rotations (librations) of one or more molecules. These types of vibrations are referred to as external or intermolecular modes,⁸ and in the case of the lattice modes of solids the vibrations are known as phonons and can involve movement of unit cell(s). Lattice modes for organic crystals frequently fall below 130 cm^-1. ⁸ Molecules with aromatic rings frequently have a series of features in this region, and have intense low- frequency Raman peaks in general.⁸

Note that the busiest region in the spectra in Figure 4 is between 50 and 150 cm^-1. One possible explanation for this observation is that in this region the energies of external and internal modes overlap, leading to vibrational interactions such as Fermi resonances that add peaks to the spectra and thus complicate their appearance. The region below 50 cm^-1 is not as complex because perhaps only lattice modes fall at this low energy. The 150–200 cm^-1 region is relatively quiet, perhaps because mostly internal modes fall in this regime and many amino acids may not have vibrations at this low energy.

A detailed discussion of each spectrum in Figure 4 and its comparison to known literature values is contained in an upcoming publication.⁹

Conclusions

A Schmidt-Czerny-Turner spectrograph was mated to a low-frequency Raman scattering module containing a laser, sample holder, amplified stimulated emission filters, and volume phase holographic gratings to filter out the Rayleigh line. The system was used to study the low-frequency Raman scattering spectra of several crystalline amino acids. The spectrum of L-cystine has a known peak at 10 cm^-1 that was observed by this system, indicating it is capable of seeing peaks as close as 10 cm^-1 to the Rayleigh line. Spectra of the amino acids between 10 and 200 cm^-1 show complicated patterns of intense peaks unique to each compound from what are probably lattice modes, low-frequency internal modes, and their interaction. The level of detail in this region justifies calling it the “second fingerprint region.”

References

Ultrasensitive ICCD Camera Enables the Study of Rapidly Evolving Material Deformation in Extreme Environments

Introduction

Understanding the response of materials under the rapidly evolving extreme conditions induced by shock-compression is of great relevance to many industries whose work involves high-strain-rate phenomena, such as aerospace design, advanced materials processing and mining, renewable energy research, and defense technology.

 

“The study of dynamic material deformation via synchrotron x-ray radiography promises to reveal new details on fundamental damage
processes in condensed matter…”

In extreme environments like these, where loading conditions typically last for only a few microseconds, the bulk mechanical behavior of materials is governed by the interplay of numerous mesoscopic damage processes. However, the evolution of these fundamental deformation processes (e.g., local phase changes, strain localization, and crack pattern growth) is inaccessible via commonly employed diagnostics based on visible radiation, such as photonic Doppler velocimetry and high-speed imaging. Therefore, to better understand the performance of current materials and assist in the intelligent design of materials with predefined properties, new techniques capable of providing an unobstructed view of in-material damage are required.

In-house x-ray radiography has long been employed in extreme environments to image subsurface deformation in materials. In recent years, third-generation synchrotron light sources and free electron lasers have extended the capabilities of dynamic x-ray imaging to the submicron and sub-nanosecond scales in macroscopic samples. These increased capabilities not only allow finer insight into material deformation but also place higher demands on detector technology.

In this application note, the use of high-energy synchrotron x-ray radiography to study shock-induced deformation in high-Z materials is discussed. Key to this research is the Princeton Instruments PI-MAX4:1024i intensified CCD (ICCD) camera with a Gen III filmless intensifier, which permits fast imaging in low-photon scenarios.

X-ray Imaging of Shock-Compression Experiments

High-energy, high-resolution x-ray imaging experiments were performed at Beamline I12 at the Diamond Light Source synchrotron. Well-defined, reproducible shock waves were driven into targets by projectiles (cylindrical steel or copper flyer plates: 2 mm thick, 12.5 mm diameter) launched from a purpose-designed portable gas gun¹. Impact velocities ranged from 250–850 ms^-1, generating shock pressures and material motion on the order of 5–20 GPa and many hundreds ms^-1, respectively. Following impact, target damage was examined via high-energy (50–250 keV) x-ray radiography and complementary velocimetry diagnostics. Figure 1 illustrates the experimental setup, showing how radiographs are captured using a single-crystal scintillator, optical relay, and the PI-MAX4:1024i ICCD camera.

Figures 2 and 3 show a photograph of the experimental apparatus assembled at Beamline I12 and a photograph of the PI-MAX4 ICCD camera’s optical relay, respectively. The large size (11 x 7 x 4 m) of the second experimental hutch at Beamline I12 facilitates large-scale experiments, allowing several diagnostics to be fielded simultaneously. To protect against damage from Compton-scattered x-rays, the optical relay and camera are shielded with 2 mm of lead.

Periodic 3D-printed Structures for Energy Absorption

Periodic structures with well-defined, customizable porosity are of great interest to safety applications where impact or blast stresses may be dissipated by the process of successive pore collapse. To examine the intricacies of the blast mitigation process in more detail, well-defined stainless steel lattice structures were produced by selective laser melting (SLM) and subject to dynamic radiography testing. Figure 4(a) shows a photograph of a typical SLM lattice structure and Figure 4(b) presents an illustration of the impact experiment.

Although visible radiation methods such as high-speed imaging in-silhouette may reveal the kinetics of the pore collapse process in open structures, penetrating x-ray radiation is required to resolve density changes throughout the system. These measurements may then be fed back into material models to guide the design of better-performing systems. Figure 5 shows a series of in-situ radiographs captured before and during the impact experiments. One radiograph was captured per shot with a 12.5 x 12.5 mm field of view and 500 ns exposure. In each dynamic image, the cylindrical projectile enters from the left. Repeat experiments were performed to stroboscopically step through the deformation process.

The dynamic radiographs reveal the localized propagation of damage throughout the lattice structures. Future work will continue to compare deformation observed in the radiographs to the predictions of 3D hydrocode simulations, providing novel data to evaluate material strength and design models.

Important New ICCD Technologies

The highly advanced PI-MAX4:1024i scientific camera (see Figure 6) used in the experiments discussed above features an exclusive picosecond gating technology from Princeton Instruments. By employing state-of-the-art electronics and fiberoptically bonding the intensifier to the CCD sensor, this technology enables new PI-MAX4:1024i cameras to gate conventional image intensifiers (which normally achieve ~2 to 3 ns gating) at <500 ps without sacrificing quantum efficiency. An integrated programmable timing generator, SuperSynchro, built into the PI-MAX4:1024i further enhances the camera’s utility for high-precision, time-resolved applications.

It should be noted that another new addition to the PI-MAX®4 series, the PI-MAX4:2048f, now provides four times the imaging area and resolution of any other scientific ICCD camera currently available. This large-format camera, which employs a 2k x 2k CCD fiberoptically coupled to one of several 25 mm diameter Gen II or Gen III filmless intensifiers, offers SuperSynchro, high frame rates (6 MHz / 16-bit digitization), and a 1 MHz sustained gating repetition rate.

Complete control over all PI-MAX4:1024i and PI-MAX4:2048f hardware features is simple with the latest version of Princeton Instruments’ LightField® data acquisition software (available as an option). Precision intensifier gating control and gate delays, as well as a host of novel functions for easy capture and export of imaging data, are provided via the exceptionally intuitive LightField user interface.

Summary

The study of dynamic material deformation via synchrotron x-ray radiography promises to reveal new details on fundamental damage processes in condensed matter, which may be used to evaluate the results of leading numerical models. This interplay between novel experimental data and advanced modeling can assist in the design of better-performing materials for the advanced materials processing, aerospace, defense, and renewable energy industries.

The PI-MAX4:1024i ICCD camera from Princeton Instruments allows researchers to probe material behavior on ultrafast timescales with resolution sufficient to evaluate leading predictive models. Using Princeton Instruments’ SuperSynchro technology in combination with a Gen III filmless intensifier offers unparalleled flexibility in gating and sensitivity, thereby facilitating imaging in low signal-to-noise scenarios.

Resources

To learn about research being conducted by the Institute of Shock Physics, Imperial College London, please visit: https://www.imperial.ac.uk/shock-physics

Reference

D.E. Eakins and D.J. Chapman, Review of Scientific Instruments 85, in press, 2014.

Authors

Mr. Michael E. Rutherforda, Dr. David J.
Chapmana, Dr. Daniel E. Eakinsa
Dr. Michael Drakopoulosb
Mr. Manjul Shahc

a: Institute of Shock Physics, Imperial
College London, London, UK, SW7
2BW

b: Diamond Light Source, Beamline I12
JEEP, Didcot, Oxfordshire, UK, OX11
0DE

c: Princeton Instruments, 3660
Quakerbridge Road, Trenton, NJ 08619
USA

Acknowledgements

The authors would like to acknowledge the continued support of this project by Imperial College London, EPSRC, STFC, and AWE. Additionally, the authors would like to extend their gratitude to all who assisted in performing experiments at the Diamond Light Source: Steven Johnson and David Pittman (sample preparation); Mark Collison, David Jones, Jasmina Music, Sam Stafford, Gareth Tear, Thomas White, and John Winters (experimental assistance).

Introduction

Over the past several years, biomedical researchers and engineers working in labs, hospitals, and universities around the world have developed an extensive set of spectroscopy-based methods — including a new class of non-invasive in vivo techniques utilizing near-infrared spectroscopy (NIRS) — to facilitate the rapid and accurate detection and diagnosis of disease and injury.¹

Recently, a British research group affiliated with both the Department of Medical Physics and Bioengineering, University College London, as well as the Institute for Women’s Health, University College London and Neonatal Unit, University College London Hospitals Trust, designed and tested a NIRS system to aid in the diagnosis of neonatal brain injury.²

The novel bedside system built by the researchers in London simultaneously measures cerebral changes in tissue oxygenation and hemodynamics by estimating the changes in hemoglobin concentration. The portable system also tracks oxygen utilization by measuring the oxidation state of cytochrome-c-oxidase (CCO), which is responsible for >95% of oxygen metabolism in the body.²

Use of the London team’s lens-based broadband NIRS system, known as CYRIL (short for “CYtochrome Research Instrument and AppLication”), was demonstrated in a cohort of six newborn infants with neonatal encephalopathy in the Neonatal Intensive Care Unit (NICU) for continuous measurement periods of up to five days; the NIRS measurements commenced immediately after being granted parental consent.^2*

CYRIL Setup

The CYRIL system (see Figure 1) utilizes two discrete channels, each employing a single light source with four detectors. An optical fiber illuminator with a thermally stable broadband white light source is used for each channel and an aspheric lens collimates the lamp’s high-intensity NIR output, focusing it onto the fiber input. A shutter and iris control the amount of light entering the fiber. Use of a long-pass (610 nm) and a short-pass (950 nm) filter in the collimated region narrows the spectrum to the detected wavelengths.²

*Ethical approval for the Baby Brain Study at University College London Hospitals Trust (UCLH) was obtained from the North West Research Ethics Centre (REC reference: 13/LO/0106). Term infants born at or transferred to UCLH for treatment of acute brain injury were eligible for investigation; only babies without congenital malformations and considered likely to survive were considered.²

Optical fibers and optode holders connect the light source to the tissue, as well as the tissue to the spectrograph. Each 3 meter fiber bundle consists of multiple high–numerical-aperture fibers (NA = 0.57) with 30 μm diameters. At the tissue, the source fiber bundle branches into a pair of fiber heads (bundle diameter = 2.8 mm). The detectors, meanwhile, comprise eight individual fiber bundles (diameter = 1 mm). The detector fibers can be arranged in any combination, thus enabling multi-distance measurements, spatially resolved spectroscopy (SRS), and/or image acquisition. The fibers are arranged vertically into a ferrule that inputs the light to the spectrometer, which allows the spectrum from each fiber to be detected on a two-dimensional CCD individually as well as simultaneously.²

As the light throughput of lens-based spectrographs is superior (>99%) to that of mirror-based spectrographs, the CYRIL system utilizes an Acton Series LS-785 lens-based spectrograph from Teledyne Princeton Instruments (see Figure 2). After light is collected from the tissue surface by CYRIL’s detector fibers, it is input into a fiber-adapted entrance in the LS-785 and passes through a variable slit to prevent overexposure. The optimal slit opening in this study was determined to be 20 μm. To reduce loss, the light is collimated and then guided to a diffraction grating (blazed at 1000 nm; 830 grooves per mm) that yields a wavelength resolution of 0.7 nm and a bandwidth of 136 nm. Because the diffraction grating is mounted on a rotating platform within the LS-785, it is easy for CYRIL users to select the range of wavelengths to be resolved (770 – 906 nm in this study). Lastly, the light is focused onto the CCD via an f/2 focusing lens. Focusing the light in the y-direction reduces cross-talk between detector channels.²

The CCD, a scientific-grade sensor utilized by the CYRIL system’s Teledyne Princeton Instruments PIXIS:512F camera, features a photosensitive array whose peak quantum efficiency is in the NIR region. This two-dimensional array of 512 x 512 pixels measures 12.3 x 12.3 mm, with each pixel measuring 24 x 24 μm. Light from all eight of CYRIL’s detector fibers can be detected simultaneously on the sensor, whose front-illuminated architecture is ideal for the low to moderate photonic signal levels associated with near-infrared spectroscopy. To reduce dark noise, the CCD is thermoelectrically cooled to –70°C (thermally stable to ± 0.05°C) during camera operation; the dark current at this temperature is 0.002 electrons/pixel/sec. When the camera is run at 1000 kHz, the read noise is 5 electrons RMS. The researchers used Teledyne Princeton Instruments LightField® acquisition software (in concert with the IntelliCal® calibration package) to calibrate each pixel in the x-direction to its corresponding wavelength bin. After calibration, the CCD resolution is 0.27 nm. Therefore, CYRIL’s wavelength resolution is 0.27 ± 0.70 nm.²

Data Acquisition and Analysis

Examples of the intensity spectra recorded (Figure 3a) and the corresponding change in attenuation between them (Figure 3b) are presented below. The change in the tissue’s levels of hemoglobin chromophore concentration is reflected by the intensity spectrum’s change in shape. During systemic oxygen desaturation (SpO2), the peak of the spectra shifts from ~780 nm to ~785 nm due to the decrease in Δ[oxCCO] (i.e., the oxidation state of the metabolic enzyme cytochrome-c-oxidase) and Δ[chromophore HbO2], as well as the increase in Δ[chromophore HHb].²

National Instruments’ LabVIEW^® 2011 software was utilized to create the custom program that allows CYRIL users to control the PIXIS:512F, collect raw data, and calculate the corresponding concentrations. The special LabVIEW program offers real-time views of the detected intensity spectra and concentration changes of the chromophores for each channel, right at the bedside. Its wavelength binning functionality lets CYRIL system users acquire and display an intensity-weighted image of the CCD’s 512 x 512 array and select the region of interest (ROI) for each individual horizontal strip to bin. Each strip corresponds to one of the detector fibers input into the LS-785 spectrometer. All of the strips can be adjusted to maximize intensity spectra per channel without saturation and to minimize cross-talk between channels. The London researchers report no significant cross-talk in the study’s setup.²

LabVIEW drivers from R Cubed Software’s scientific imaging tool kit enable the PIXIS:512F to acquire data separately from the ROIs and to bin the data by wavelength per user-defined settings. Recorded light intensities peaked at >60000 counts/sec in the study. Typical photon counts were >50000 for the detectors placed 1 cm from the source, >35000 for detectors 1.5 cm from the source, >30000 for detectors 2.0 cm from the source, and >24000 for detectors 2.5 cm from the source. The dark count was 2 orders of magnitude lower (~400 counts). Mathworks’ MATLAB^® software was used to perform data analysis and the NIRS data were processed with an automatic wavelet de-noising function capable of reducing high-frequency noise whilst maintaining trend information.²

Summary of Results

The researchers utilized CYRIL both to measure changes in cerebral tissue oxygenation simultaneously with hemodynamics by estimating changes in hemoglobin concentrations as well as to track metabolism and oxygen utilization by measuring the oxidation state of CCO. Quantitative near-infrared spectroscopy data were acquired at the same time as systemic data, permitting multimodal data analysis, and the researchers studied NIRS variables in response to global pathophysiological events (NIRS analysis was focused on spontaneous oxygen desaturations). The study indicates that the relationship between hemoglobin oxygenation changes and CCO oxidation changes during the desaturation events was significantly correlated with a magnetic-resonance-spectroscopy–measured biomarker of injury severity.²

Enabling Technologies

CYRIL’s Acton Series LS-785 from Teledyne Princeton Instruments is a lens-based spectrograph (see Figure 4) that delivers the highest throughput of any commercially available NIR Raman spectrometer on the market today. Performance advantages of the LS-785 include easy wavelength adjustment for 750 – 830 nm laser excitations, custom-designed anti-reflective coatings, unique f/2 lenses (whose proprietary coatings provide >99% throughput), simple interfacing to fiberoptic probes and microscopes, and 5 cm^.-1 resolution.

CYRIL also employs a Teledyne Princeton Instruments PIXIS:512F scientific CCD camera. This low-noise camera, which has been engineered for optimal performance in quantitative UV-to-NIR imaging and spectroscopy applications, uses Teledyne Princeton Instruments’ exclusive XP cooling technology to minimize thermally generated dark current (thereby preserving high signal-to-noise ratios even during many hours of continuous operation). Thanks to this innovative and field-proven technology, PIXIS is the only scientific camera platform to offer deep cooling with an all-metal, hermetically sealed design that carries a lifetime vacuum guarantee. For convenient control and image/spectra acquisition using a laptop, the PIXIS:512F is equipped with a USB 2.0 data interface.

Both the LS-785 lens-based spectrograph and the PIXIS:512F scientific CCD camera are fully compatible with Teledyne Princeton Instruments’ 64-bit LightField imaging and spectroscopy software (see Figure 5). LightField serves as a full-feature “command center” — affording users complete control of spectrometer and camera operating parameters, spectral acquisition, data processing, and more. Among its many benefits, such as a powerful new math engine, LightField allows direct data acquisition into National Instruments’ LabVIEW and MathWorks’MATLAB software packages.

Furthermore, LightField supports Teledyne Princeton Instruments’ exclusive IntelliCal wavelength and intensity calibration package (see Figure 6). Yet another capability leveraged to good effect by the creators of CYRIL, IntelliCal provides up to 10x improvement in wavelength accuracy and instrument-independent intensity calibration.

Future Directions

By providing unique and real-time in vivo patient data, new bedside spectroscopy-based measurement systems such as CYRIL hold the potential not only to aid in disease and injury diagnosis but to help guide treatment.

In the coming years, commercially available high-precision spectrometers and scientific cameras like those from Teledyne Princeton Instruments will continue to offer biomedical researchers and engineers a broad range of easy-to-integrate features and functions to meet the evolving sensitivity, speed, and resolution requirements of these novel systems.

Resources

To learn more about the latest NIRS-enabled research being performed by the Multimodal Spectroscopy group at University College London, please visit: www.ucl.ac.uk/medphys/research/borl/nirs/mms

References

Acknowledgment

Teledyne Princeton Instruments would like to thank Dr. Ilias Tachtsidis, University College London, for his invaluable contributions to this application note.

“The PI-MAX4:1024EMB emICCD camera seamlessly combines the rapid gating capabilities of an image intensifier and the excellent linearity of a back-illuminated, frame-transfer, 1024 x 1024 EMCCD detector…”

Introduction

Laser-induced breakdown spectroscopy (LIBS) is considered one of the most convenient and efficient analytical techniques for trace elemental analysis in gases, solids, and liquids. Both non-invasive and non-destructive, LIBS requires little-to-no sample preparation and can easily be performed either in the lab or in the field in hazardous industrial environments in real-time. Remote measurements can be done from more than 50 meters’ distance.

During LIBS, typically, but not always, a short laser pulse generated at 1064 nm by an Nd:YAG laser is focused on a sample (see Figure 1). Laser energy heats, vaporizes, atomizes, and ionizes target material, generating a small area of plasma. Excited atoms and ions in the plasma emit secondary light that is collected and resolved by a spectrometer and directed to a high-speed, high-sensitivity photodetector. Each chemical element has a unique spectral signature that can be discerned from the obtained spectra. As a result, the multi-elemental composition of the sample can be determined instantly.

To satisfy LIBS’ competing requirements of high resolution and broad wavelength range, an echelle spectrometer is recommended. Unlike Czerny-Turner and Littrow spectrometers, echelle spectrometers use two dispersive elements, usually a prism and a grating. Thus, the spectrum is divided not only in the x-direction but also the y-direction. The primary advantage of this assembly is that the entire two-dimensional detection array is utilized and a full spectrum is captured at high resolution, not just a small region.

In LIBS applications where a relatively low power laser can be used, the less intense background continuum emission means a conventional CCD camera with a high-speed mechanical chopper affords a suitable imaging solution. However, in LIBS applications where a high-power, short-duration laser pulse is required (i.e., the lone laser pulse lasts femtoseconds to nanoseconds, 10-15 to 10-9 sec), only a comparatively small amount of laser energy is transferred to the sample. This short-duration pulse produces a weak emission signal over the continuum that is very difficult, or even impossible, for a conventional CCD camera to capture with good signal-to-noise (S/N) ratio — which is why gated, intensified CCD (ICCD) cameras are widely utilized to perform LIBS.

For challenging applications where ICCD camera sensitivity and dynamic range are not enough, an emICCD camera is recommended. This document presents data demonstrating the improved performance of a new emICCD camera from Princeton Instruments (used in concert with an echelle spectrometer from Lasertechnik Berlin, or LTB) that delivers ultrahigh sensitivity for demanding LIBS applications on the nanosecond and picosecond timescales.

Princeton Instruments’ new emICCD camera technology overcomes ICCD sensitivity and dynamic range limitations by fiberoptically coupling an image intensifier to a back-illuminated, electron-multiplying CCD (EMCCD) detector. Dual-gain control optimizes the emICCD camera’s performance by adjusting both the intensifier and the EMCCD gains, thus permitting better linearity and dynamic range (subsequently increasing S/N ratio).

Experimental Setup

LIBS setups for laboratories and in-the-field industrial settings rely on compact, relatively inexpensive scientific-grade instruments. Figure 2 is a diagram of a typical LIBS setup, which includes a pulsed laser, a mirror, a focusing lens, a sample holder/chamber, collecting optics, an optical fiber, an echelle spectrometer, a gated ICCD camera, and a computer (not shown).

The LIBS spectra presented in the following section were acquired at Lasertechnik Berlin’s test lab utilizing a setup similar to the one depicted in Figure 2. The experimental setup in Berlin included an LTB ARYELLE 200 echelle spectrometer, which was mated to a PI-MAX4:1024EMB emICCD camera from Princeton Instruments configured with a super-red image intensifier fiberoptically bonded to a back-illuminated, frame-transfer EMCCD detector. See Figure 3.

Spectra and Observations

Figure 4 presents a LIBS spectrum of a solid sample (i.e., 402 stainless steel) acquired using a PI-MAX4:1024EMB emICCD camera mated to an ARYELLE 200 echelle spectrometer. By utilizing the echelle spectrometer to divide the spectrum in the x-direction as well as the y-direction, the charge-coupled device’s entire 1024 x 1024 detection array can be leveraged to capture the full spectrum, matching the spectral resolution of the ARYELLE 200.

Figure 5 compares LIBS spectra of 402 stainless steel acquired utilizing the LTB echelle spectrometer paired with an emICCD camera (spectrum rendered using a gold line) and a standard ICCD camera (spectrum rendered using a blue line), respectively.

The emICCD camera (PI-MAX4:1024EMB) was fitted with a super-red intensifier, whereas the standard ICCD camera (PI-MAX4:1024f) was fitted with a red-blue intensifier. The emICCD camera, with its dual-gain mechanism, was observed to be ~3x to ~5x more sensitive. Had the same super-red intensifier been used for both cameras, it is expected that the sensitivity with the emICCD camera across the full spectral range, from 200 nm to 900 nm, would have been ~3x better than with the standard ICCD camera. The solid sample, the megapixel detector array size, the intensifier gate/delay times, the laser excitation wavelength, and other key experimental parameters were identical.

Figures 6 and 7 provide a closer look at these spectra in the UV and NIR regions, respectively. Thanks to its use of a CCD featuring electron-multiplying detection technology, the emICCD camera system clearly demonstrates superior sensitivity and resolution in the spectral regions of interest when compared to the standard gated ICCD camera.

Enabling Technology

The PI-MAX4:1024EMB emICCD camera seamlessly combines the rapid gating capabilities of an image intensifier and the excellent linearity of a back-illuminated, frame-transfer, 1024 x 1024 EMCCD detector in order to deliver quantitative, ultra-high-sensitivity performance for LIBS applications executed on nano- and picosecond timescales.

This fiberoptically bonded Princeton Instruments PI-MAX^®4 camera system provides <500 psec gate widths using standard fast-gate intensifiers while preserving quantum efficiency. Its integrated SuperSynchro timing generator allows camera users to set gate pulse widths and delays under GUI software control, and significantly reduces the inherent insertion delay (~27 nsec).

Complete control over all PI-MAX4:1024EMB hardware features is simple with the latest version of Princeton Instruments’ LightField^® data acquisition software (available as an option). Precision intensifier gating control and gate delays, as well as a host of novel functions for easy capture and export of spectral data, are provided via the exceptionally intuitive LightField user interface.

The PI-MAX4:1024EMB uses a high-bandwidth (125 MB/sec or 1000 Mbps) GigE data interface to afford camera users real-time image transmission. This interface supports remote operation from more than 50 meters away.

Summary

Laser-induced breakdown spectroscopy generally requires little if any sample preparation and is easy to perform in hazardous industrial environments in real-time. Hence, it is a very attractive analytical tool. LIBS applications are utilized in many diverse fields. Areas that benefit from LIBS include geology, mining, construction, exploration of planets, environmental monitoring, analysis of biological samples, archaeology, architecture, military/defense, forensic science, combustion processes, the metal industry, and the nuclear industry.

Unique PI-MAX4 emICCD cameras from Princeton Instruments provide high-performance, cost-efficient, and user-friendly solutions for performing ultra-high-sensitivity LIBS applications on nanosecond and picosecond timescales. All instrumentation-related functions are fully controlled via software that includes quantitative algorithms as well as script language for complex automated measurement tasks.

Overview

Although the use of sprays for industrial processes such as material deposition, cutting, and mixing is widespread, the design and testing of most spray devices is still predominantly phenomenological, owing not only to limitations in computing power but to gaps in the fundamental understanding of the multiphase fluid phenomena that drive spray breakup and morphology ^1,2. One key to developing a first-principles understanding of sprays, therefore, is to devise methods for making measurements in turbid spray regions, particularly measurements that can offer insight into fluid motion and the forces active in spray mixing.¹

Recently, researchers at Chalmers University of Technology in Sweden demonstrated a three-pulse configuration for time-gated ballistic imaging (BI) applied to a turbulent, steady spray; this technique, which utilizes a pair of ultrafast scientific ICCD cameras, permits the acquisition of time-correlated image data.¹

The researchers report that the new approach images the liquid-gas interface of a spray, where shear forces and turbulence instabilities act to break the liquid apart. Region-matching analysis is subsequently applied to the time-correlated image data so as to compute the velocity and acceleration of resolved fluid structures.¹

This application note will outline some of the novel work that has been performed by the group, which is led by Dr. Mark Linne, Professor of Applied Mechanics in the Combustion Division at Chalmers.

Experimental Setup

The Chalmers University experimental setup (shown in Figure 1) uses intense, ultrafast light pulses and specialized detection technologies to separate high-quality imaging light from diffusely scattered background light. The region of interest is illuminated with a series of coherent ultrafast pulses generated from a single oscillator that seeds three matched regenerative amplifiers ¹.

Final collection of the high-quality imaging light from each source pulse is effected by dividing the optical signal exiting the Kerr gate into two detection paths. The images are acquired by two Princeton Instruments PI-MAX^®4 scientific ICCD cameras equipped with intensifiers fiberoptically bonded to interline-transfer CCDs. The gain response of the intensifiers can
be gated to 5 ns and, using the interline-transfer sensors’ masked storage areas in concert with specialized PI-MAX4 readout modes, each of the two cameras can generate a pair of images separated by as little as 450 ns. Thus, each individual PI-MAX4 can be set to capture consecutive image pulses (i.e., two images per camera). With this arrangement, each pulse train yields three time-correlated images of the spray.¹

Data & Results

The Chalmers University steady-spray rig was employed to set up a controlled test case for measuring fluid velocity and acceleration in a turbid environment. Image-triplets of a water spray (P_inj = 19 bar) exiting a plain-orifice nozzle (L/d = 25) were acquired for dynamic ballistic imaging analysis. Figure 2 shows one of these image-triplets.¹

The primary objective of the Chalmers University three-pulse BI system is to enable the acquisition of spatially resolved time-series data that can be analyzed to track motion and changes in the imaged structure. Consecutive images in the series are compared and velocity vectors are computed for each time-step by selectively matching structure from targeted image regions. By computing coordinated sets of velocity vectors, the time-resolved displacement information acquired from multiple time-steps can be utilized to estimate higher order motion components.¹

The velocity results for the spray imaged with the three-pulse BI system indicated large streamwise motion with small variations in the radial direction. The acceleration results indicated that the liquid structures were subjected to acceleration in the positive radial direction, and retardation in the streamwise direction. These results are indicative of both turbulence and shear forces affecting the breakup of the liquid jet.¹

For a more detailed description of the experimental setup as well as an in-depth discussion of the results and their derivation, please refer to D. Sedarsky, M. Rahm, and M. Linne, “Visualization of acceleration in multiphase fluid interactions,” Opt. Lett. 41, 1404–1407 (2016). Related investigations being conducted at Chalmers University of Technology include the use of a short-pulse, time-gated, backscattering setup designed to provide information on the spray-formation region of atomizing sprays.³

New ICCD Camera Technologies

Thanks to innovative picosecond gating technology developed by Princeton Instruments, the PI-MAX4 scientific ICCD camera platform (see Figure 3) is now even faster. This exclusive gating technology, which fully leverages the performance advantages of Princeton Instruments’ state-of-the-art electronics and intensifier-to-CCD fiberoptic bonding, allows PI-MAX4:1024i ICCD cameras to gate conventional image intensifiers (which normally achieve ~2 to 3 ns gating) at <500 ps without compromising quantum efficiency. Princeton Instruments’ built-in SuperSynchro programmable timing generator further enhances PI-MAX4 camera utility for high-precision, time-resolved applications.

Another addition to the PI-MAX4 series, the PI-MAX4:2048f, provides four times the imaging area and resolution of any other scientific ICCD camera currently available on the market. This large-format camera, which uses a 2k x 2k CCD fiberoptically coupled to one of several 25 mm diameter Gen II or Gen III filmless intensifiers, offers built-in SuperSynchro, high frame rates (6 MHz / 16-bit digitization), and a 1 MHz sustained gating repetition rate.

Complete control over all PI-MAX4:1024i and PI-MAX4:2048f hardware features is simple with the latest version of Princeton Instruments’ 64-bit LightField® data acquisition software, available as an option. Precision control of intensifier gate widths and delays, as well as a host of novel functions for easy capture and export of imaging data, are provided via the exceptionally intuitive LightField user interface.

References

Advanced Instrumentation Enables Improvements in High-Speed, Label-Free Imaging

Overview

Popular molecular imaging techniques are only able to reveal the distribution or behavior of specific molecules within the human body that have been labeled with pigments or fluorescent proteins. Raman spectroscopy, however, allows researchers to identify the components of unlabeled molecules via spectral analysis. A vibrational (Raman) spectrum is therefore commonly referred to as a molecular fingerprint. The use of such label-free molecular imaging increases the likelihood that unexpected changes and abnormalities in the body will be discovered.¹

One innovative Raman-based technique, known as ultra-multiplex CARS spectroscopic imaging, demonstrates great utility for miniaturized, non-invasive, real-time, and cellular-level molecular diagnostics in experimental and clinical applications. This technique, which enables biological structures and processes to be observed and analyzed via quantitative photonic data collected simultaneously across a broad spectral range, continues to be refined and improved.

For example, Associate Professor Hideaki Kano of the University of Tsukuba, Japan, and his colleagues recently performed ultra-multiplex CARS spectroscopic imaging of living cells (18 colors corresponding to 18 wavenumbers) with an effective exposure time per pixel of 1.8 msec, the fastest speed reported to date for broadband CARS whose spectral coverage is ~3000 cm^-1.²

CARS Basics

Coherent anti-Stokes Raman scattering (CARS) derives its name from the fact that rather than using a traditional single laser this nonlinear spectroscopy technique utilizes a pair of very strong coherent laser beams to irradiate a sample, thereby producing an anti-Stokes frequency signal. Whereas the first laser’s frequency is usually constant, the frequency of the second can be tuned so that the frequency difference between the two lasers equals the frequency of a Raman-active mode of interest. This particular mode will be the only extremely strong peak in the Raman signal.³

CARS is orders of magnitude stronger than normal Raman emission. A monochromator is not necessarily required to perform CARS; a wideband interference filter with a detector placed behind the filter may work instead. Mathematical and schematic descriptions of CARS are presented next. See Figure 1a.

Two laser beams with frequencies ω₁ and ω₂ (ω₁ > ω₂ ) interact coherently to produce a strong scattered light with frequency 2ω₁ – ω₂ . If the frequency difference between the two lasers (ω₁ – ω₂ ) equals the frequency ω_m of a Raman-active vibrational, rotational, or other mode, then a strong light with frequency ω₁.

In other words, to obtain a strong Raman signal, the second laser frequency should be tuned in such a way that ω₂ = ω₁ – ω_m. Then the frequency of strong scattered light will be 2ω₁ – ω₂ = 2ω₁ – (ω₁ – ω_m) = ω₁ + ω_m, which is higher than the excitation frequency ω₁.

In addition to overcoming the low signal intensities of most biomolecules, CARS provides directional emission and narrow spectral bandwidth, without interference from autofluorescence. CARS studies include microbial cells, eukaryotic cells, trapped cells, medical tissue, epithelial tissue, muscle tissue, nervous tissue, lung tissue, breast tissue, bone, and skin.⁴

The rapid scanning capability of CARS microscopy is invaluable to researchers investigating real-time dynamics. Furthermore, as opposed to fluorescence microscopy techniques, cells can be repeatedly imaged in CARS without the fading problems attributable to photobleaching.⁵

Multiplex CARS and Ultra-Multiplex CARS

Multiplex CARS (see Figure 1b) is an enhanced CARS technique that employs a multichannel detector (e.g., a scientific CCD or CMOS camera) coupled to a spectrometer in order to cover the broad spectral range of the coherent Raman signal.

By using a broadband laser source (e.g., a supercontinuum light source or a femtosecond laser source) instead of the tunable laser ω₂, the typical spectral coverage of the multiplex CARS signal can be further extended, reaching approximately 3000 cm^-1. This ultra-multiplex CARS spectroscopic imaging approach is broad enough to detect all vibrational fundamental modes — including the critically important fingerprint region (in which unknown organic compounds can be identified as well as distinguished from one another) along with the C-H and O-H stretching regions (useful for identifying the rough molecular distribution of lipids, proteins, and nucleic acids).²

New Ultra-Multiplex CARS Experiment

When Dr. Hideaki Kano attended the CLEO Conference in San Francisco as a doctoral student back in the spring of 2000, he was attracted to supercontinuum generation using photonic crystal fibers — which require only a femtosecond laser oscillator to generate a supercontinuum (SC). Later, after beginning his academic career in the Hamaguchi Lab at the University of Tokyo, he developed a home-built inverse Raman (now frequently referred to as a stimulated Raman loss, or SRL) spectroscopic system utilizing SC generation.⁶

During his experiment, Dr. Kano happened to find the CARS signal, which was very intense and actually much easier to detect than the inverse Raman (i.e., SRL) signal. He quickly realized that combining SC generation with a spectroscopic apparatus would be a breakthrough not only in the field of fundamental spectroscopy but in the life sciences as well, where spectroscopic imaging was just emerging.

Experimental Setup

Two laser sources were incorporated in Dr. Kano’s experimental setup, as were a new highreadout- speed CCD camera with ultrahigh NIR sensitivity and a high-throughput spectrometer. See Figure 2.

The first laser source, which was based on a master oscillator fiber amplifier (MOFA) configuration, involved a microchip oscillator — a passively Q-switched laser comprising an Nd:YVO₄ crystal bonded with a saturable absorber mirror — and a Yb-doped fiber amplifier. The second source was a passively Q-switched microchip Nd:YAG laser. The researchers were able to switch the laser sources based on the type of sample being studied, affording good experimental flexibility in terms of wavelength, temporal duration, repetition rate, and output average power.²

A newly available scientific CCD camera provided significantly higher NIR sensitivity than that afforded by the state-of-the-art CCD camera used by Dr. Kano and his associates during their previous ultra-multiplex CARS spectroscopic imaging study of living cells.⁷

As well as its greater sensitivity at important NIR in vivo imaging wavelengths, the new camera delivered the high-speed operation needed to synchronize with the ultrafast lasers used in Dr. Kano’s time-resolved studies. The researchers coupled the camera to a high-throughput spectrograph, which dispersed the lens-collected CARS signal for detection by the camera.

Before being utilized to image living cells, the experimental system was first evaluated by measuring the CARS signal of a polystyrene bead (diameter: 10 μm). The ultra-multiplex CARS signal of the bead was detected in the range from 600 cm^-1 to 3600 cm^-1 with resolution <10 cm^-1. The pixel dwell time was ~1 msec and CARS images of 161 × 161 pixels were acquired of the beads with a total data-acquisition time of ~28 sec (Figure 3).²

Imaging of Live A549 Cells

After verifying the performance of the system by imaging polystyrene beads, the researchers imaged a living cell (A549) using the high-speed, high-sensitivity CCD camera and the MOFA laser source (Figure 4). The sharp peak observed at 2850 cm^-1 on the spectrum rendered in red corresponds to one of the bright spots inside the cell, which in turn corresponds to the CH₂ stretching vibrational mode observed primarily in intracellular lipid droplets. Note that a raw CARS signal consists of a vibrationally resonant signal and a non-resonant background. These two components interfere with each other and produce dispersive line shapes.²

Dr. Kano and his colleagues next extracted the pure vibrationally resonant signal to obtain spontaneous-Raman-equivalent spectral profiles by performing numerical analysis. They found the main features of these resultant spectral profiles to be in good agreement with those of intracellular lipids and proteins, respectively. More than a dozen characteristic Raman bands between 3427 cm^-1 and 1009 cm^-1 corresponded to the vibrational modes.²

For additional data and a more detailed discussion of this study, please refer to the following article: Hideaki Kano, Takumi Maruyama, Junko Kano, Yuki Oka, Daiki Kaneta, Tiffany Guerenne, Philippe Leproux, Vincent Couderc, and Masayuki Noguchi, “Ultramultiplex CARS spectroscopic imaging with 1-millisecond pixel dwell time,” OSA Continuum 2, 1693–1705 (2019).

Progress and Outlook

The researchers performed ultra-multiplex (600 cm^-1 – 3600 cm^-1) CARS spectroscopic imaging on living cells, reporting the fastest time to date for broadband CARS whose spectral range is ~3000 cm^-1. Based on the clear molecular fingerprint, Dr. Kano and his colleagues visualized intracellular molecular distribution with more than 15 vibrational bands. The exposure time of 1 msec on software equated to an effective exposure time per pixel of ~1.8 msec, a significant reduction when compared to Dr. Kano’s previous study on living-cell imaging (in which the exposure time per pixel was 50 msec). The improvement was mainly attributable to the higher sensitivity and readout speed of the newly available CCD camera.^2,7

The combination of this high-speed technique and multivariate analysis methods can help life scientists and medical doctors gain meaningful insights into the dynamics of intracellular metabolic activity. Dr. Kano’s research group is currently collaborating with a pathologist to develop a new spectroscopic diagnostic tool using the molecular fingerprint.

Key Technologies

To facilitate their most recent work, the researchers selected a Teledyne Princeton Instruments BLAZE^® camera with a proprietary “super-deep-depletion” CCD manufactured from high-resistivity bulk silicon⁸. Designed to yield the highest near-infrared quantum efficiency of any silicon device available, the camera featured a back-illuminated 1340 x 400 spectroscopic array format (20 μm square pixels) capable of being cooled down to -95°C in air, without chillers or liquid assist, for low-dark-current performance.

BLAZE also offered the researchers the fastest ADC speeds available in a CCD camera. The new sensor platform’s dual 16 MHz readout ports were engineered to enable unprecedented spectral rates of more than 1600 spectra/second with full vertical binning and up to 215 kHz when operated in kinetics mode.

The spectrograph used in concert with the BLAZE camera was a Teledyne Princeton Instruments LS-785. Owing to its fast f/2 optical system with proprietary AR-coated lenses for optimum NIR transmission, the LS-785 can achieve up to 4x the throughput of a standard f/4 mirror-based spectrograph. All functions and timing of the CCD camera and the lens-based spectrograph were controlled and coordinated within the experimental setup via Teledyne Princeton Instruments LightField^® 64-bit software

Acknowledgment

Teledyne Princeton Instruments wishes to thank Dr. Hideaki Kano (University of Tsukuba, Japan) for his invaluable contributions to this application note.

To learn more about Dr. Hideaki Kano’s research at the University of Tsukuba, please visit: http://www.bk.tsukuba.ac.jp/~CARS/en/index.html

References