Open Access
Add to BibSonomyAbstract
An environmentally-stable low-repetition rate fiber oscillator is developed to produce narrow-bandwidth pulses with several tens of picoseconds duration. Based on this oscillator an alignment-free all-fiber laser for multi-photon microscopy is realized using in-fiber frequency conversion based on four-wave-mixing. Both pump and Stokes pulses for coherent anti-Stokes Raman scattering (CARS) microscopy are readily available from one fiber end, intrinsically overlapped in space and time, which drastically simplifies the experimental handling for the user. The complete laser setup is mounted on a home-built laser scanning microscope with small footprint. High-quality multimodal microscope images of biological tissue are presented probing the CH-stretching resonance of lipids at an anti-Stokes Raman-shift of 2845 cm−1 and second-harmonic generation of collagen. Due to its simplicity, compactness, maintenance-free operation, and ease-of-use the presented low-cost laser is an ideal source for bio-medical applications outside laser laboratories and in particular inside clinics.
©2012 Optical Society of America
1. Introduction
Coherent Anti-Stokes Raman scattering (CARS) microscopy is a potent technique for chemical selective tissue imaging [1]. By probing vibrational molecule resonances chemical information is obtained without the use of any labels. Hence, CARS allows for microscopy of living cells and is a promising technique for real time in vivo imaging, e.g. during brain cancer surgery [2]. However, CARS signal generation requires two synchronized picosecond pulse trains with their frequency difference matching the resonance frequency of interest. These are commonly generated using Ti:Sapphire or frequency doubled Nd-based bulk lasers in combination with a bulk optical parametric oscillator. Such systems are not only expensive and complex, but also require constant maintenance and alignment. A wider use of CARS in real-world applications, such as medical imaging in clinical environments, is crucially dependent on the development of compact, turn-key laser sources which are reliable and easy to use. Therefore, several attempts have been made to transfer the compactness and ruggedness of fiber laser technology to CARS laser sources. Nevertheless, while alignment-free fiber lasers are routinely operated in many industrial applications, the critical point for CARS still is the fiber based generation of a synchronized pulse train at a second wavelength. Besides electronic locking of two separate lasers [3] all-optical frequency conversion of a mono-laser source has been demonstrated [4–8], resulting in intrinsic passive synchronization of the two pulse trains. The fiber based frequency conversion is typically realized by soliton self-frequency shift [4–6]. However, in addition to the fiber sections all of these sources rely on free-space setups to adapt the generated pulses to suit the CARS process. Only recently optical parametric generation (OPG) by four-wave-mixing (FWM) in photonic-crystal fibers (PCF) [9,10] has been demonstrated as new approach to generate passively synchronized pump and Stokes pulses for CARS [7]. The proposed scheme offers several advantages since the fiber-generated long-picosecond pulses can directly be used for CARS and complete fiber integration including fiber delivery becomes feasible. Also continuous-wave seeded FWM using short few ps pulses has been applied [8]. Using a cw-seed reduces the bandwidth of the FWM generated pulses [11], which is advantageous for applications requiring high spectral resolution, e.g. probing narrow resonances in the fingerprint region. However, the use of 1-2 ps pulses as in [8] only results in moderate resolution; moreover, the short pulses hinder fiber integration both for the fiber amplifier due to increased nonlinear effects as for the CARS setup due to the extra delay lines needed to precisely control the temporal overlap.
In this contribution we present, to the best of our knowledge, the first alignment-free all-fiber laser source for CARS microscopy delivering synchronized pump and Stokes pulses from a single fiber end. Similar to the source proposed in [7] optical parametric frequency generation via in-fiber FWM is used to generate the CARS pump, while the residual fundamental laser pulses serve as CARS Stokes. The complete setup including the FWM stage is constructed of polarization maintaining (PM) fiber components, which are fusion spliced eliminating any free-space propagation. The ultra-compact setup is enabled by a novel ps master oscillator with low MHz repetition rate. This environmentally-stable, SESAM mode-locked all-fiber oscillator generates spectrally narrow pulses with 50ps duration. It is therefore perfectly adapted to the requirements, reducing the number of components and, thus, the complexity and the footprint of the system. The turn-key CARS laser source has great benefit for the user as the fiber delivery eliminates alignment of the spatial and temporal pulse overlap. Microscopic multimodal nonlinear imaging is performed by probing the CARS CH-stretching vibration of lipids around 2845cm−1 and by probing second harmonic generation in the sample simultaneously. High quality images of atherosclerotic plaques at the inner human arterial wall are presented.
2. The alignment-free all-fiber laser system
CARS microscopy of biological samples is limited to a certain window of laser parameters. Since CARS is a three-photon process the accumulated signal is increasing with the cubic of the peak power (Ppeak) multiplied by the duty-cycle (ηduty) of the laser. The sensitivity of the detection system hence gives a minimum power requirement (Ppeak3∙ηduty > Limsens). At the same time, the onset of cell damage limits the maximum power on the sample, while the damage mechanism is depending on tissue type, peak power and wavelength. Assuming, that cell damage is predominantly two-photon induced [12] the maximum condition is depending on the square of the peak power (Ppeak2∙ηduty < Limdamage). Hence, a high peak power and a rather low duty cycle extend the useful operation window. Consequently, a short pulse duration (τ) and low repetition rate (frep) is advantageous (ηduty = τ∙frep). In contrast, besides the decrease in spectral resolution, undesired NL effects and issues with temporal overlap and dispersion in optical elements, delivery fibers or in the case of deep tissue imaging become more and more critical for short pulses with few ps or sub-ps duration. A very low repetition rate, in turn, limits the scanning speed.
Pulses with tens of ps duration represent a good compromise. They are ideally suited for practical CARS systems as they facilitate the experiment drastically. At the same time, they allow for very high spectral resolution and reach the required peak power of typically 0.1-1 kW in combination with MHz repetition rates (required for video rate imaging) at a moderate average power (Pavg = Ppeak∙ηduty). Indeed, for example a 20 ps system with its repetition rate lowered to 2 MHz has the same duty cycle as a system with 500 fs at 80 MHz, and, hence, under the above assumptions it is fully equivalent in terms of efficient CARS signal generation and the onset of cell damage.
Such pulse durations and peak power levels can be generated by FWM in PCFs [7,9–11], where a launched pump pulse generates two new, equally separated frequency components at shorter (signal) and longer (idler) wavelengths. We decided on a target pulse duration of 20 ps as it is appropriate both for CARS imaging, due to the above mentioned reasons, as for pulse generation via FWM. Since the generated signal pulse is significantly shorter [7], this corresponds to a pump pulse between 50 and 60 ps, which is still long enough to keep spectral broadening via self-phase-modulation (SPM) reasonable and also long enough to generate a clean signal spectrum with suitable bandwidth, even without the use of a cw-seed keeping the system very simple.
To drive the FWM process a suitable all-fiber laser system was developed using only 6μm-core PM single-mode fiber (SMF). The following subsections describe the ps oscillator and the amplification and FWM stages.
Environmentally-stable all-fiber oscillator with low repetition rate
As outlined above the target system requires a pulse source, which delivers around 50 ps pulses with narrow bandwidth and a low MHz repetition rate. Since pulse picking and spectral filtering as in [7] complicate the system it is highly desirable to generate such pulses directly from a simple oscillator.
The simplest Yb-fiber oscillator concept is the all-normal dispersion laser omitting the dispersion compensation. Such lasers have been shown to operate at low repetition rate in a highly nonlinear regime with broad bandwidth stabilized by a several nm wide spectral filter [13]. On the other hand ps pulses with relatively narrow bandwidths have been obtained without spectral filter in linear regimes at high dispersion [14,15]. However, in these cases the time-bandwidth product has still been relatively large. Here, a 10 mm long uniform fiber Bragg grating (FBG) with a width of 0.06 nm is used to spectrally narrow the pulses and stabilize the mode-locking. The FBG is, moreover, crucial to fix the wavelength to exactly match the requirements of the FWM and finally the CARS resonance. The laser is constructed in a simple linear configuration as shown in Fig. 1 . The FBG represents one of the cavity end mirrors and with a peak reflectivity of 50% it also serves as output coupler. The other end mirror is formed by the semi-conductor saturable absorber mirror (SESAM) which is directly butt-coupled to the fiber. It has a modulation depth of 24%, a relaxation time of 9ps and a saturation fluence of 100 µJ/cm2. The fiber section, realized with PM Panda fiber, consists of a delay section of about 50 m, a polarizer to select the oscillation axis and a 1.2 m active fiber section with a pump absorption of 80 dB/m which is core-pumped through the FBG by a 976 nm single-mode diode.
Fig. 1 Schematic of the all-fiber CARS laser source, showing the ps oscillator (upper part) and the amplification and frequency conversion stages (lower part).
With a low repetition rate of 1.9 MHz this simple oscillator generates a pulse train of 45 to 55 ps pulses depending on the pump level. At a launched pump power of 11 mW it produces 0.13 nJ pulses with a duration of 51 ps and a spectral width of 0.07 nm as shown in Fig. 2 . The oscillator output is spliced to an optical isolator which is delivering the pulses to the amplification stages.
Fig. 2 Oscillator pulse characterization: (a) Temporal characteristic measured with a fast photo diode (19ps response time) and a 70GHz sampling oscilloscope; (b) Spectrum measured with a resolution of 0.02nm.
Core-pumped SMF amplifier and all-fiber frequency conversion by FWM
The low repetition rate reduces the average power requirements for the whole laser system; hence, the amplifiers can be constructed only of SMF components, making the system very compact. A 1m active fiber with a pump absorption of 250 dB/m is used to pre-amplify the pulses. This results in 7.5 mW seed launched into the 0.3 m long main amplification fiber (1200 dB/m pump absorption) which can deliver up to 0.5 W output power. With a resulting peak power of up to 5 kW the amplified pulses can easily drive the FWM process.
Efficient parametric signal and idler generation requires good overlap of the involved fiber modes and a suitable dispersion characteristic of the fiber to realize phase matching. Both can be fulfilled in endlessly single-mode PCFs which easily allow one to tailor the dispersion by changing the air-hole structure [10]. As the majority of biomedical CARS imaging applications focusses on probing the resonances of lipids [16] we looked for a commercially available PCF with appropriate dispersion characteristic to generate the FWM signal at a frequency difference matching the C-H-stretching vibrational resonance of methylene groups around 2845 cm−1 when being pumped around 1030 nm by an Yb-based fiber laser. For this purpose simulations analog to those in [7] were carried out and as a result a PCF offered by NKT Photonics named “LMA-5 PM” was identified to meet our requirements. The fiber has 4 rings of 1.2 µm air holes with a hole-to-hole separation of 3.2 µm. On the basis of the fiber’s phase-matching characteristic (see Fig. 3(a) ) the central wavelength of the pump laser was chosen accordingly to generate the desired frequency shift of 2845 cm−1. As can be deduced from the slope in Fig. 3(a) the frequency shift changes by 6.5 cm−1 per 0.1 nm pump wavelength variation. Due to manufacturing tolerances and uncertainties concerning the fiber parameters an experimental FWM measurement is required to test the frequency shift of the actual piece of PCF. With this measurement the simulation can be adapted to precisely design the target wavelength of the oscillator to finally meet the desired CARS excitation frequency.
Fig. 3 (a) FWM phase-matching diagram for the PCF showing the frequency shift and the corresponding signal wavelength over pump wavelength; (b) Microscope image of the splice between Yb-doped SIF (left) and PCF (right), the splice transmission is 74%.
The PCF was directly spliced to the main amplifier fiber using arc fusion splicing. Previous alignment of the Panda structure of both fibers was executed to ensure coupling into the slow axis of the PCF. The relatively similar mode-field diameters of nominal 6.0 μm (SIF) and 4.4 μm (PCF) allow for a high theoretical coupling efficiency of 91% assuming Gaussian mode profiles. After optimization of the splice procedure parameters we obtain an excellent experimental value of 74% for the splice shown in Fig. 3(b). As can be seen the collapse length of the PCF air holes is kept very small. At the same time the splice is still strong enough to withstand pulling and bending during experimental handling. Finally, the PCF is cut to a length of 0.6 m and the fiber end is collapsed and angle polished.
At the end of this alignment-free configuration we measure 10.5 mW FWM signal at a pump power of 290 mW launched into the PCF. By turning up the pump this increases to 50 mW signal @ 345 mW of pump. Due to the increasing parametric gain bandwidth, SPM-broadening of the pump spectrum and nonlinear broadening of the signal pulse itself, the bandwidth of the generated signal grows with increasing power. This dependency is shown in Fig. 4(d) where the spectral width of both signal and residual pump is plotted for different signal levels. The graph demonstrates that power levels of several ten mW which are more than sufficient for CARS imaging applications can easily be generated with reasonable spectral widths of several ten wavenumbers.
Fig. 4 (a)-(c) Characterization of the generated pulses at 26mW of signal power: Spectral shape of FWM signal (a) and corresponding residual pump (b), temporal shapes accordingly, measured with a fast photo diode (c) and signal pulse autocorrelation trace (c, inset). (d) Spectral width of FWM signal and residual pump for different signal power.
The spectral shape of the FWM signal (serving as CARS pump) and the spectrum of the residual FWM pump (serving as CARS Stokes) are shown in Figs. 4(a), and 4(b) exemplarily for a signal power of 26 mW, which corresponds to the power level used in the subsequent CARS experiments. The spectra are centered around 797.9 nm and 1032.1 nm matching the desired frequency separation of 2845 cm−1. Due to the high peak power the pump spectrum shows significant SPM structures and it is broadened to 0.8 nm. The temporal pulse shapes were measured with a fast photo diode with a response time of 18.5 ps and a 70 GHz sampling oscilloscope. The recorded traces for the partially depleted pump pulse and the signal pulse are shown in Fig. 4(c). As the conversion starts only at the center of the pulse the generated signal is significantly shorter than the pump. It therefore reaches the detection limit of the equipment, hence, an additional autocorrelation measurement was performed (inset of Fig. 4(c)) measuring a pulse duration of 19 ps under the assumption of a Gaussian shape. From this we can estimate the peak power to 0.7 kW. The temporal walk-off between pump and signal in the PCF due to group-velocity mismatch is about 0.1 ps/cm and, hence, it is negligible regarding the pulse duration and fiber length. In consequence, a good temporal overlap of the pulses is automatically ensured for the CARS process.
3. CARS imaging
To obtain a complete and compact CARS imaging system a scanning microscope was developed, which is simple to use and does not require an optical table. The optical layout has been simplified since a confocal microscope design is not required; instead the nonlinear processes provide intrinsic 3D sectioning capabilities. Due to the nonlinear nature of CARS the spatial overlap of the pump and Stokes pulses in the focal spot is critical. Thus, conventional systems require careful adaptation and overlay of the separate pump and Stokes beams including control of the temporal pulse overlap, which make the alignment time-consuming and difficult and increase space requirements and system complexity. For the experiment reported here, the fiber output of the laser is directly attached to the microscope as illustrated in Fig. 5 . Pump and Stokes pulses are emitted simultaneously eliminating any need for alignment. The collimated beam is deflected by a pair of scanning galvanometric mirrors and focused using a 20x microscope objective with NA 0.4 (Mitutoyo Plan Apo NIR). Finally, the forward generated CARS signal is detected with a photo-multiplier tube (PMT). All elements in the illumination path are designed for the near infrared reducing reflection losses. Furthermore, by minimizing the distance of the sample to the detector also scattered signal photons can be efficiently detected. For medical applications a large field of view (FOV) is beneficial, whereas most commercial microscopes have only a limited FOV, which is optically highly corrected. In contrast, the FOV of the microscope reported here has been maximized and exceeds 1mm2 when using a 20x objective.
Fig. 5 Schematic of the CARS microscopy setup, SP short pass, LP long pass, BP band pass, PMT photo multiplier tube.
To reduce the total power on the biological sample, the output of the PCF containing FWM signal, residual pump and idler is spectrally filtered. A short pass with its edge at 1050 nm blocks the long-wavelength idler. Tilting this filter, moreover, allows one to continuously decrease the power fraction transmitted at 1032 nm. Hence, the amount of Stokes available for CARS can be adjusted. An additional long-pass filter blocks any light emitted at the CARS wavelength, which is generated in small amounts by non-phase-matched FWM in the PCF [17]. High-quality images are obtained with a total power of less than 20 mW on the sample. In front of the PMT the residual laser light is blocked by two short-pass filters, moreover, a band-pass filter ensures that only CARS photons are detected.
Multimodal nonlinear microscopic images utilizing CARS, SHG and TPEF of a human aorta section are presented in Fig. 6 . These were recorded with a total average power on the sample of 50 mW (a)-(c) and 20 mW (d) respectively. Sections of 1.2 x 1.2 mm2 were scanned with a pixel dwell time of 1 μs. Figure 6(a) shows the image obtained when the CARS signal at 650 nm is detected. Probing the CH-stretching vibration with a laser frequency difference of 2844 cm−1, lipid plaques are clearly visible in this image indicating atherosclerosis. The non-resonant signal contribution in the image is small; the ratio of resonant signal to non-resonant background for the lipid plaques is about 10:1, which has been measured by tuning off-resonance with a slightly modified setup. By exchanging the band-pass 650 nm filter in front of the detector with a short-pass 600 nm, the combined second harmonic generation (SHG) signal of collagen at 399 and 516 nm and the two-photon excited fluorescence signal (TPEF) of elastin is recorded (Fig. 6(b)). Due their isotropic nature the lipid droplets appear dark in 6(b); while fibers composed of the structural proteins collagen and elastin become visible [18]. The multimodal composite image is shown in Fig. 6(c).
Fig. 6 Microscopic images of atherosclerotic plaque deposition at the inner human arterial wall. (a)-(c) show the same sample section: (a) CARS image probing lipids at 2844 cm−1, the atherosclerotic plaques appear bright due to their high lipid content, (b) combined SHG and TPEF signal of the elastic fibers of the arterial wall predominantly composed of collagen (SHG) and elastin (TPEF), (c) Multimodal composite image showing an overlay of the CARS and the SHG/TPEF signals. The images were sampled with 16.8 Mpixels with a pixel dwell time of 1 μs, 2 scans averaged, 50 mW total laser power on the sample. (d) Multimodal composite image as in (c) showing a different aorta section, recorded with 20 mW total laser power on the sample.
4. Conclusion
In this contribution a completely alignment-free, simple and potentially low-cost all-fiber laser source for nonlinear and especially CARS microscopy is presented. The CARS pump pulses are produced by optical parametric generation via four wave mixing in a photonic-crystal fiber, which gives intrinsic passive synchronization of the pump and Stokes pulses. The laser relies on a long-cavity passively mode-locked Yb fiber oscillator, developed to produce spectrally narrow 50 ps pulses with a low repetition rate of 2 MHz. These parameters allow for direct amplification to 5 kW peak power using standard core-pumped single-mode fibers, which gives sufficiently high intensities to drive the frequency generation in the PCF. The complete setup is constructed of commercially available, polarization-maintaining fiber components which are fusion spliced eliminating any alignment.
CARS pump (798 nm) and Stokes (1032 nm) pulses are emitted from a single fiber end which is directly attached to a home-built laser scanning microscope. The single-fiber delivery drastically eases the experimental effort as the pulses are perfectly overlapped in space and time. This reduces the alignment to a minimum which is of particular importance for the user. The relatively long pulse durations are advantageous as they ensure temporal overlap even after additional optical elements, e.g. microscope objective lenses.
High-quality multimodal nonlinear microscopic imaging is demonstrated, presenting resonant imaging of lipids at a wavenumber shift of 2845 cm−1 in combination with SHG and TPEF showing plaque deposition by arteriosclerosis in a human aorta section.
With its hands-off operation and compact dimensions the presented light source makes the application of CARS microscopy interesting for a wider community. Additional features as the fiber delivery and the extremely simplified handling particularly pave the way to the application of CARS in clinics.
Acknowledgments
This work was supported by the German Federal Ministry of Education and Research (BMBF) [13N10773], [13N10774]; the European network of excellence P4L (Photonics4Life); and the Ministerio de Economía y Competitividad of Spain [TEC2008-05490]. M.B. acknowledges support from the Carl-Zeiss-Stiftung. The authors like to thank Christian Matthäus for providing the biological sample.
References and links
1. C. L. Evans and X. S. Xie, “Coherent anti-stokes Raman scattering microscopy: chemical imaging for biology and medicine,” Annu Rev Anal Chem (Palo Alto Calif) 1(1), 883–909 (2008). [CrossRef] [PubMed]
2. C. Krafft, B. Dietzek, and J. Popp, “Raman and CARS microspectroscopy of cells and tissues,” Analyst (Lond.) 134(6), 1046–1057 (2009). [CrossRef] [PubMed]
3. S. Bégin, B. Burgoyne, V. Mercier, A. Villeneuve, R. Vallée, and D. Côté, “Coherent anti-Stokes Raman scattering hyperspectral tissue imaging with a wavelength-swept system,” Biomed. Opt. Express 2(5), 1296–1306 (2011). [CrossRef] [PubMed]
4. E. R. Andresen, C. K. Nielsen, J. Thøgersen, and S. R. Keiding, “Fiber laser-based light source for coherent anti-Stokes Raman scattering microspectroscopy,” Opt. Express 15(8), 4848–4856 (2007). [CrossRef] [PubMed]
5. A. F. Pegoraro, A. Ridsdale, D. J. Moffatt, J. P. Pezacki, B. K. Thomas, L. Fu, L. Dong, M. E. Fermann, and A. Stolow, “All-fiber CARS microscopy of live cells,” Opt. Express 17(23), 20700–20706 (2009). [CrossRef] [PubMed]
6. G. Krauss, T. Hanke, A. Sell, D. Träutlein, A. Leitenstorfer, R. Selm, M. Winterhalder, and A. Zumbusch, “Compact coherent anti-Stokes Raman scattering microscope based on a picosecond two-color Er:fiber laser system,” Opt. Lett. 34(18), 2847–2849 (2009). [CrossRef] [PubMed]
7. M. Baumgartl, M. Chemnitz, C. Jauregui, T. Meyer, B. Dietzek, J. Popp, J. Limpert, and A. Tünnermann, “All-fiber laser source for CARS microscopy based on fiber optical parametric frequency conversion,” Opt. Express 20(4), 4484–4493 (2012). [CrossRef] [PubMed]
8. S. Lefrancois, D. Fu, G. R. Holtom, L. Kong, W. J. Wadsworth, P. Schneider, R. Herda, A. Zach, X. Sunney Xie, and F. W. Wise, “Fiber four-wave mixing source for coherent anti-Stokes Raman scattering microscopy,” Opt. Lett. 37(10), 1652–1654 (2012). [CrossRef] [PubMed]
9. L. Lavoute, J. C. Knight, P. Dupriez, and W. J. Wadsworth, “High power red and near-IR generation using four wave mixing in all integrated fibre laser systems,” Opt. Express 18(15), 16193–16205 (2010). [CrossRef] [PubMed]
10. D. Nodop, C. Jauregui, D. Schimpf, J. Limpert, and A. Tünnermann, “Efficient high-power generation of visible and mid-infrared light by degenerate four-wave-mixing in a large-mode-area photonic-crystal fiber,” Opt. Lett. 34(22), 3499–3501 (2009). [CrossRef] [PubMed]
11. P. J. Mosley, S. A. Bateman, L. Lavoute, and W. J. Wadsworth, “Low-noise, high-brightness, tunable source of picosecond pulsed light in the near-infrared and visible,” Opt. Express 19(25), 25337–25345 (2011). [CrossRef] [PubMed]
12. K. König, T. W. Becker, P. Fischer, I. Riemann, and K. J. Halbhuber, “Pulse-length dependence of cellular response to intense near-infrared laser pulses in multiphoton microscopes,” Opt. Lett. 24(2), 113–115 (1999). [CrossRef] [PubMed]
13. W. H. Renninger, A. Chong, and F. W. Wise, “Giant-chirp oscillators for short-pulse fiber amplifiers,” Opt. Lett. 33(24), 3025–3027 (2008). [CrossRef] [PubMed]
14. M. Baumgartl, B. Ortaç, J. Limpert, and A. Tünnermann, “Impact of dispersion on pulse dynamics in chirped-pulse fiber lasers,” Appl. Phys. B 107(2), 263–274 (2012). [CrossRef]
15. X. Tian, M. Tang, X. Cheng, P. P. Shum, Y. Gong, and C. Lin, “High-energy wave-breaking-free pulse from all-fiber mode-locked laser system,” Opt. Express 17(9), 7222–7227 (2009). [CrossRef] [PubMed]
16. J. P. Pezacki, J. A. Blake, D. C. Danielson, D. C. Kennedy, R. K. Lyn, and R. Singaravelu, “Chemical contrast for imaging living systems: molecular vibrations drive CARS microscopy,” Nat. Chem. Biol. 7(3), 137–145 (2011). [CrossRef] [PubMed]
17. M. Balu, G. Liu, Z. Chen, B. J. Tromberg, and E. O. Potma, “Fiber delivered probe for efficient CARS imaging of tissues,” Opt. Express 18(3), 2380–2388 (2010). [CrossRef] [PubMed]
18. W. R. Zipfel, R. M. Williams, R. Christie, A. Y. Nikitin, B. T. Hyman, and W. W. Webb, “Live tissue intrinsic emission microscopy using multiphoton-excited native fluorescence and second harmonic generation,” Proc. Natl. Acad. Sci. U.S.A. 100(12), 7075–7080 (2003). [CrossRef] [PubMed]
