Open Access
Add to BibSonomyAbstract
In coherent anti-Stokes Raman scattering (CARS) and second harmonic generation (SHG) imaging, backward and forward generated photons exhibit different image patterns and thus capture salient intrinsic information of tissues from different perspectives. However, they are often mixed in collection using traditional image acquisition methods and thus are hard to interpret. We developed a multimodal scheme using a single central fiber and multimode fiber bundle to simultaneously collect and differentiate images formed by these two types of photons and evaluated the scheme in an endomicroscopy prototype. The ratio of these photons collected was calculated for the characterization of tissue regions with strong or weak epi-photon generation while different image patterns of these photons at different tissue depths were revealed. This scheme provides a new approach to extract and integrate information captured by backward and forward generated photons in dual CARS/SHG imaging synergistically for biomedical applications.
© 2016 Optical Society of America
1. Introduction
Both coherent anti-Stokes Raman scattering (CARS) and second harmonic generation (SHG) imaging were proposed decades ago [1, 2], and have since been markedly developed with the emergence of ultrafast pulsed infrared lasers [3, 4]. Both have clear, potential use in clinical applications due to their label-free, nonlinear nature. CARS is a third-order nonlinear process. When the frequency difference between pump and Stokes waves matches a vibrational frequency of the target molecules, the coherently driven molecular oscillators emit strong signals at the anti-Stokes frequency within the focal volume. With the beating frequency tuned to 2845 cm−1 Raman shift, tissues rich in CH2 chemical bonds, such as single lipid bilayers [5], mouse brain tissue [6], and several types of cancer cells [7–9] can be observed at high contrast [10] using CARS microscopy. SHG, on the other hand, is a second-order nonlinear process, and only arises from materials lacking central symmetry. Collagen and fibrous tissues, such as fascia [11], tendon [12, 13], and muscle [14], thus generate strong SHG signals.
In CARS and SHG imaging, signals are generated in both the forward and backward directions. Assuming that the excitation beams are monochromatic plane waves, by solving Maxwell’s equations one knows that the signal intensity of CARS and SHG is proportional to the square of the interaction length L and the phase matching term
where represents the phase mismatch [15]. is defined as the coherence length at which the first signal maximum is generated. When perfect phase matching is achieved, i.e. , the coherence length becomes very large and the phase matching term reaches its maximum. However, the complexity of biological tissues may alter the strict phase matching condition. LaComb et al. [16] have built a mathematical model based on a relaxed phase matching condition that accounts for dispersion, randomness, and axial momentum contributions from the media. In this model, the overall signal intensity contains contributions from a series of non-zero values where small values corresponding to higher nonlinear conversion efficiency are dominant in forward signals. In contrast, backward signals are associated with large values which consequently have smaller coherence lengths. This effect has also been revealed for collagen in tendon, fascia, muscle, and cornea [11, 17–23] where small, segmented, and punctate fibrils are observed in the backward SHG channel, while these same fibrils appear to be long and continuous in the forward SHG channel.In biological imaging, tightly focused excitation beams are normally used rather than plane waves, thus a substantial revision of the theory is required. Richard and Wolf described the focal fields of tightly focused plane waves in 1959 [24]. Theoretical modeling has been demonstrated by Cheng et al. [25] for CARS and Moreaux et al. [26] for SHG under tightly focused condition. Both models consider Gouy phase shift [15] that occurs when a Gaussian beam traverses through its focus in their analysis. The resulting forward-generated photons are confined to a small cone. Volkmer et al. [27] has visualized cell nucleus with apparent substructure and the diffraction-limited features within the cytoplasm in epi-detected CARS using tightly focused laser beams. Cheng et al. [28] and Schie et al. [29] have also found that backward-generated CARS photons provide information of structures with axial length smaller than the excitation wavelength while forward-generated CARS photons provide information of objects with comparable or larger size than the excitation wavelength. Therefore, in CARS and SHG imaging, the backward-generated and forward-generated photons provide tissue information from different points of view.
There are two types of photons traveling in the backward direction: backward-generated photons and backscattered forward-generated photons. Adopting the naming rules from [30], in this manuscript, backward-generated photons are called BG-photons, forward-generated photons are called FG-photons, and backscattered forward-generated photons are called BS-photons. In the backward channel, FG-photons are detected as BS-photons. However, BG- and BS-photons are often mixed in currently existing CARS or SHG systems. For example, the BG-photons are easily overwhelmed by the BS-photons in thick tissue imaging due to the increased backscattering of FG-photons [25, 31].
As the information carried by BG- and FG-photons are both critical for pathology analysis in clinical applications [17, 20, 23, 32–34], it is necessary to develop a method to differentiate BG- and BS-photons in CARS/SHG imaging. For CARS microscopy, taking advantage of the travel time difference between BG-CARS and BS-CARS photons, Schie et al. proposed a time-correlated single photon counting detection scheme to simultaneously resolve these photons through backward collection in CARS microscopy [29]. In SHG microscopy, images of BG-photons were obtained with thin tissues around 10 μm thick, in which backscattering of FG-photons is suppressed [11]. Glycerol has been used to suppress the backscattering of FG-photons in thick tissue in SHG microscopy [17]. Additionally, optical fibers are an essential part in most in vivo endomicroscopy modalities. Small core single excitation fibers have been tested for collecting BG-photons [35]. Fiber bundle collection has been applied to CARS endomicroscopy with multimode fibers (MMF) surrounding the central excitation fiber for collection of photons [36, 37].
In this paper, we combine the model used in [38] and [30] with the fiber bundle collection scheme and present a simultaneous two channel collection scheme with the single central fiber (SCF) as one collection channel and a bundle comprising 18 MMF surrounding the SCF as the other to differentiate BG- and BS-photons in backward detection in our CARS/SHG dual modality endomicroscopy prototype. We systematically compare the collection characteristics of the SCF and the MMF bundle in CARS/SHG endomicroscopy imaging. The experimental results indicate that the SCF is suitable for collecting BG-photons while the MMF bundle is suitable for collecting BS-photons. Within the same imaging region, images formed by BG-photons and BS-photons separately were revealed by this method. In SHG imaging of 1 mm thick mouse tail tendon tissue, epi-photon images captured the fine, heterogeneous features, while scattered BG-photons revealed the framework distribution pattern. We also quantified the ratio of BG- and BS-photons collected by the SCF in SHG imaging of the mouse tail tendon, and the great difference of this value at the regions with or without strong BG-photon generation indicates its potential application as a quantitative factor in differentiating these regions and monitoring BG- and FG-photon generation status. SHG images formed by BG-photons at the surface, inside and bottom of the mouse tail tendon were acquired and showed different pattern features.
2. Materials and methods
The schematic of our CARS/SHG endomicroscope prototype is shown in Fig. 1. The light source for the pump wave is a picosecond optical parametric oscillator (OPO) based on a non-critically phase-matched parametric interaction in LBO crystal (Levante, APE, Berlin, Germany). The OPO is pumped by the second harmonic (532 nm) of a mode-locked Nd:YVO4 laser (High-Q Laser, Hohenems, Austria). The laser delivers a 7 ps, 76 MHz pulse train at both 532 nm and 1064 nm. The 1064 nm pulse is used as the Stokes beam and the 5 ps OPO signal as the pump beam ranging from 670 to 980 nm. The pump beam is tuned to 817 nm, resulting in a 2845 cm−1 Stokes shift with the 1064 nm Stokes beam. The pump and Stokes beams are overlapped with a time-delay line and a dichroic mirror (DM1) in time and space. A 1/2-λ waveplate and a linear polarizer are placed in both pump and Stokes arms for rotating the orientation of the linear states of polarization (SOPs) of the two beams. The two beams are coupled into the 1 meter long excitation fiber (i.e. PM1300-HP) with a 10 × objective #1 (NA = 0.25, Newport, Irvine, CA) at orthogonal SOPs to suppress the four-wave-mixing (FWM) effect in fiber [39]. Then, the two beams are collimated with a collimator (M-10 × , NA = 0.25, Newport). After the collimator, a customized dual wavelength waveplate (DWW) [39] is used to convert the SOPs of the two beams from orthogonal to identical, as a result of which maximal CARS generation efficiency at samples is achieved. After being reflected by a silver mirror, the two beams are focused on samples with a 1.1-NA water immersion objective lens (Olympus, Japan). Samples are mounted on a high precision XYZ translation stage (Nano-LPS300, Mad City Labs, Madison, MI) for submicrometer-level fine scanning. The XYZ translational stage is mounted on a long travel XY translation stage (MAX202, Thorlabs) for millimeter-level rough scanning. The BG- and BS-photons are coupled back into the system by the 60 × objective, reflected by the mirror, and focused onto the common endface of the excitation fiber and the fiber bundle by the collimator.
Fig. 1 Schematic of CARS/SHG endomicroscope prototype with simultaneous SCF and MMF bundle collection.
At the center of the common endface is the SCF (PM1300 excitation fiber, 8 μm core diameter, 125 μm cladding diameter, 245 μm coating diameter, NA = 0.12). This fiber has been proven to effectively deliver the 817 nm and 1064 nm excitation lasers [40]. The four-wave-mixing noise in the excitation fiber is suppressed with our polarization control scheme [41]. Surrounding the SCF is the 18 MMF bundle. The 18 MMF (200 μm core diameter, 220 μm cladding diameter, transmission range: 350-2400 nm, NA = 0.22, length 1 meter) are separated from the excitation fiber near the common end by bifurcation. The 18 MMF are arranged to form a hexagonal active area with a diameter of 1.26 mm and a filling factor of 48% at the common endface.
Two photomultiplier tubes (PMT, R3896, Hamamatsu) are used for simultaneous detection of photons collected by the SCF and the MMF bundle. To detect photons collected by the MMF bundle, PMT1 is placed at its collection end. To detect photons collected by the SCF, a dichroic mirror (DM2) is inserted before microscopy objective #1 to redirect signal photons detected by SCF to PMT2. We use four HQ660/40 nm (Chroma, Bellows Falls, VT) bandpass filters for 663 nm CARS signal collection and switch to two ZET405/20x bandpass filters (Chroma Technology) to collect 406 nm SHG signals generated with the 812 nm pump beam from the tuned OPO. An NI PCI-6115 data acquisition system (DAQ) is used to receive synchronization signals from the translation stage, supply gain control voltage to the PMT, and acquire the voltage signal from the PMT. This DAQ possesses simultaneous sampling ability with four 12-bit analog input channels. Special attention should be paid to the problem of missing samples when using an external clock for this DAQ, which is caused by the pipeline first-in first-out (FIFO) buffer built into the analog to digital converter of the 611x family of DAQs [42]. A computer is used to control the DAQ and the scanning stage, and process data. LabVIEW is used to build the control and imaging software. Limited by the translation stage, pixel dwell time for scanning is 267 μs.
3. CARS imaging results and discussions
3.1 Collection analysis in CARS
We first tested the performance of the 18 MMF bundle collection, including collection efficiency and resolution performance. Polystyrene beads (PEBs) with 10 μm diameter were used for collection efficiency measurement. Similar to our previous study [37], we acquired CARS images by the 60 × objective and the MMF bundle, as shown in Fig. 2(a) and Fig. 2(b) respectively. The image collected by the 60 × objective was obtained by inserting a dichroic mirror above the 60 × objective and focusing the CARS photons to a PMT with a lens. Intensity profiles along the green line in the images are shown in Fig. 2(c). By integrating intensity along the green line, we can determine the collection efficiency of the MMF bundle compared to the 60 × objective along that line. By integrating the intensity of the entire image, the influence of noise is decreased. The collection efficiency of the MMF bundle is measured to be 14.7% relative to the 60 × objective. The collection efficiency of the single central fiber was measured as 0.73% in our previous work [37].
Fig. 2 CARS images of 10 μm polystyrene beads (PEB) collected by (a) 60 × objective and (b) MMF bundle. (c) Intensity profiles along green lines in (a) and (b). Scale bar is 10 μm.
The resolution of MMF bundle collection was measured with 1 µm PEB. Figures 3(a) and 3(b) show CARS images of 1 µm PEB collected by the 60 × objective and by the MMF bundle, respectively. Lateral and axial resolution measurement results of 60 × objective collection are shown in Figs. 3(b) and 3(c) while that of the MMF bundle collection are shown in Figs. 3(e) and 3(f). For 60 × objective collection, the lateral resolution is 1.04 µm and axial resolution is 6.42 µm. For MMF bundle collection, the lateral resolution deteriorates to 1.29 µm due to coupling loss, whereas the axial resolution is improved to 5.27 µm due to the pinhole effect of the MMF bundle at the active area of the common endface. Based on our scanning setup, we measured the axial resolution by Z-stack scanning of the 1 µm PEB and averaging the intensity of each image, and noise was greatly reduced in the axial intensity curve relative to the lateral intensity curve of the MMF bundle, as shown in Figs. 3(f) and 3(e), respectively. We could barely detect a CARS image of 1 µm PEB with SCF collection due to the weak signal generated in partial focal volume and large coupling loss, and thus could not precisely measure its collection efficiency and resolution from 1 µm PEB (data not shown).
Fig. 3 CARS imaging resolution measurement with 1 μm PEB. 60 × objective collection results: (a) CARS image of 1 μm PEB, (b) lateral profile of 1 μm PEB and (c) averaged intensity of Z scan images of 1 μm PEB. MMF bundle collection results: (d) CARS image of 1 μm PEB, (e) lateral profile of 1 μm PEB and (f) averaged intensity of Z scan images of 1 μm PEB.
3.2 Comparison of SCF and MMF bundle collection in CARS imaging
CARS images of mouse skin adipocytes simultaneously collected by the SCF and the MMF bundle are shown in Fig. 4. Figures 4(a)-4(d) show a Z-scan stack with a step size of 5 μm collected by the MMF bundle. The 5 μm step size was set according to the 5.27 μm axial resolution of the MMF bundle collection measured above. The laser focus is at the upper surface of the adipose cell in Fig. 4(a), and the edges of the cells are clearly illustrated. Several air bubbles, which were generated during the preparation of the sample, are shown as dark spots located at the surface of the cell on the left. The laser focus is 5 μm deeper into the adipose cells in Fig. 4(b). At the edges of the cells, the signals generated across the focal volume overlap and the edges do not show as sharp as that shown in Fig. 4(a). As the laser is focused to a depth of 10 μm in Fig. 4(c), the bubbles shown in Fig. 4(a) leave blurred shades, while new bubbles with sharp edges emerge, indicating that the focus has reached the bottom of the cell. The edges of the cells are farther from each other at the bottom. In Fig. 4(d), the adipose cells become out of focus and show blurred images at the depth of 15 μm, while the edges of the adipose cells in the next layer begin to show up, indicating that the laser focus is at a plane between these two layers. The images collected by the SCF are shown in Figs. 4(e)-4(h).
Fig. 4 CARS images of mouse adipocytes with Z-axis step of 5 μm collected by the MMF bundle (a-d), the single central fiber (e-h), and color overlapped images (i-l) with the SCF images in green and the MMF bundle images in red. General tissue structure is revealed by MMF bundle (a-d), fine structure by SCF in (e-h), especially the blood capillary-like structure, indicated by the yellow arrows in (g, h). Scale bars are 20 μm.
Comparing the images collected by the SCF with those collected by the MMF bundle, it can be noticed that the signal to noise ratio (SNR) shown in Figs. 4(e)-4(h) is much lower due to the weak collection efficiency (0.73%) of the SCF [37]. The image patterns are also displayed differently; some fine structures (indicated by the yellow arrows) similar to blood capillaries begin to emerge in Fig. 4(g) and become more distinct in Fig. 4(h), which are not discernable in Figs. 4(c) and 4(d). The diameter of these structures is estimated to be 8 μm, matching the size of a typical blood capillary. The signals inside the capillaries should originate from erythrocytes [10]. As discussed previously, the laser focus has reached the intermediate plane between the two layers of adipocytes at the depth of 15 μm. It is reasonable to assume the existence of blood capillaries between these adipocytes, as each adipose cell must be connected to at least one blood capillary or vessel to obtain nutrition and pump out metabolites. Therefore, the SCF collection channel can reveal important biological information that the MMF bundle collection channel lacks. The intrinsic difference between backward and forward generated photons is the leading contribution to this phenomenon, which will be further discussed later in this section. Figures 4(i)-4(l) represent the superimposed images of these two collection channels, where the SCF images are colored in green and the MMF bundle images are colored in red.
The distribution of the MMF bundle at the common endface is shown in Fig. 5(a). The fibers are grouped and labeled according to their radial location; groups 1, 2, and 3 are located 245 μm, 424.35 μm, and 490 μm from the center, respectively. The collection light cones with an NA of 0.22 at the common endface are shown in Fig. 5(b). The SCF collects photons that fall and propagate backward in its collection cone after all the scattering events they experience before arriving at the objective lens. This case was already proved in [35], in which a fiber with 5.6 µm mode field diameter (MFD) collected BG-CARS photons generated by 0.75 µm PEBs. BS-CARS photons rarely enter the collection cone of the SCF due to the pinhole effect of the small core. On the other hand, the photons that fall and propagate backward within the collection cones of the MMF bundle are collected by the MMF bundle. BG-CARS photons that are generated at the edges of large scatterers, due to incomplete destructive interference, are largely collected by the MMF bundle, as shown in BG-CARS images of large PEBs [40] and human adipocytes [29]. As the quantity of FG-CARS photons can be orders of magnitude larger than BG-CARS photons [25] and backscattering of FG-CARS photons occurs frequently in tissue [31], BS-CARS photons which fall within and propagate in the collection cones of the MMF bundle are the dominant source of photons collected by the MMF bundle, as well as a factor causing higher collection efficiency for MMF bundle collection than for SCF collection.
Fig. 5 Modeling of MMF bundle and single central fiber collection. (a) Distribution of the collection fiber bundle at the common endface. (b) Collection light cones of fiber bundle at the common endface.
Therefore, the SCF is responsible for collecting BG-CARS photons, and the MMF bundle surrounding the SCF is suitable for collecting the majority of BS-CARS photons, and some BG-CARS photons at the edge of scatterers. As BG-CARS photons are sensitive to objects of sub-wavelength size due to incomplete destructive interference [25], the SCF collection can be used to reveal fine tissue structures which correspond to small coherence length of scatters. As large scatterers can generate strong BS-CARS photons in tissue, the MMF bundle collection is suitable for revealing general tissue structures which correspond to large coherence length of scatters. Thus, the signals that reveal blood capillaries in Figs. 4(g) and 4(h) should originate from BG-CARS photons. Besides the low SNR, the other white speckles in Figs. 4(e)-4(h) may also come from segmented small structures revealed by BG-CARS photons. In contrast, the MMF bundle images are overwhelmed by BS-CARS photons generated by larger scatterers. Similarly, the BG-CARS image of adipocytes in [29] also shows strong intensity at the edges due to strong BG-CARS generation at the interface with different third-order susceptibility, and weak intensity at the center of the adipocyte due to destructive interference of BG-CARS waves induced by large objects [25].
4. SHG imaging results and discussions
4.1 Collection analysis in SHG
The model proposed by Han et al. [38] and extended in [30] describes the intensity distribution of BG-SHG and BS-SHG at the confocal plane. We apply this model and obtain the intensity distribution of B-SHG at the confocal plane in Fig. 6(a), which is a combination of BG-SHG distributed in a Gaussian profile and BS-SHG distributed uniformly. The distribution of BG-SHG and the cores of the MMF bundle at the common endface are shown in Fig. 6(b). As the cores of the MMF bundle are located 145 μm to 590 μm away from the center of the common endface, and due to the sharp peak of the Gaussian distribution of BG-SHG, the portion of BG-SHG collected by the MMF bundle can be neglected.
Fig. 6 (a) Intensity distribution of backward SHG (B-SHG) at the confocal plane. It is a combination of uniformly distributed backward-scattered forward-generated SHG (BS-SHG) and the backward-generated SHG (BG-SHG) with Gaussian distribution. The peak intensity of BG-SHG is set as 1, 1/e2 half width ω is set as 52.9 μm and intensity of BS-SHG is set as 0.2. (b) Intensity distribution of BG-SHG in Gaussian profile at the confocal plane. The three hollow disks represent the cores of three groups of multimode fibers (MMFs) shown in Fig. 5(a) with core center at 245 μm, 424.35 μm, and 490 μm radial position respectively. Parameters of BG-SHG are set same as (a). The MMF bundle is located far from the 1/e2 half width of the Gaussian profile and thus can collect little BG-SHG compared to the single central fiber (SCF) located at the center of the Gaussian profile.
The intensity of total backward SHG (B-SHG) at the confocal plane can be expressed as a combination of BG-SHG and BS-SHG in the following form [30]:
where is the radial distance to center of the confocal plane, is the peak intensity of BG-SHG at , is the 1/e2 half width of BG-SHG, and is the intensity of BS-SHG at the confocal plane. The power collected by the SCF is the integration of in the core area of the SCF,where is the radius of the SCF core, is the collection solid angle of the SCF, and is the solid angle of the collimator light cone. The first term in Eq. (3) represents the power of BG-SHG collected by the SCF, shown as in the following Eq. (4). The second term in Eq. (3) represents the power of BS-SHG collected by the SCF, shown as in the following Eq. (4):Similarly, the power collected by the MMF bundle can be expressed with only the BS-SHG term as follows:
where is the radius of the core of MMF, is the collection solid angle of MMF. Due to the uniform distribution of BS-SHG at the confocal plane, the BS-SHG collected by the SCF can be obtained from the BS-SHG collected by the MMF bundle:where the solid angle has been expressed as . is the numerical aperture of the SCF, and is the numerical aperture of the MMF. Therefore, the BG-SHG collected by the SCF can be extracted from the total power collected by the SCF by removing the BS-SHG term:here we define BG-SHG to BS-SHG ratio as BGSR at the SCF core area in the focal plane,
4.2 SHG imaging experimental results and discussion
We first imaged 40 μm mouse tail tendon fascicle sections to test the system’s SHG performance. The images collected by the MMF bundle, the SCF, and the 60 × objective in free space are shown in Figs. 7(a) and 7(e), Figs. 7(b) and 7(f), and Figs. 7(c) and 7(g), respectively. Figures 7(e)-7(h) are zoomed-in scans of the yellow dashed rectangular region in 7(a). Comparing the first and third column, we see that the images collected by the MMF bundle maintain most of the features of the images collected by the 60 × objective. To illustrate the differences in image patterns collected, color overlapped images are shown in Figs. 7(d) and 7(h) with images collected by the SCF in green and that collected by the MMF bundle in red. Similar to the analysis in the CARS imaging part, the MMF bundle collects mostly BS-SHG photons which represent general fibrillar structures with large coherence length as shown in Figs. 7(a) and 7(e). The SCF collects mainly BG-SHG photons at the center of confocal plane which represent fine tissue architectures with small coherence length as shown in Figs. 7(b) and 7(f). The different features between FG-SHG and BG-SHG are in accordance with the results reported in [11].
Fig. 7 SHG imaging of mouse tail tendon collected by (a) MMF bundle, (b) SCF, and (c) 60 × objective. (d) Color overlapped image with SCF collection in green and MMF bundle collection in red. (e-h) Zoomed-in scan corresponding to (a)-(d) in the yellow dashed rectangular region in (a). Large, fibrillar structures are revealed by the MMF bundle collection (a, e), while fine segmented architectures are revealed by the SCF collection (b, f). Scale bars are 20 μm.
We then used thick mouse tail tendon tissues (≥ 1 mm) which provide many BS-SHG photons at the confocal plane. Mouse tails were obtained from three C57Bl/6J male mice aged 12 weeks sacrificed for other experiments. Tissue was then transferred to a coverslip that was supported by a carrier slide with glass at the center removed to eliminate scattering of photons by the carrier slide below the tissue. The images collected by the SCF and the MMF bundle at 10 μm depth are shown in Figs. 8(a) and 8(b) respectively. The MMF bundle collection shows images of continuous fibrillar structures which are generated by BS-SHG photons. The BG-SHG image collected by the SCF, obtained with Eq. (7) and shown in Fig. 8(c), reveals fine segmented architectures. BG-SHG and BS-SHG images are overlapped in green and red respectively in Fig. 8(d), showing the distribution of scatterers generating BG-SHG among the general fibrillar structures generating BS-SHG. By applying Eq. (8) we get BGSR in SCF collection in Fig. 8(e) at each pixel. A zoomed-in image of BGSR at the yellow rectangular region in Fig. 8(e) is shown in Fig. 8(f). It can be seen from Figs. 8(e) and 8(f) that at regions where BG-SHG is strong, the BGSR collected by the SCF has high value. This BGSR can tell us the SCF’s performance in collecting BG-SHG and BS-SHG photons. Note that the BGSR shows high values at the bottom left region where both BG-SHG and BS-SHG are weak, because the weak BS-SHG is placed as the denominator for calculating BGSR. We calculate the average BGSR in the regions where strong BG-SHG is observed, shown in yellow rectangles in Fig. 8(d), and this value is represented in the form of average value (standard deviation) as 6324 (663). For those areas where BG-SHG is weak while BS-SHG is strong, shown in white rectangles in Fig. 8(d), this value is 441 (56). This result shows in the areas where strong BG-SHG is generated, BG-SHG can be 6000 times of BS-SHG. This can explain why the SCF collected Fig. 8(a) strongly resembles the BG-SHG image shown in Fig. 8(c), as the BS-SHG rarely enter the collection cone of the SCF. Therefore, BGSR can be used as a quantitative factor to differentiate these regions and monitor BG- and FG-photon generation.
Fig. 8 Backward SHG images of 1 mm thick mouse tail tendon tissue collected by (a) the SCF and (b) the MMF bundle. The MMF bundle collection shows image of BS-SHG. (c) BG-SHG image obtained by subtracting normalized BS-SHG intensity from SCF collected image. (d) Color overlapped image with BG-SHG in green and the MMF bundle collected BS-SHG in red. The ratio of BG-SHG and BS-SHG (BGSR) collected by the SCF at the yellow rectangular regions with strong BG-SHG and the white rectangular regions with weak BG-SHG and strong BS-SHG is calculated. (e) BGSR at each pixel of the image. The pixels with strong BG-SHG have high BGSR, showing that the SCF collects much more BG-SHG than BS-SHG and is thus suitable for revealing BG-SHG image pattern. (f) BGSR at the yellow region in (e). Imaging depth is 10 μm. Scale bars are 10 μm.
Figure 9 shows the SHG imaging results at 10 μm depth from another thick mouse tail tendon tissue. The BG-SHG image is shown in Fig. 9(a), and the BS-SHG image collected by the MMF bundle is shown in Fig. 9(b). Similar to Fig. 8, the SCF collected image strongly resembles Fig. 9(a) and is therefore not shown. The BG-SHG and BS-SHG images are shown in green and red respectively in Fig. 9(c). Figure 9(c) shows that the BG-SHG scatterers are distributed as a framework for the tendon, and the FG-SHG scatterers are located within this framework. This framework pattern of BG-SHG might be related to the linear and lateral assembly of fibril intermediates in collagen fibrillogenesis [43, 44]. At the assembly regions, the structures are more heterogeneous than the fibril intermediates and might generate strong BG-SHG. Figure 9(d) shows the conventional SHG microscopy imaging result collected by the free space 60 × objective in the backward direction. We also find that this framework pattern exists within the next 10 μm depth, as shown in Fig. 9(e) at 10 μm, Fig. 9(f) at 15 μm and Fig. 9(g) at 20 μm.
Fig. 9 (a) BG-SHG image of 1 mm thick mouse tail tendon tissue. (b) MMF bundle collected BS-SHG image. (c) Color overlapped image with BG-SHG in green and BS-SHG in red. The fine structures generating BG-SHG are distributed in a framework pattern. (d) Epi-SHG image collected by the 60 × objective. (e) Color overlapped image at 10 μm depth, (f) 15 μm depth, and (g) 20 μm depth. (h) The ratio of BG-SHG and BS-SHG (BGSR) collected by the SCF at Regions 1 to 6 in (e) at different depth. The yellow rectangular Regions 1, 2, and 3 are where strong BG-SHG is generated and the white rectangular Regions 4, 5 and 6 are where weak BG-SHG and strong BS-SHG are generated. BGSR shows large value difference for these two types of regions. Scale bars are 20 μm.
We also tested the change of BGSR with depth at regions with bright BG-SHG (marked 1, 2, 3) and regions without bright BG-SHG (marked 4, 5, 6), as depicted in Fig. 9(e). In Fig. 9(h), the BGSR for Regions 4, 5 and 6 increases from 560 to around 2500 as depth goes from 10 μm to 20 μm. This should be caused by decreased BS-SHG due to increasing imaging depth, as proved in [30], which contributes to the increase of BGSR. Region 1 has the strongest BG-SHG in Fig. 9(e) compared with Regions 2 and 3, and therefore has the largest BGSR. Then, it becomes very weak in Fig. 9(f), and as a result, its BGSR decreases. From Fig. 9(f) to Fig. 9(g), though BG-SHG decreases a little, the rapid decrease of BS-SHG at the same area leads to the increase of BGSR for Region 1. A similar trend is observed for Region 2, as the brightness of BG-SHG is not decreased appreciably in Fig. 9(f), it has a lower decrease rate from 10 μm to 15 μm than Region 1. The increase of BGSR for Region 3 is mainly due to the decrease of BS-SHG at that part of tissue. From Fig. 9(h), it can be observed that though BGSR changes at different depths for each region, significant difference still exists among different regions of the same depth.
Next, we investigated the images feature of BG-SHG at different depths within a tendon by performing a Z-section imaging with 2 μm step size and total depth of 20 μm from surface to bottom in Fig. 10. The green color shows images collected by the SCF that exhibit a pattern of BG-SHG as discussed previously, while the red color indicates images collected by the MMF bundle, which are generated by BS-SHG. The images collected by the SCF and the MMF bundle are detected simultaneously with two PMTs. The brightness of the red color image is darkened to better illustrate the features of BG-SHG expressed in green. It can be observed that in Figs. 10(a)-10(c), from the tendon surface to 4 μm depth, the fine structures of tendon generating BG-SHG show up as small points. In Figs. 10(d)-10(g), 6 μm to 12 μm below the tendon surface, the fine structures show BG-SHG as segmented fibrillar shapes oriented in the same direction with the general fibrillar structures generating BS-SHG. In Figs. 10(h)-10(k), as the laser focus goes to 14 μm and 20 μm and reaches the bottom of tendon, most of the BG-SHG turns into small spots with no obvious shapes or orientation, with only small regions showing intensity in Figs. 10(i)-10(k).
Fig. 10 (a) – (k) Section SHG imaging of a mouse tail tendon from surface to 20 μm depth. SCF collected images are shown in green to show image pattern of BG-SHG, and MMF bundle collected BS-SHG is shown in red. The BG-SHG shows different pattern features at the surface and inside of tendon. In (a)-(c), at the surface of the tendon, BG-SHG shows the pattern feature as weak small points. In (d)-(g), inside of the tendon, BG-SHG shows the feature as segmented fibrillar shapes oriented in the same direction with the general fibrillar structures generating BS-SHG. In (h)-(k), at the bottom of the tendon, BG-SHG displays the feature as weak small points. (l) Intensity change of BG-SHG and BS-SHG through the section imaging at the yellow polygon region in (g). The intensity is normalized relative to the maximum intensity in the scanning series. The peak intensity of BS-SHG appears near the surface of the tendon while the peak intensity of BG-SHG appears at depth inside of the tendon. Scale bars are 10 μm.
We further investigated the intensity of BG-SHG and BS-SHG through section scanning in the region marked yellow in Fig. 10(g). We ignored the tiny BS-SHG power collected by the SCF and took the SCF collected intensity as the intensity of BG-SHG in this comparison. We averaged the raw data of BG-SHG in this region at each depth to obtain average pixel intensity of BG-SHG and normalized this average intensity to the largest value in the whole series of images. Similarly, we acquired intensity of BS-SHG normalized relative to the maximum intensity. The results are shown in Fig. 10(l). The strongest BS-SHG appears at 2 μm depth which is close to the tendon surface. Below 2 μm, as the laser focus goes deeper into the tissue, due to the absorption and scattering of excitation laser, the blurring of focus [25] and decreased backscattering of BS-SHG [30], the intensity of BS-SHG is decreased. In contrast, the strongest BG-SHG appears at 6-8 μm depth inside of the tendon. This can be explained by the fact that the scatterers generating BG-SHG might be highly localized at this depth inside of the tendon, or they are distributed in a spatial frequency suitable for generating strong BG-SHG at this depth [13, 45]. The relatively faster decrease of intensity of BG-SHG compared to that of BS-SHG in 12-20 μm should be due to the pinhole effect of the SCF, and fewer BG-SHG scatterers at this depth.
This novel dual modality imaging system has an advantage for revealing BG-SHG and BS-SHG image patterns for thick tissues. Conventionally, BG-SHG images can only be obtained for tissues less than 10 μm thick, since 10 μm has much less backscattering of forward-SHG [11, 30], and for thick tissues, the increased backscattering of FG-SHG increases the BS-SHG so much that the BG-SHG is often overwhelmed by BS-SHG in backward detection. With our setup, as BG-SHG and BS-SHG are separated at the confocal plane by the SCF and the MMF surrounding it, and due to the Gaussian distribution of BG-SHG and uniform distribution of BS-SHG, the BG-SHG image can be extracted from the BS-SHG photons especially in thick tissue imaging so that the image pattern difference between BG-SHG and BS-SHG can be obtained.
This method has certain limitations however. First of all, we found that deeper than 20 μm, the SCF has limited ability to reveal BG-SHG images. The second limitation is that this method can only quantify the ratio of BG-SHG and BS-SHG collected by the SCF, but not that at the entire confocal plane of the system, and this quantification can only be applied to thick tissue imaging, because thin tissue cannot provide enough backscattering of BS-SHG to form uniform distribution across the common endface of this system. Lastly, our proof-of-concept approach uses a scanning stage to move the sample for imaging. In order to enable implementation for clinical endomicroscopy, this collection scheme must be miniaturized and coupled into a small compact probe with micro-electro-mechanical systems (MEMS) or piezoelectric tube scanning mechanism inside. Graded-index (GRIN) lens or microlens is also needed to replace the objective in our current setup. From the experience gained from this study, our group has recently reported a miniaturized endomicroscope probe design which opens up the possibility of incorporating this prototype for in vivo imaging applications [46].
5. Conclusions
In this paper, we have developed and tested a multimodal scheme using an integrated single central fiber and multimode fiber bundle to simultaneously collect and differentiate images formed by backward and forward generated photons in CARS and SHG. We have demonstrated that the SCF is suitable for collecting backward-generated photons with small coherence length and revealing fine structures in tissues, and the MMF bundle is suitable for collecting backscattered forward-generated photons with large coherence length and revealing general tissue structure. A framework structure feature of backward-generated SHG in regions of mouse tail tendons is observed. Different features of backward-generated SHG photons at the surface, inside and bottom of a tendon are also revealed and discussed. In the single central fiber collection, the ratio of backward-generated SHG photons and backscattered forward-generated SHG photons is quantified, and it shows a large difference at regions with or without strong backward-generated SHG, indicating its potential to be used as a quantitative factor to differentiate these regions and monitor photon generation status. As the generation of backward photons and forward photons strongly depends on the size and arrangement of tissue, the backward and forward generated photons provide information of tissue from different points of view, with backward-generated photons providing information of structures with axial length smaller than the excitation wavelength and forward-generated photons providing information of objects with comparable or larger size than excitation wavelength. Our image acquisition scheme has the potential to enhance the functionality and imaging capability of in vivo CARS/SHG endomicroscopy for biomedical applications.
Acknowledgments
This research is supported by DOD PC11860, DOD MR130311, TT & WF Chao Foundation, and John S Dunn Research Foundation (STCW). The authors would like to acknowledge the help of James Mancuso in proofreading the manuscript, the help of Zhen Zhao, Yuanxin Chen and Richard Federley in preparing the tissue, the discussion with Michael Thrall on CARS endomicroscopy application, the discussion with Xuping Li on tendon morphology and the discussion with other members of translational photonics lab, Zhengfan Liu, Zachary Satira, Xi Wang, and Zhiyong Wang in the experimentation.
References and links
1. P. D. Maker and R. W. Terhune, “Study of optical effects due to an induced polarization third order in the electric field strength,” Phys. Rev. 137(3A), A801–A818 (1965). [CrossRef]
2. C. J. R. Sheppard and R. Kompfner, “Resonant scanning optical microscope,” Appl. Opt. 17(18), 2879–2882 (1978). [CrossRef] [PubMed]
3. I. Freund and M. Deutsch, “Second-harmonic microscopy of biological tissue,” Opt. Lett. 11(2), 94–96 (1986). [CrossRef] [PubMed]
4. M. D. Duncan, J. Reintjes, and T. J. Manuccia, “Scanning coherent anti-Stokes Raman microscope,” Opt. Lett. 7(8), 350–352 (1982). [CrossRef] [PubMed]
5. E. O. Potma and X. S. Xie, “Detection of single lipid bilayers with coherent anti-Stokes Raman scattering (CARS) microscopy,” J. Raman Spectrosc. 34(9), 642–650 (2003). [CrossRef]
6. C. L. Evans, X. Xu, S. Kesari, X. S. Xie, S. T. C. Wong, and G. S. Young, “Chemically-selective imaging of brain structures with CARS microscopy,” Opt. Express 15(19), 12076–12087 (2007). [CrossRef] [PubMed]
7. L. Gao, A. A. Hammoudi, F. Li, M. J. Thrall, P. T. Cagle, Y. Chen, J. Yang, X. Xia, Y. Fan, Y. Massoud, Z. Wang, and S. T. C. Wong, “Differential diagnosis of lung carcinoma with three-dimensional quantitative molecular vibrational imaging,” J. Biomed. Opt. 17(6), 066017 (2012). [CrossRef] [PubMed]
8. Y. Yang, F. Li, L. Gao, Z. Wang, M. J. Thrall, S. S. Shen, K. K. Wong, and S. T. Wong, “Differential diagnosis of breast cancer using quantitative, label-free and molecular vibrational imaging,” Biomed. Opt. Express 2(8), 2160–2174 (2011). [CrossRef] [PubMed]
9. L. Gao, H. Zhou, M. J. Thrall, F. Li, Y. Yang, Z. Wang, P. Luo, K. K. Wong, G. S. Palapattu, and S. T. Wong, “Label-free high-resolution imaging of prostate glands and cavernous nerves using coherent anti-Stokes Raman scattering microscopy,” Biomed. Opt. Express 2(4), 915–926 (2011). [CrossRef] [PubMed]
10. 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]
11. F. Légaré, C. Pfeffer, and B. R. Olsen, “The Role of Backscattering in SHG Tissue Imaging,” Biophys. J. 93(4), 1312–1320 (2007). [CrossRef] [PubMed]
12. W. R. Zipfel, R. M. Williams, and W. W. Webb, “Nonlinear magic: multiphoton microscopy in the biosciences,” Nat. Biotechnol. 21(11), 1369–1377 (2003). [CrossRef] [PubMed]
13. R. M. Williams, W. R. Zipfel, and W. W. Webb, “Interpreting Second-Harmonic Generation Images of Collagen I Fibrils,” Biophys. J. 88(2), 1377–1386 (2005). [CrossRef] [PubMed]
14. P. J. Campagnola and L. M. Loew, “Second-harmonic imaging microscopy for visualizing biomolecular arrays in cells, tissues and organisms,” Nat. Biotechnol. 21(11), 1356–1360 (2003). [CrossRef] [PubMed]
15. M. Born and E. Wolf, Principles of Optics: Electromagnetic Theory of Propagation, Interference and Diffraction of Light (Cambridge University Press, 1999).
16. R. LaComb, O. Nadiarnykh, S. S. Townsend, and P. J. Campagnola, “Phase matching considerations in second harmonic generation from tissues: Effects on emission directionality, conversion efficiency and observed morphology,” Opt. Commun. 281(7), 1823–1832 (2008). [CrossRef] [PubMed]
17. C. P. Pfeffer, B. R. Olsen, and F. Légaré, “Second harmonic generation imaging of fascia within thick tissue block,” Opt. Express 15(12), 7296–7302 (2007). [CrossRef] [PubMed]
18. O. Nadiarnykh, R. B. Lacomb, P. J. Campagnola, and W. A. Mohler, “Coherent and incoherent SHG in fibrillar cellulose matrices,” Opt. Express 15(6), 3348–3360 (2007). [CrossRef] [PubMed]
19. C. P. Pfeffer, B. R. Olsen, F. Ganikhanov, and F. Légaré, “Multimodal nonlinear optical imaging of collagen arrays,” J. Struct. Biol. 164(1), 140–145 (2008). [CrossRef] [PubMed]
20. S.-W. Chu, S.-P. Tai, T.-M. Liu, C.-K. Sun, and C.-H. Lin, “Selective imaging in second-harmonic-generation microscopy with anisotropic radiation,” J. Biomed. Opt. 14(1), 010504 (2009).
21. M. Rivard, M. Laliberté, A. Bertrand-Grenier, C. Harnagea, C. P. Pfeffer, M. Vallières, Y. St-Pierre, A. Pignolet, M. A. El Khakani, and F. Légaré, “The structural origin of second harmonic generation in fascia,” Biomed. Opt. Express 2(1), 26–36 (2011). [CrossRef] [PubMed]
22. G. Tjin, P. Xu, S. H. Kable, E. P. W. Kable, and J. K. Burgess, “Quantification of collagen I in airway tissues using second harmonic generation,” J. Biomed. Opt. 19(3), 036005 (2014). [CrossRef] [PubMed]
23. M. Han, G. Giese, and J. Bille, “Second harmonic generation imaging of collagen fibrils in cornea and sclera,” Opt. Express 13(15), 5791–5797 (2005). [CrossRef] [PubMed]
24. B. Richards and E. Wolf, “Electromagnetic diffraction in optical systems. II. Structure of the image field in an aplanatic system,” Proc. R. Soc. Lond. A Math. Phys. Sci. 253(1274), 358–379 (1959). [CrossRef]
25. J.-X. Cheng, A. Volkmer, and X. S. Xie, “Theoretical and experimental characterization of coherent anti-Stokes Raman scattering microscopy,” J. Opt. Soc. Am. B 19(6), 1363–1375 (2002). [CrossRef]
26. L. Moreaux, O. Sandre, and J. Mertz, “Membrane imaging by second-harmonic generation microscopy,” J. Opt. Soc. Am. B 17(10), 1685–1694 (2000). [CrossRef]
27. A. Volkmer, J.-X. Cheng, and X. S. Xie, “Vibrational imaging with high sensitivity via epidetected coherent anti-Stokes Raman scattering microscopy,” Phys. Rev. Lett. 87(2), 023901 (2001). [CrossRef] [PubMed]
28. J.-X. Cheng and X. S. Xie, “Coherent Anti-Stokes Raman Scattering Microscopy: Instrumentation, Theory, and Applications,” J. Phys. Chem. B 108(3), 827–840 (2004). [CrossRef]
29. I. W. Schie, T. Weeks, G. P. McNerney, S. Fore, J. K. Sampson, S. Wachsmann-Hogiu, J. C. Rutledge, and T. Huser, “Simultaneous forward and epi-CARS microscopy with a single detector by time-correlated single photon counting,” Opt. Express 16(3), 2168–2175 (2008). [CrossRef] [PubMed]
30. M. Shen, Y. Tian, S. P. Chong, J. Zhao, H. Zeng, and S. Tang, “Quantifying the backscattering of second harmonic generation in tissues with confocal multiphoton microscopy,” J. Biomed. Opt. 18(11), 115003 (2013). [CrossRef] [PubMed]
31. C. L. Evans, E. O. Potma, M. Puoris’haag, D. Côté, C. P. Lin, and X. S. Xie, “Chemical imaging of tissue in vivo with video-rate coherent anti-Stokes Raman scattering microscopy,” Proc. Natl. Acad. Sci. U.S.A. 102(46), 16807–16812 (2005). [CrossRef] [PubMed]
32. S.-P. Tai, T.-H. Tsai, W.-J. Lee, D.-B. Shieh, Y.-H. Liao, H.-Y. Huang, K. Zhang, H.-L. Liu, and C.-K. Sun, “Optical biopsy of fixed human skin with backward-collected optical harmonics signals,” Opt. Express 13(20), 8231–8242 (2005). [CrossRef] [PubMed]
33. W. Lo, H.-Y. Tan, M.-G. Lin, C.-M. Hsueh, W.-L. Chen, S.-J. Lin, S.-H. Jee, and C.-Y. Dong, “Forward- and backward-second harmonic generation imaging of corneal and scleral collagen,” Proc. SPIE 6860, 68600F (2008). [CrossRef]
34. N. Morishige, N. Yamada, S. Teranishi, T. Chikama, T. Nishida, and A. Takahara, “Detection of subepithelial fibrosis associated with corneal stromal edema by second harmonic generation imaging microscopy,” Invest. Ophthalmol. Vis. Sci. 50(7), 3145–3150 (2009). [CrossRef] [PubMed]
35. F. Légaré, C. L. Evans, F. Ganikhanov, and X. S. Xie, “Towards CARS Endoscopy,” Opt. Express 14(10), 4427–4432 (2006). [CrossRef] [PubMed]
36. M. D. I. Veilleux, P. Coté, S. Verreault, M. Fortin, P. Paradis, S. Leclair, R. S. Da Costa, B. C. Wilson, and E. J. Seibel, “Design and modeling of a prototype fiber scanning CARS endoscope,” Proc. SPIE 7558, 75580D (2010). [CrossRef]
37. Z. Liu, Z. Wang, X. Wang, X. Xu, X. Chen, J. Cheng, X. Li, S. Chen, J. Xin, and S. T. C. Wong, “Fiber bundle based probe with polarization for coherent anti-Stokes Raman scattering microendoscopy imaging,” Proc. SPIE 8588, 85880F (2013). [CrossRef]
38. X. Han and E. Brown, “Measurement of the ratio of forward-propagating to back-propagating second harmonic signal using a single objective,” Opt. Express 18(10), 10538–10550 (2010). [CrossRef] [PubMed]
39. Z. Wang, Y. Liu, L. Gao, Y. Chen, P. Luo, K. K. Wong, and S. T. C. Wong, “Use of multimode optical fibers for fiber-based coherent anti-Stokes Raman scattering microendoscopy imaging,” Opt. Lett. 36(15), 2967–2969 (2011). [CrossRef] [PubMed]
40. Z. Wang, Y. Yang, P. Luo, L. Gao, K. K. Wong, and S. T. Wong, “Delivery of picosecond lasers in multimode fibers for coherent anti-Stokes Raman scattering imaging,” Opt. Express 18(12), 13017–13028 (2010). [CrossRef] [PubMed]
41. Z. Wang, L. Gao, P. Luo, Y. Yang, A. A. Hammoudi, K. K. Wong, and S. T. C. Wong, “Coherent anti-Stokes Raman scattering microscopy imaging with suppression of four-wave mixing in optical fibers,” Opt. Express 19(9), 7960–7970 (2011). [CrossRef] [PubMed]
42. http://digital.ni.com/public.nsf/allkb/D64CD277A6B739A186256A73007E7BCC
43. D. E. Birk, M. V. Nurminskaya, and E. I. Zycband, “Collagen fibrillogenesis in situ: fibril segments undergo post-depositional modifications resulting in linear and lateral growth during matrix development,” Dev. Dyn. 202(3), 229–243 (1995). [CrossRef] [PubMed]
44. G. Zhang, B. B. Young, Y. Ezura, M. Favata, L. J. Soslowsky, S. Chakravarti, and D. E. Birk, “Development of tendon structure and function: regulation of collagen fibrillogenesis,” J. Musculoskelet. Neuronal Interact. 5(1), 5–21 (2005). [PubMed]
45. J. Mertz and L. Moreaux, “Second-harmonic generation by focused excitation of inhomogeneously distributed scatterers,” Opt. Commun. 196(1-6), 325–330 (2001). [CrossRef]
46. X. Chen, X. Xu, D. T. McCormick, K. Wong, and S. T. Wong, “Multimodal nonlinear endo-microscopy probe design for high resolution, label-free intraoperative imaging,” Biomed. Opt. Express 6(7), 2283–2293 (2015). [CrossRef] [PubMed]
