1. Introduction

The combination of two-photon microscopy (TPM) and optical coherence tomography (OCT) is useful for in-vivo tissue imaging. TPM is 3D fluorescence microscopy based on two-photon excitation. TPM has the advantages of high imaging depths down to a few hundred microns in typical tissues and minimal photo-damage [1]. TPM has been used for in-vivo cellular study within tissues in the fields of neuro-biology, immunology, and cancer biology. Although TPM has enhanced our knowledge on how cells behave within intact tissue environments, it still has limitations inherent to fluorescence microscopy in that it can only visualize fluorescing features with the field of view (FOV) of a few hundred microns on one side. This limitation of TPM can be compensated by combining it with OCT, which provides complementary information about tissues. OCT is a 3D imaging technique based on light back-reflection from within tissues. OCT provides information about tissue micro-structure at sub-tenth micron resolutions in larger tissue regions than TPM [2]. OCT has been developed mainly as a non-invasive diagnostic technique for clinical studies in the fields of ophthalmology, dermatology, and gastroenterology etc. Identifying structural information with OCT can be useful in gaining better understanding of tissue environments, in which cells reside, for in-vivo tissue study. Optical coherence microscopy (OCM), which is a high resolution version of OCT, can visualize cells with a few micron resolutions [3]. The combination of TPM and OCT provides more information regarding in-vivo tissues in the molecular, cellular, and structural levels.

Several combined methods have been reported. These methods are usually combinations of TPM and OCM by using single light sources, such as Ti-Sapphire lasers [4–6]. Structural and biochemical contrasts of live drosophila embryos [4], fibroblast cells cultured on a structured substrate [5], and fibroblast cells on an organotypic RAFT tissue model [6] have been demonstrated using various combined methods. Tri-modal imaging of OCM, two-photon fluorescence, and coherent detection of second harmonic generation in the ex-vivo human skin has also been demonstrated [7]. Although these methods showed images with augmented contrasts, these images had the same FOV and resolution for both TPM and OCM by sharing the same light sources. Recently, broad spectral band generation using a nonlinear photonic crystal fiber from a Ti-Sapphire laser has been demonstrated for simultaneous OCM and TPM imaging [8]. This method is useful for high-resolution cellular imaging within tissues. Visualization of microscopic collagen distribution by TPM and macro-scale tissue structure by OCT has also been demonstrated in a wound healing study with a skin-equivalent tissue model [9].

We developed a new combined method, which uses separate light sources and scanners for optimal imaging conditions of individual modalities. TPM used a Ti-Sapphire laser for molecular and cellular imaging based on fluorescence at sub-cellular resolutions. OCT used a wavelength-swept source centered at 1300 nm and a separate scanner to get information of tissue micro-structures in the deeper and wider tissue regions than TPM at sub-tenth micron resolutions. This method was designed to do simultaneous imaging by combining light from both sources and using the same objective lens. This combined method was applied to image small intestine and ear tissues of mouse models ex-vivo. System development, performance test, and applications to tissue imaging are described in detail in the next sections.

2. Material and methods

2.1 System configuration

A schematic of the system is shown in Fig. 1 . TPM used a Ti-Sapphire laser (Chameleon Ultra II, Coherent), which had 140 fs pulse width and 80 MHz pulse repetition rate, as an excitation light source. Output from the source first passed through a combination of a half wave plate and a polarizer for power control. The excitation beam went to a combination of a resonant scanner (GSI Lumonics, 8 KHz resonant frequency) and a galvanometric scanner (6215H, Cambridge Technology), which can do high-speed raster scanning. After these scanners, the excitation beam was expanded by a pair of scan and tube lenses, which have focal lengths of 50 mm and 250 mm respectively. The excitation beam was transmitted through a dichroic (DM1, 1025DCSP, Chroma), and coupled into an upright microscope (BX51, Olympus). DM1 was used to combine TPM excitation beam with OCT illumination beam. In the microscope, the excitation beam was reflected on a dichroic mirror in the upper position (DM2, 680DCSP, Chroma) of the microscope, transmitted through another dichroic mirror in the lower position (DM3, 680DCLP, Chroma), and slightly under-filled the back aperture of a 20x objective lens (XLUMPLFLN, NA 1, Olympus). The objective lens focused the excitation beam into a sample. The excitation focus scanned in the x-y plane of the sample by the scanners, and in the z axis by an objective translator (P-725.4CL, PI). Emission light from the sample was collected back by the objective lens and was reflected on DM3 toward photomultiplier tubes (PMTs, R5929, Hamamatsu). Signals from the PMTs were collected by a frame grabber (Alta, Bitflow) and images were displayed in real time.

 

Fig. 1 System configuration of combined TPM and OCT. SM: scanning mirror, PDBD: polarization-diverse balanced detection, PMT: photomultiplier tube, HWP: half-wave plate, PL: polarizer, SL: scan lens, TL: tube lens, OL: objective lens, SP: sample, TS: translation stage, DM: dichroic mirror. TPM excitation and emission beams were depicted in red and green respectively, and OCT beam was depicted in gray.

Download Full Size | PDF

OCT used a custom-built wavelength swept source based on a polygonal wavelength filter and a semiconductor optical amplifier (BOA-5785, Covega) as the gain medium [10,11]. This wavelength swept source had the center wavelength of 1312 nm, bandwidth of 107 nm, spectral resolution of 0.17 nm, output power of 45 mW, and sweeping speed of 47.6 K depth-scans per second. Output from the source was coupled to an interferometer setup, in which light was split to the reference and sample arms by a fiber splitter in the ratio of 10/90 respectively. In the sample arm, light delivered via the fiber was collimated (f = 6.2 mm, Oz optics), reflected on an x-y galvanometric scanner (6215H, Cambridge Technology), transmitted through a scan lens (f = 150 mm), reflected on DM1, and then combined with TPM excitation beam. OCT beam was expanded to 1.6 mm in diameter by the combination of SL2 and TL1 on the back aperture of the objective lens, and was focused into the sample with effective NA of 0.1. Reflected light from the sample was collected back by the objective and was traced in the reverse direction of the OCT illumination beam to the sample arm fiber. Coupled light in the fiber went to the detection arm of the interferometer via a circulator. In the detection arm, the reflected light from the sample was combined with the one from the reference arm, which had a matched path length for interference generation. Interference signal was collected by a polarization-diverse balanced detection setup and digitized by a data acquisition board (PDA14, Signatec). Image was displayed in real time with standard OCT signal processing.

TPM had FOV of 300 μm x 300 μm with the 20x objective lens. The resolution was characterized by imaging 0.2 μm diameter fluorescent microspheres in 3D and by analyzing full width at half maximum (FWHM) intensity in both lateral and axial directions. The measured FWHMs were 0.47 μm and 2.5 μm on average in the lateral and axial directions respectively at the excitation wavelength of 800 nm. Imaging speed was 40 frames per second in case of 400 x 400 pixel images, and actual imaging speed varied depending on the number of frames to be integrated. TPM obtained 3D images by acquiring x-y plane images sequentially, while moving its focal plane stepwise in the z axis. OCT obtained 3D images by acquiring 250 x 250 depth-scans in the x-y plane. Size of the 3D image was 800 μm x 800 μm in the x-y plane and 2.5 mm in the z axis composed of 312 pixels. OCT resolution was characterized by imaging 2 μm diameter fluorescent microspheres and a mirror, and by analyzing FWHM. OCT had resolutions of 5 μm and 8 μm on average in the lateral and axial directions in the air respectively, and depth of focus (DOF) of approximately 150 μm by using 0.1 NA. Only a small fraction out of the imaging depth range was in focus, because there was trade-off between lateral resolution and DOF. In order to image larger depth ranges than 150 μm in focus, multiple images were acquired by moving the objective in the axial direction by 150 μm step size, and then focused regions of these images were combined. Later on, NA of OCT beam was decreased to 0.06 by changing focal length of SL2 from 150 to 250 mm. DOF and lateral resolution were adjusted to 400 μm and 8 μm respectively. Acquisition time was approximately 1.25 seconds per volume. Sensitivity of 97 dB was obtained with the incident power of 11.4 mW on the sample. This relatively low sensitivity was because the objective lens had a low transmission efficiency of 45% around 1300 nm.

2.2 Tissue sample preparation

Two types of tissue samples were prepared for imaging. One was a small intestine sample from a CX3CR1gfp mouse, in which immune cells with CX3CR1 receptors express green fluorescent protein (GFP) [12]. The other was an ear tissue of a C57BL/6-Tg(CAG-EGFP)1Osb/J mouse, which expresses GFP in all its body except erythrocytes and hair. Both mouse models were obtained from the Jackson Laboratory, and bred at the animal facility of POSTECH Biotech Center under specific pathogen-free conditions.

For preparation of small intestine samples, CX3CR1gfp mice were euthanized and the small intestines were taken out. After removing surrounding fat tissues carefully, the small intestines were divided into proximal, middle and distal parts, and the middle parts were used for imaging. Inner surfaces of the tubular intestine samples were exposed and flatten carefully on coverslips, glued with Vetbond epoxy (3M Animal Care Products), washed gently with phosphate buffered saline (PBS), and then fixed with 4% paraformaldehyde. In order to visualize the morphology of intestine villi with TPM, these samples were labeled with 2 μM 5-(and-6)-(((4-chloromethyl)benzoyl)amino) tetramethylrho- damine (CMTMR, Invitrogen) for 30 min at 37 °C. Mouse ear samples were obtained by cutting a small piece of ear tissue from the GFP mice.

3. Results

3.1 Microsphere imaging

The combined system was tested first by imaging simple fluorescent microsphere samples, which were made by immobilizing yellow-green fluorescent microspheres in 3D with 2% agarose gel. These samples were prepared by either mixing two different size microspheres (2 μm and 10 μm in diameter) or using single size ones (6 μm in diameter). These samples were imaged and results are shown in Fig. 2 . Figure 2(a) is an overlay of TPM and OCT images, which are color-coded as blue and red colors respectively. This overlay image is a 2D x-y plane image, generated by projecting maximum intensity of the 3D images which had 150 μm in-focus depth range. Fluorescent microspheres appear in both images, because they scatter light as well as fluoresce. OCT and TPM images have different FOVs as outlined by inner and outer square lines. Figure 2(b) and 2(c) show TPM and OCT images respectively in the FOV of TPM. Individual microspheres are well co-registered in both images overall, though some spheres appear only in OCT image. These microspheres might not have fluorescent coating, or the agarose gel might not be completely dissolved. In OCT image (Fig. 2(c)), 2 μm microspheres appear blurry due to the resolution limit of OCT. Figure 2(d) is a 3D reconstructed image of 6 μm fluorescent microspheres in the FOV of TPM, and this image shows good 3D co-registration. These microsphere images were used as calibration data to co-register TPM and OCT images.

 

Fig. 2 2D TPM and OCT images of mixed fluorescent microspheres of 2 μm and 10 μm in diameter (a-c) and a 3D image of fluorescent microspheres of 6 μm in diameter (d). (a) Combined image, (b) TPM image, and (c) OCT image in the FOV of TPM. The scale bar is 100 μm.

Download Full Size | PDF

3.2 Tissue imaging

The combined system was applied to image the small intestine of CX3CR1gfp mice ex-vivo. For OCT, a single volumetric image was acquired with the focal plane of OCT beam positioned at half thickness down from top of the villi. Since the tissue was approximately 600 μm thick, some sections above and below the DOF were slightly out of focus. For TPM, a 3D image was acquired by moving the imaging plane down to 250 μm deep from the surface. 2 channel images were acquired to distinguish the CMTMR-labeled surface and GFP expressing cells. Excitation wavelength was 840 nm, and excitation power of TPM was approximately 45 mW. Imaging speed was 2 frames per second by integrating 20 frames. These small intestine samples were imaged from the inner surface and results are shown in Fig. 3 .

 

Fig. 3 TPM and OCT images of the ex-vivo small intestine of a mouse model. (a) 3D reconstruction of TPM and OCT images, (b) histology, (c) x-y plane image of GFP TPM, (d) OCT, (e) CMTMR TPM, and (f) Overlay of GFP TPM, OCT, and CMTMR TPM image at 100 micron deep from the surface. The scale bar is 100 μm.

Download Full Size | PDF

Figure 3(a) is an overlay of 3D reconstructed TPM and OCT images. OCT image shows distinctive morphology of intestine villi in a large FOV, and TPM image shows CMTMR-labeled surface in a small FOV. The shape of intestine villi shown in Fig. 3(a) is consistent with the histology image in Fig. 3(b). Figures 3(c) and 3(d) show x-y plane images of TPM and OCT respectively, at 100 micron deep from top of the villi. TPM image shows GFP expressing immune cells inside the villi, and OCT image shows structures of villi. These villi appear to have two layers in OCT image: the outer layer shows stronger scattering than the inner layer. Morphology information about the OCT image was verified by comparing with CMTMR labeled TPM image in Fig. 3(e). Both OCT- and CMTMR-labeled TPM images show good co-registration. Figure 3(f) shows an overlay of OCT, GFP TPM, and CMTMR TPM images in the x-y plane. As expected, these images are well co-registered. This overlaid image shows that GFP expressing immune cells are located in the inner layer of the villi.

The combined system was applied to image the mouse ear tissue ex-vivo, and results are shown in Fig. 4 . For OCT, a single volumetric image was acquired with the 0.06 NA and the whole thickness of the ear was in focus. For TPM image, 3D images were acquired by moving the imaging plane down to 150 μm deep from the surface. Excitation wavelength was 930 nm, and excitation power was 37.5 mW. Imaging speed was 2 frames per second by integrating 20 frames, and TPM images were 220 μm on one side. Figure 4(a) shows overlaid 3D reconstructed images of TPM and OCT along with a cross-sectional OCT image. OCT cross-sectional image shows microstructures in the whole thickness: the epithelium, dermis, and cartilage layers are visible. Figure 4(b)-4(j) show x-y plane images of TPM and OCT in TPM FOV at various depth locations. TPM images at different layers show various molecular cellular structures: cells in the epithelium (Fig. 4(b)), fibrous structures and cell clusters at the base of hair follicles in the dermis (Fig. 4(e)), and honeycomb structures in the cartilage (Fig. 4(h)). These features shown by TPM are not clearly visible in the corresponding OCT images partly due to resolution limitation and different contrast. However, OCT shows tissue structures nicely in the cross-sectional image. Overall, combined TPM and OCT provides molecular, cellular information together with tissue microstructures.

 

Fig. 4 TPM and OCT images of the ex-vivo ear tissue of a mouse model. (a) 3D reconstructed TPM and OCT images and an OCT cross-sectional image, (b-d) x-y plane images of GFP TPM, OCT, and overlay respectively in the epithelium, (e-g) x-y plane images in the dermis, (h-j) x-y plane images in the cartilage. The scale bar is 100 um.

Download Full Size | PDF

4. Discussion

The new combined method was implemented successfully, and its capabilities of providing more tissue information were demonstrated in ex-vivo tissue images. One significant difference compared with previous combined methods is that this method utilizes full functionalities of OCT. Therefore, OCT provides structural, morphological information of tissues in larger and deeper tissue regions than TPM. This structural information can be overlaid with molecular and cellular information from TPM in this method. This combined information will be useful for better understanding of in vivo tissue environments, because it covers a wide range from molecular, cellular up to tissue levels. Spatial co-registration of the two imaging modalities was demonstrated in the overlaid microsphere images in Fig. 2. However, we experienced spatial registration errors in the range of several microns maximum partly due to chromatic aberration and tissue property variation. Better co-registration methods need to be developed in the future.

The current combined system was not optimized. DOF of OCT was 150 μm originally with 0.1 NA and it was adjusted to 400 μm later by reducing NA to 0.6. Further improvement can be achieved by using an axicon lens [13] or by applying an image processing algorithm [14]. TPM can have higher imaging depths by using longer excitation wavelengths or by adapting adaptive optics [15]. Current OCT provided information of tissue structure only. Additional information, such as tissue vasculature including flow and tissue birefringence, can be obtained by adapting Doppler OCT and polarization-sensitive OCT.