1.

Introduction

Optical-resolution photoacoustic microscopy (OR-PAM) is capable of sensing endogenous optical absorption in biological bodies with fine lateral resolution provided by optical focusing. Pioneered by Maslov et al.,¹ OR-PAM imaging has been successfully applied to both structural and functional imaging.^2–9 Since both oxy- and deoxyhemoglobin are the dominant absorbing components in blood, the distribution of optical absorption, detected by the OR-PAM system, can be used to recover oxygen saturation (${\mathrm{SO}}_2$). For example, Hu et al.¹⁰ used OR-PAM to image ${\mathrm{SO}}_2$ with capillary-level resolution in mouse brain with dual-wavelength measurements. Hu et al.¹¹ also imaged the healing process of laser-induced microvascular lesions in a small animal model in vivo in terms of morphological and ${\mathrm{SO}}_2$ mapping. These studies demonstrated the capability of OR-PAM as a potential powerful tool in microcirculatory physiology and pathophysiology.

Recently our group reported the first label-free, fiber-based OR-PAM (F-OR-PAM) imaging systems.^12,13 The system retains many of the powerful properties of our previously proposed table-top system.¹⁴ It takes advantage of a flexible image guide and has significant potential to serve as a new micro-endscopic imaging technique for clinical use. Visualization of microvasculature in living mouse ears was demonstrated with the system.

Fluorescence contrast high-resolution fiber-optic micro-endoscopy (HRME) was introduced by Muldoon et al.¹⁵ This technique, which is inspired by the goal of virtual histopathology, utilizes a microscopic setup to image fluorescently-labelled cellular structures with lateral resolution on the order of microns. Fluorescent signals excited by a wide-field light source are magnified and sensed by a CCD camera. Rosbach et al.¹⁶ justified the feasibility of the imaging modality in evaluating lymph nodes from breast cancer patients with morphological variation. Muldoon et al.¹⁷ verified the validity of the technique in assisting diagnosis of oral lesions by observing and analyzing morphology of epithelia in human oral cavities.

In the present paper, we propose a combined system based on the HRME technique and our previous F-OR-PAM system, with the purpose of engineering a hybrid imaging platform to visualize capillary vasculatures along with cellular context.

2.

Methods

The dual modality imaging platform consists of two modules (Fig. 1). The F-OR-PAM system employs a 532-nm diode-pumped Ytterbium-doped fiber laser system as the optical source (GLP-10, IPG Photonics Corporation). Pulse repetition rate (PRR) of the laser system is tunable within the range from 20 k to 600 kHz. Pulse duration is $\sim 1\mathrm{ns}$ and the pulse energy can reach up to 20 μJ. A pair of 2D galvanometer scanning mirrors (6230H, Cambridge Technology, Inc.) is utilized to accomplish optical scanning. The scanners are driven by two analog sinusoidal waves ($x$- and $y$-direction) from a function generator. The scanned beam is then focused by an objective lens ($f=18\mathrm{mm}$, K16033703, Mitutoyo Co.) and directed into an image-guide fiber (FIGH-30-850, Myriad Fiber Imaging Tech., Inc.). Photoacoustic signals are sensed by a 3.5-MHz transducer (19-mm focus, 6-mm active element, f#=3.17, CD International Technology, Inc.), amplified, and pre-filtered by a pulser-receiver (5900 PR, Olympus NDT, Inc.) and then digitized by a 12-bit, 8-channel, high-speed digitizer (CS8289, Gage Applied Technologies, Inc.) along with the higher-frequency ($x$-axis) scanning feedback signals. For the fluorescent sub-system, a 447.5-nm-centered diode laser is utilized as the excitation light source. A standard fluorescent filter unit is employed in the system. An exciter band-pass filter with a center wavelength (Thorlabs, Inc.) of 445 nm is utilized. A 475-nm cut-off dichroic mirror (Chroma Technology Corp.) is used to reflect excitation light while transmitting excited fluorescence signals back. Excited fluorescence signals travel back through the image guide, reflected by the beam splitter positioned above the objective lens, and then follow the light path to the scientific-grade, high-resolution Electron-Multiplied CCD (EM-CCD, Andor iXon 885, Andor Technology). To rule out noise in the fluorescence signals, we use a 500-nm long-pass barrier filter (Thorlabs, Inc.) between dichroic mirror and tube lens. Pixel size of the CCD is $8\mathrm{\mu m}\times 8\mathrm{\mu m}$ and there are $1002\times 1004\mathrm{pixels}$ in total. The objective lens and the tube lens function together as a microscopic system. We use a plano-convex lens with a focal length of 150 mm (Thorlabs, Inc.) as the tube lens and therefore the magnification factor of the system is $\sim 8.3$. Proflavine, an FDA-approved drug for human medical use, is utilized as the contrast agent. In soft tissue, proflavine stains nuclei. This is important, since nuclei are often enlarged in cancer cells. Since the 532-nm laser is away from the absorption spectrum of Proflavine, no fluorescence signals will be generated by our OR-PAM optical source.

Fig. 1

Experimental setup for the combined photoacoustic and fluorescence micro-endoscopy imaging system (RM: reflective mirror; GS: glass slide; PD: photodiode; $\mathrm{SM}1/2$: scanning mirrors 1 and 2; FG: function generator; DX/Y: galvanometer scanning mirror divers; BS: beam splitter; OL: objective lens; UST: ultrasound transducer; P-R: pulser-receiver; TL: tube lens; EM: emission filter; DM: dichroic mirror; EX: excitation filter; DL: diode laser; DAQ: data acquisition card; HSD: high-speed digitizer; FL: fiber laser). The system is composed of an OR-PAM subsystem and a fluorescence microscopy system. A 2D galvanometer scanning mirror pair is used for raster optical scanning of the focused laser pulse. In the fluorescence imaging subsystem, excitation light is selected with an excitation filter and a dichroic mirror and directed by a beam splitter to share the same light path as the laser beam through the objective lens and the image guide to the target. Fluorescence energy is then reflected back through the dichroic mirror and then focused by a tube lens to the high-resolution CCD camera.

The image guide we used, which consists of 30,000 single-mode fibers in a bundle, has a diameter of $\sim 800\mathrm{\mu m}$. Considering the magnification factor of the microscope system, each single fiber covers 4.7 pixels of the EM-CCD on average, which is above the Nyquist limit.

The software package for data acquisition and data transfer is written in C/C++, and data processing and analysis is conducted with Matlab (Mathworks, Inc.).

3.

Results

3.1.

System Characterization

Figure 2 illustrates the system characterization of the imaging system. Figure 2(a) shows the entire footprint of the image guide we use, which has a diameter of $\sim 800\mathrm{\mu m}$. When the fluorescent imaging module is well focused, single elements of the image guide can be visualized, as is shown in Fig. 2(b). This demonstrates that the module is capable of imaging features at a cellular level. To examine the lateral resolution of the fluorescent module, we imaged a 1951 United Stated Air Force resolution target (USAF 1951). Features with a dimension of 4.38 µm can be resolved, which suggests that the imaging module has a lateral resolution that is on the order of the F-OR-PAM sub-system. For properties of our F-OR-PAM module, readers can refer to Ref. 12 for details. Figure 2(d) and 2(e) are images of a carbon fiber network obtained with both the two modules with fluorescent dye applied to the background. Contents in the two images are consistent. While carbon fibers appear as dark features in the fluorescent images, the F-OR-PAM image relies on photoacoustic signals from the fiber structures. The dual-modality imaging platform is capable of simultaneously providing two kinds of information that are supplementary to each other.

Fig. 2

System characterization. Panel (a) is the whole footprint of the image guide, with a diameter of $\sim 800\mathrm{\mu m}$ Panel (b) is a zoomed-in version of the region in the dashed rectangle in (a). When the imaging system is well focused, single elements of the image guide can be visualized. Panel (c) is the image of the USAF 1951 resolution target taken with the fluorescent imaging system. Bars with a width of around 4.38 μm are resolved. Panels (d) and (e) are the same carbon fiber network imaged with the fluorescent imaging module and the F-OR-PAM module sequentially. Consistent contents are visualized in the two images.

3.2.

In Vivo Imaging

In vivo studies were conducted to demonstrate imaging capability of our system. An ear of a 7-week-old SCID hairless outbred mouse was imaged for in vivo studies. Experimental procedures followed the laboratory animal protocol approved by the University of Alberta Animal Use and Care Committee. An animal anesthesia system was utilized while image data were collected. Before the imaging experiment, we topically applied proflavine-saline solution with a concentration of 0.01% (w/v) to the ear skin surface. To avoid contamination of fluorescence image by our fiber laser, the fluorescent and F-OR-PAM images were taken sequentially in a few minutes after the dye was applied. But acquisition of the two images can be interlaced. The measured power of the excitation 455-nm light at distal end of the image guide is around 0.45 mW. The 532-nm laser works with PRR of 160 kHz. The average power measured at the distal end when conducting F-OR-PAM imaging is around 35 mW. Results of the experiment are shown in Fig. 3. The scanning frequency of the galvanometer mirror system is 400 Hz and 1 Hz for the $x$- and $y$-axis, respectively. With the laser repetition rate of 160 kHz, the frame rate of the F-OR-PAM imaging sub-system is two frames per second, resulting in images with 160,000 points. Figure 3(a) and 3(b) are the F-OR-PAM maximum amplitude projection (MAP) and fluorescent images of the mouse ear obtained at the same location, respectively. A two dimensional Hessian-based Frangi vesselness filter was used for data processing.¹⁸ Figure 3(c) is the coregistered image.

Fig. 3

Results of in vivo experiments. Panel (a) is the fluorescence image of the mouse ear taken at the same location where the F-OR-PAM image was obtained; panel (b) is the F-OR-PAM image. (c) Co-registered image of (a) and (b). (d) Fluorescence image of normal human oral mucosa with topical application of Proflavine.

Figure 3(d) is a fluorescence image of normal human oral mucosa taken with our system. After topical application of the fluorescent dye, the distal end of the image guide was placed in contact with normal human oral mucosa to acquire fluorescent images. Staining of nuclei and cell membrane can be visualized clearly in the image.

4.

Discussion

We demonstrated a hybrid imaging system for both F-OR-PAM-based vasculature and fluorescently labeled cellular structure imaging. We further provide cellular information on the micron scale based on the OR-PAM system reported by our group.^12,13 We believe that the concept shown in the present paper may lead to a more powerful tool for biomedical imaging.

A potential application of our system is the study and diagnosis of early-stage cancer. Whereas F-OR-PAM can provide images of vasculature (and potentially hemoglobin oxygenation) in tissue, cellular information may provide morphological context, which is crucial for researchers and clinicians. For example, morphological anomalies in tumor tissue and variations in the microvasculatures due to cancer angiogenesis can confirm each other. This enables the technique to serve as a form of noninvasive virtual biopsy for identification of superficial tumor tissues in human cavities in diagnosis by clinicians. Estimate of ${\mathrm{SO}}_2$ at capillary level, which is not possible with pure fluorescence imaging, can also be accomplished with dual-wavelength OR-PAM imaging.¹⁰ This could help reveal the metabolic status of the cancer tissue, which is proven to be of importance for both fundamental studies and clinical practices. Another example is vasa vasorum imaging. Angiogenesis in this vascular layer of the lumen of atherosclerotic vessels has been correlated with plaque vulnerability. Visualizing the vessels could help provide one determinant of plaque vulnerability while fluorescent labeling could target immune cell invasion as an additional surrogate measure of rupture vulnerability.

We would like to note that though the fluorescent dye we use for the present project is not cell specific, various dye materials can be utilized to label different cells, cell markers, or sub-cellular features in the soft tissue to generate contrast between tissue components—for example, the amine-reactive derivatives of fluorescein isothiocyanate (FITC), which is widely used for various applications such as antibody labeling in immunofluorescence. The FDA-approved dye indocyanine green (ICG)^19,20 has the advantage of near-infrared excitation and could also be considered as a myriad of existing and emerging fluorescent reporters. Investigation of these dyes is warranted in our future work for novel potential applications.