The feasibility of photoacoustic microscopy (PAM) for evaluation of angiogenesis inhibitor was investigated on a chick embryo model in vivo. Different concentrations of the angiogenesis inhibitor, Sunitinib, were applied to the chorioallantoic membrane (CAM) of the chick embryos. Imaging of microvasculature in embryo CAMs was acquired using a laser-scanning PAM system; while the optical microscopy (OM) capturing the microvascular images of the same set of CAMs for comparison served as a gold standard for validating the results from PAM. The microvascular density as a function of applied Sunitinib concentration has been quantified in both PAM and OM images. The results from these two modalities have a good agreement, suggesting that PAM could provide an unbiased quantification of microvascular density for objective evaluation of anti-angiogenesis medication. In comparison with conventional OM which enables only two-dimensional enface imaging, PAM is capable of three-dimensional analysis of microvessels, including not only morphology but also functions, as demonstrated in part by the imaging result on a canine bladder model. The emerging PAM technique shows promise to be used in clinical and preclinical settings for comprehensive and objective evaluation of anti-angiogenesis medications.
© 2013 Optical Society of America
Angiogenesis, or the formation of new blood vessels out of pre-existing capillaries, has been associated with various physiological and pathological processes such as cancer, arthritis and exudative age-related macular degeneration . For example, it has been shown that cancer cells grow in three dimensional (3D) spheroids, and further growth of the tumor requires neovascularization and the tumor mass maintains itself by perfusion . Thus, the study of 3D angiogenesis can provide a unique comprehension on the process of cancer and other major diseases. It has been demonstrated that therapeutic interference with neovasculature formation can prevent diseases with excessive vessel growth. In recent years, an increasing number of anti-angiogenic drugs are under development aiming at treating diseases related to angiogenesis such as cancer. Therefore, 3D functional imaging of angiogenesis with high temporal and high spatial resolution may be of considerable help for not only better cancer diagnosis but also optimized treatment, e.g. objectively assessing the efficacy of anti-angiogenic drugs.
Several methodologies have been demonstrated to study 3D angiogenesis. The layer-by-layer biopsies of bone marrows were used for computer-aided 3D reconstruction images to reveal a complex arborizing and branching of microvessels in bone marrow with leukemia . The serial sectioning of tissue samples mechanically restricts the biopsy method from applications in vivo. Laser scanning confocal microscopy can capture stacks of confocal optical slices to digitally reconstitute a 3D projection of the vascular endothelial cells with high resolution. This technology, however, usually requires staining of vascular endothelial cells for angiogenesis imaging . 3D power Doppler ultrasound enables 3D study of cancer associated hemodynamic changes, and can render both anatomic (e.g. vascularization) and functional images (e.g. blood flow) information [5,6]. However, improving the spatial resolution of high-frequency power Doppler ultrasound beyond 100 μm is challenging, which makes it difficult to study the morphologies and functions of small capillaries from angiogenesis . Optical coherence tomography (OCT), with the ability to quantitatively assess hemodynamic parameter such as blood vessel diameter and flow using Doppler algorithms, has been adapted to noninvasive microangiography of the human retina and other organs such as brain [7–9]. Neither OCT nor Doppler ultrasound, however, has the direct access to the blood oxygenation, one of the key functional parameters of tissue micro-environment.
Photoacoustic microscopy (PAM) is an emerging hybrid imaging modality based on the detection of laser generated ultrasound signals [10–14]. Presenting the intrinsic optical absorption contrast among soft tissues with excellent sensitivity and spatial resolution, PAM holds the promise to provide high-quality 3D angiogenesis images. Based on how the focus is achieved for generating microscopic images, PAM has now been divided into two categories including acoustic-resolution PAM (AR-PAM) and optical-resolution PAM (OR-PAM) [10,11,13]. The former has achieved lateral resolution of 45 μm and axial resolution of 15 μm with imaging depth beyond 3 mm. In comparison with AR-PAM, OR-PAM, although with limited imaging depth (> 0.7 mm ), enables further elevated spatial resolution which is crucial for quantitative evaluation of 3D distributed small capillaries. The OR-PAM system with axial resolution of 15 µm has been shown as a promising tool for mapping neovascular network architecture during angiogenesis inhibitor responsiveness . Potentially, the OR-PAM system with axial resolution down to the size of capillaries (5-10 µm) should furnish better quantification of the 3D change of the capillary network in response to angiogenesis inhibitor. For both AR-PAM and OR-PAM, the axial resolution is determined by the bandwidth of detected photoacoustic signals instead of the confocal parameter. Thus, low NA lenses may be used without sacrificing the depth resolution. Similar to the situation in OCT, working with relatively low NA objectives for PAM allows imaging with large fields of view, enabling the synthesis of microscopic and macroscopic information . Using ultrasonic detectors with extremely high receiving bandwidth, axial resolution less than 10 μm has been realized [17,18]. This excellent axial resolution enables the sectioning along the depth direction of tissues in vivo, facilitating the visualization of 3D features in an intact biological object from tissue level to cellular level.
In this work, we explored the feasibility of OR-PAM for objective assessment of an angiogenesis inhibitor through quantitative imaging of microvasculature. The capability of our imaging system in 3D rendering and analysis of microvasculature was first demonstrated through the experiment on a canine bladder model, paving the road toward in vivo noninvasive biopsy of bladder cancer. Then by using a chick embryo model, we explored the performance of this system in quantifying the microvascular density and its change in response to angiogenesis inhibitor at different doses. The chick embryo model not only provides a reliable and easy-to-operate assay for evaluating agents with antiangiogenic activity but also enables conventional optical microscopy (OM) and digital image analysis (DIA) to be used as a gold standard to verify PAM results [19,20].
The schematic of our laser-scanning OR-PAM imaging system has been demonstrated in our previous publications [17,21,22]. Pulsed laser illumination is provided by an Nd:YAG laser (Spot-10-200-532, Elforlight Ltd., UK), working at 532 nm wavelength with a pulse duration of 2 ns. To obtain a 3D photoacoustic image, a 2D raster scan of the focused laser beam on the sample surface is conducted quickly by using high-speed 2D scanning mirrors while the ultrasound detector and the sample are kept static. Laser induced photoacoustic signals are detected by either a conventional PZT needle hydrophone or a microring resonator based ultrasound detector. The lateral resolution of this system is determined by the NA of the objective lens. When an achromatic lens with a focal length of 50 mm and NA of 0.25 was used as the objective lens, the system lateral resolution is 5 µm. The system axial resolution is determined by the bandwidth in photoacoustic signal detection. For example, when custom-built needle hydrophone with a center frequency of 35 MHz and a −6 dB bandwidth of 100% was used, the system can achieve an axial resolution of ~22 µm. When employing our microring resonator based ultrasound detector which facilitates an extremely broad detection bandwidth up to 100 MHz, this system can realize an excellent axial resolution of 8 µm.
To demonstrate the capability of this system in 3D rendering of microvascular structures, the imaging result from the study on a canine bladder model was analyzed . The microvascular structures in the canine bladder provide a good model for examining the system performance including spatial resolution (both lateral and axial), contrast-to-noise ratio, image continuity, and preliminary quantitative analysis. However, since conventional OM has no depth penetration and is not able to image the 3D vasculature in an optically scattering biological sample, the quantitative results from PAM of canine bladder cannot be validated by the OM and the established DIA technology. For this reason, a well-established chick embryo chorioallantoic membrane (CAM) model was employed for verifying the capability of OR-PAM in objective evaluation of angiogenesis inhibitor medication. CAM is one of the most popular models to study angiogenesis-related phenomenon due to its very dense capillary network lining on the surface [1,19,20,23–27]. Since all the vessels are distributed on a 2D surface, a commonly-used reflection-mode OM equipped with a digital camera can be used to capture the images of the CAM superficial vessels. The results from the DIA can then be adopted to verify the feasibility of PAM in quantification of vessel density.
Fertilized chick eggs were obtained from a hatchery (Townline, Zeeland, MI). An ex ovo chick embryo culture method was used for easy access to the chick embryo CAM . Briefly, eggs were removed from incubator after 72 hours of incubation and embryos were transferred to a sterile container such as a petri dish. Then, the embryos were returned to the incubator in a static position until further treatment. On embryo developmental day (EDD) 5, 10 μl of the angiogenesis inhibitor, Sunitinib, or 10 μl of 0.9% NaCl as a control was topically applied on the CAM. The position of placing the Sunitinib (or NaCl) drop was marked by a small piece of filter paper (~1 mm diameter) which was deposited carefully on the CAM using a sterile tweezers. After flow and perfusion, the Sunitinib (or NaCl) drop affected an area of ~5 mm in diameter, as examined by the changing of the vasculature in the CAM. At least two imaging areas, each with a size of 1.22 mm × 1.22 mm and adjacent to the filter paper, were selected randomly within the treated region. The embryos were then returned to the incubator in a static position for 24 hrs until the imaging experiment on EDD 6. In order to search for the optimal span of the concentrations of Sunitinib facilitating vessel densities at various levels, the pretest was performed and the best span of 30 to 300 µM was determined. The chick embryos were treated with Sunitinib at three different concentrations of 30, 100, and 300 µM, or with the control solution (0.9% NaCl), as shown in Fig. 1. To maintain the life of chick embryos, an IR lamp as a heating source was used to illuminate the chick embryos during acquisition of images .
The PAM images were obtained over a region of 1.22 mm × 1.22 mm with 256 × 256 pixels, and were displayed in 256 grey levels, i.e., 8 bits. A 4X objective was used in the OM to capture optical images with a size comparable to PAM images. The 8-bits color OM images were stored in BMP format. In general, we took OM images and then PAM images of the same embryo at the place marked by the filter paper. The quantification of both PAM and OM images from the chick embryos was realized following the DIA method described in the literature . Before analysis, the OM color images were converted into grayscale. Then, an adaptive thresholding was applied to convert the grayscale PAM and OM images to black/white images, i.e. ‘1’ and ‘0’ representing vessels and surrounding tissues, respectively. During adaptive thresholding, we partitioned the original image into many sub-images, and normalized every pixel value to between 0 and 1 in each sub-image. The mean value M in that sub-image was calculated. Then, for every pixel value p, if p > M + C, where C is a user-defined constant, p was set to ‘1’ as the blood vessels; otherwise ‘0’ as the surrounding tissues was assigned. Finally, the vessel density was determined by calculating the ratio of the number of pixels with ‘1’ to that of total pixels (i.e., the area of white pixels over the total imaging area). In the adaptive thresholding, different values of the number of sub-images and the constant C were selected to optimize the similarity of the vascular pattern between the input grayscale images and the output black/white images, and were dependent on the characteristics of images including lighting or signal uniformity, vessel density, and vessel texture. The DIA method for calculating the vessel density following the description in Ref . can be directly applied to 2D images of chick embryos. For the calculation of vessel density in the 3D image of the canine bladder, same idea was followed and the 3D vessel density was calculated by counting the volume of white pixels over the total imaging volume.
Label-free PAM images presenting the 3D microvasculature in the inner wall of a canine bladder and the image analysis are shown in Fig. 2. The quantitative evaluation of vessel density using the DIA method described above was conducted. In Figs. 2(a)-2(c), the left column shows the imaging result in grayscale which, after image processing, was turned into black/white images, as shown in the right column. The constant value C of 0.2 and the number of sub-images of 36 (partitioned along X and Y, i.e., a 2D partition) were chosen in image conversion. Figures 2(b) and 2(c) show the 2D cuts from the 3D PAM image along the two lines indicated in Fig. 2(a). In these 2D B-scan images, the microvasculature with 2-3 layers in the sample can be clearly observed. A vessel density in 3D of 2.4% over a volume of 1.04 mm × 1.04 mm × 0.52 mm was extracted from the converted black/white images. Besides, facilitated by the excellent spatial resolution of PAM, the vasculature from large to small vessels can all be recognized clearly with satisfactory contrast-to-noise ratio, as can be seen from the maximum amplitude projection (MAP) image in Fig. 2(a). The highest axial resolution of this system was quantified as 8 μm . The shortest length scale of imaged vessels was ~20 μm, as shown in the A-line photoacoustic signal in Fig. 2(d). Figure 2(e) shows the depth-encoded MAP image on the XY plane, demonstrating the sectioning capability of PAM.
Figure 3(a) shows an example of co-registered PAM and OM images of the vasculature on a chick embryo CAM on EDD 5. As can be seen, the two images from PAM and OM have good match for either large vessels of ~160 µm or networks of micro-vessels down to ~10 µm, as well as the branching points. Figure 3(b) depicts the characteristic microvasculature of the embryo on EDD 6 after 100-µM Sunitinib was applied on EDD 5. The significantly decreased vessel density on the CAM as a result of the applied angiogenesis inhibitor can be clearly recognized in both PAM and OM images, which verifies that the anti-angiogenesis medication works well on the chick embryo CAM model. The signal-to-noise ratios (SNRs) of Figs. 3(a) and 3(b) are 22 dB and 24 dB, respectively.
For quantitative study, chick embryos were divided into four groups and treated on EDD 5 with 0.9% NaCl (control), or 10 µl of Sunitinib at 30, 100, and 300 µM concentrations respectively. For each group, the number of embryos was ≥ 3. PAM and OM images were taken 24 hrs after the treatment (EDD 6). On each embryo, images were taken from at least two different locations both within the treated area. Therefore, for each condition, we have at least 2 × 3 = 6 samples which facilitate the calculation of the average and the standard deviation (STD). Figure 4 shows the representative PAM images from two individual chick embryos as examples, demonstrating clearly the gradual changes in microvascular density from plentiful to sparse as a result of the increased concentration of the Sunitinib applied on the CAM.
The DIA method described above for the extraction of angiogenic density was pretested through the observation by the eye to determine the most suitable value of C and the number of sub-images . More specifically, we visually compared the images before (grayscale) and after the thresholding process (black/white) and saw if the processed images still rendered the characteristic vascular features. This visual assessment was conducted over all the PAM and OM images from all the four sample groups in order to decide the optimal parameters which include the constant value C of 0.2 and the number of sub-images of 36. Figure 5 illustrates the examples of PAM image conversion from grayscale to black/white using the selected parameters, where Figs. 5(a) and 5(b) show the results from embryos with high and low vascular densities respectively. For both of these two cases, the majority of vascular features were kept well after the conversion. The converted images in black/white were then used for quantification of the vessel densities in each specimen.
Figure 6 shows the statistical results of the vessel density in each group extracted from PAM and OM images. By comparing the two curves from PAM and OM respectively, a similar dependence (i.e. slope) of the vessel density over applied drug concentration from 30 to 300 µM can be seen. A lower-than-expected vessel density and a relatively large STD are observed in the control condition of the PAM curve, which is believed to be a result of the large sample-to-sample difference. Moreover, the inherent large variation in vessel density exists among different regions on the same CAM surface. The discrepancy in absolute values of vessel density between PAM and OM will be discussed later. For better comparison, the two curves from PAM and OM are normalized respectively, as shown in the inset of Fig. 6. For normalization, the average value of vessel densities over three conditions (i.e. 30, 100 and 300 µM) is subtracted from the vessel density at each condition. The data of the control condition is excluded due to the issue mentioned above. The two curves after normalization showing the relative vessel densities are in good agreement except at the control condition.
4. Discussion and conclusions
In this work, we demonstrated ex vivo imaging of canine bladder (Fig. 2) using the microring resonator, to prove the depth sectioning capability, and in vivo imaging of chick embryos (Figs. 3-5) using the PZT needle hydrophone, to validate the capability in quantitative evaluation of vascular density in response to angiogenesis inhibitor. Microring, although providing a better axial resolution , was not used for chick embryos mainly because it comes with a large footprint (~1 cm wide) and, therefore, is not convenient for reflection-mode imaging. PZT needle hydrophone, with a diameter of 1 mm, facilitates a convenient imaging setup on chick embryos; while the limited axial resolution of hydrophone was not a problem for imaging 2D vasculature in CAM of chick embryos. Further development on the microring resonator to facilitate reflection-mode operation is part of the future work .
Due to the spatial sensitivity of the PZT hydrophone, PA images in Figs. 3-5 show brighter in the center region and dimmer on the edge. The adopted adaptive thresholding helps alleviate the issue of nonuniform spatial sensitivity of PA images by converting the original grayscale PA images to black/white images. In the work described in Ref , the issue of nonuniform lighting effect (i.e., nonuniform spatial sensitivity) was also effectively addressed by using the same adaptive thresholding method.
As we can see from Fig. 3 as an example, there is no obvious difference between PAM and OM in describing large vessels. However, there is noticeable discontinuity in small vessels and capillaries in the PAM images, which does not exist in the OM images. The main reason is that OM images are captured by a complementary metal–oxide–semiconductor sensor which has a much longer temporal integration; while in PAM, when no signal averaging is involved, the signal at each image pixel is an integration over a very short period of 2 ns as determined by laser pulse duration. Therefore, unlike OM, PAM shows the transient image of the red blood cells sparsely distributed in the capillaries, which is the main reason for the discontinuity in the PAM images. This also, at least in part, explains why the blood vessel densities quantified by PAM are lower than those by OM, as observed in Fig. 6.
Drugs that interrupt angiogenesis have shown promise in treating cancer and other diseases including age-related macular degeneration. Developing powerful tools for objective and comprehensive assessment of the efficacy of anti-angiogenesis drugs are necessary. The imaging tools enabling 3D quantification of angiogenesis levels and inhibition status is considered favorable for examination of drugs in anti-angiogenic procedures. In this work, we explored whether the emerging PAM has the potential to enable noninvasive, label-free and 3D imaging for the determination of angiogenesis level and the evaluation of inhibition agents. The experiments on canine bladders and chick embryos demonstrated that PAM is able to quantify the microvascular density and its change in response to angiogenesis inhibitor. The imaging results from PAM on chick embryos were verified by the established OM and DIA method which was employed as a gold standard. Compared to conventional OM which is a 2D imaging method and has no penetration in optically scattering tissues, PAM, when facilitated with high frequency ultrasonic detectors, could be more practical for clinical and preclinical applications due to its 3D imaging and sectioning capability. PAM visualizes the intrinsic optical absorption contrast in tissues. As a result of the strong optical absorption of hemoglobin in the visible to near-infrared spectrum, PAM is able to present the smallest capillaries and even individual red blood cells . The great sensitivity of PAM in imaging microvasculature without need of any contrast agents will benefit the research or clinical evaluation of anti-angiogenesis medications.
In this work, we have focused on the validation of PAM for evaluation of only the morphology of vasculature and its change in response to angiogenesis inhibitor. One of the unique advantages of PAM is its functional imaging ability, including the high sensitivity to blood flow and blood oxygen saturation. We believe these functional hemodynamic properties of tissue microenvironment could render more direct and faster response to medications, and quantitative study of these properties in vivo may shed light on the understanding of the performance of new drugs. In the next step, our PAM system will be developed further, e.g. by adding a tunable laser, to expend its functional imaging capability. In addition, PAM can also be integrated with other imaging modalities including confocal fluorescence microscopy  so that a single scan on a biological sample can render multiple contrasts each sensitive to different physiological or morphological information in the sample, providing more comprehensive evaluation of drugs.
Supports from the National Institutes of Health grant R01AR060350, University of Michigan-Shanghai Jiao Tong University (UM-SJTU) Joint Grant, and Samsung GRO Program are gratefully acknowledged.
References and links
1. A. W. Griffioen and G. Molema, “Angiogenesis: potentials for pharmacologic intervention in the treatment of cancer, cardiovascular diseases, and chronic inflammation,” Pharmacol. Rev. 52(2), 237–268 (2000). [PubMed]
3. A. R. Perez-Atayde, S. E. Sallan, U. Tedrow, S. Connors, E. Allred, and J. Folkman, “Spectrum of tumor angiogenesis in the bone marrow of children with acute lymphoblastic leukemia,” Am. J. Pathol. 150(3), 815–821 (1997). [PubMed]
4. I. Rajantie, M. Ilmonen, A. Alminaite, U. Ozerdem, K. Alitalo, and P. Salven, “Adult bone marrow-derived cells recruited during angiogenesis comprise precursors for periendothelial vascular mural cells,” Blood 104(7), 2084–2086 (2004). [CrossRef] [PubMed]
5. H. Pairleitner, H. Steiner, G. Hasenoehrl, and A. Staudach, “Three-dimensional power Doppler sonography: imaging and quantifying blood flow and vascularization,” Ultrasound Obstet. Gynecol. 14(2), 139–143 (1999). [CrossRef] [PubMed]
6. J. W. Xuan, M. Bygrave, H. Jiang, F. Valiyeva, J. Dunmore-Buyze, D. W. Holdsworth, J. I. Izawa, G. Bauman, M. Moussa, S. F. Winter, N. M. Greenberg, J. L. Chin, M. Drangova, A. Fenster, and J. C. Lacefield, “Functional neoangiogenesis imaging of genetically engineered mouse prostate cancer using three-dimensional power Doppler ultrasound,” Cancer Res. 67(6), 2830–2839 (2007). [CrossRef] [PubMed]
9. J. A. Izatt, M. D. Kulkarni, S. Yazdanfar, J. K. Barton, and A. J. Welch, “In vivo bidirectional color Doppler flow imaging of picoliter blood volumes using optical coherence tomography,” Opt. Lett. 22(18), 1439–1441 (1997). [CrossRef] [PubMed]
10. H. F. Zhang, K. Maslov, G. Stoica, and L. V. Wang, “Functional photoacoustic microscopy for high-resolution and noninvasive in vivo imaging,” Nat. Biotechnol. 24(7), 848–851 (2006). [CrossRef] [PubMed]
12. L. V. Wang, “Tutorial on photoacoustic microscopy and computed tomography,” IEEE J. Sel. Top. Quantum Electron. 14(1), 171–179 (2008). [CrossRef]
15. S. S. Oladipupo, S. Hu, A. C. Santeford, J. Yao, J. R. Kovalski, R. V. Shohet, K. Maslov, L. V. Wang, and J. M. Arbeit, “Conditional HIF-1 induction produces multistage neovascularization with stage-specific sensitivity to VEGFR inhibitors and myeloid cell independence,” Blood 117(15), 4142–4153 (2011). [CrossRef] [PubMed]
16. V. J. Srinivasan, A. C. Chan, and E. Y. Lam, “Doppler OCT and OCT angiography for in vivo imaging of vascular physiology,” Chapter 2 in Selected Topics in Optical Coherence Tomography, book edited by Gangjun Liu (2012).
18. C. Zhang, K. Maslov, J. Yao, and L. V. Wang, “In vivo photoacoustic microscopy with 7.6-µm axial resolution using a commercial 125-MHz ultrasonic transducer,” J. Biomed. Opt. 17(11), 116016 (2012). [CrossRef] [PubMed]
19. C. Peifer and G. Dannhardt, “A novel quantitative chick embryo assay as an angiogenesis model using digital image analysis,” Anticancer Res. 24(3a), 1545–1551 (2004). [PubMed]
20. C. N. Doukas, I. Maglogiannis, and A. A. Chatziioannou, “Computer-supported angiogenesis quantification using image analysis and statistical averaging,” IEEE Trans. Inf. Technol. Biomed. 12(5), 650–657 (2008). [CrossRef] [PubMed]
21. S.-L. Chen, Z. Xie, P. L. Carson, X. Wang, and L. J. Guo, “In vivo flow speed measurement of capillaries by photoacoustic correlation spectroscopy,” Opt. Lett. 36(20), 4017–4019 (2011). [CrossRef] [PubMed]
22. S.-L. Chen, Z. Xie, T. Ling, L. J. Guo, X. Wei, and X. Wang, “Miniaturized all-optical photoacoustic microscopy based on microelectromechanical systems mirror scanning,” Opt. Lett. 37(20), 4263–4265 (2012). [CrossRef] [PubMed]
25. A. Mazzocca, E. Fransvea, G. Lavezzari, S. Antonaci, and G. Giannelli, “Inhibition of transforming growth factor beta receptor I kinase blocks hepatocellular carcinoma growth through neo-angiogenesis regulation,” Hepatology 50(4), 1140–1151 (2009). [CrossRef] [PubMed]
26. I. Goze, A. Cetin, and A. Goze, “Investigation of effects of essential oils of Origanum minutiflorum O Schwarz PH Davis and Cyclotrichium niveum (Labiatae) plants on angiogenesis in shell-less chick embryo culture,” Afr. J. Biotechnol. 9, 2156–2160 (2010).
27. P. Nowak-Sliwinska, A. Weiss, J. R. van Beijnum, T. J. Wong, J.-P. Ballini, B. Lovisa, H. van den Bergh, and A. W. Griffioen, “Angiostatic kinase inhibitors to sustain photodynamic angio-occlusion,” J. Cell. Mol. Med. 16(7), 1553–1562 (2012). [CrossRef] [PubMed]
28. H. C. Yalcin, A. Shekhar, A. A. Rane, and J. T. Butcher, “An ex-ovo chicken embryo culture system suitable for imaging and microsurgery applications,” J. Vis. Exp. 44(44), 2154 (2010). [PubMed]
29. S.-L. Chen, Z. Xie, L. J. Guo, and X. Wang, “A fiber-optic system for dual-modality photoacoustic microscopy and confocal fluorescence microscopy using miniature components,” Photoacoustics 1(2), 30–35 (2013). [CrossRef]