J. Biophotonics 8, No. 9, 730–739 (2015) / DOI 10.1002/jbio.201400086
FULL ARTICLE
Polarization second harmonic generation microscopy provides quantitative enhanced molecular specificity for tissue diagnostics Rajesh Kumar*, 1, Kirsten M. Grønhaug2, Elisabeth I. Romijn1, Andreas Finnøy1, Catharina L. Davies1, Jon O. Drogset3, and Magnus B. Lilledahl1 1
Department of Physics, Norwegian University of Science and Technology (NTNU), N-7491, Trondheim, Norway Levanger Hospital, Norway 3 Trondheim University Hospital, Norway 2
Received 30 July 2014, revised 29 September 2014, accepted 6 October 2014 Published online 4 November 2014
Key words: polarization second harmonic generation (p-SHG) microscopy, collagen fiber, susceptibility (χ) tensor imaging, nonlinear optical imaging, multiphoton microscopy, osteoarthritis, human knee articular cartilage, biomedical optical diagnosis
Due to specific structural organization at the molecular level, several biomolecules (e.g., collagen, myosin etc.) which are strong generators of second harmonic generation (SHG) signals, exhibit unique responses depending on the polarization of the excitation light. By using the polarization second harmonic generation (p-SHG) technique, the values of the second order susceptibility components can be used to differentiate the types of molecule, which cannot be done by the use of a standard SHG intensity image. In this report we discuss how to implement p-SHG on a commercial multiphoton microscope and overcome potential artifacts in susceptibility (χ) image. Furthermore we explore the potential of p-SHG microscopy by applying the technique to different types of tissue in order to determine corresponding reference values of the ratio of second-order χ tensor elements. These values may be used as a bio-marker to detect any structural alterations in pathological tissue for diagnostic purposes.
1. Introduction Second harmonic generation (SHG) is a nonlinear optical process, in which two photons of the same frequency interact with a nonlinear material and generate another photon having twice the energy of * Corresponding author: e-mail: [email protected] © 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The SHG intensity image (red) in (a) shows the distribution of collagen fibers in ovary tissue but cannot determine the type of collagen fiber. However, the histogram distribution (b) for the values of the χ tensor element ratio can be used to quantitatively identify the types of collagen fibers.
the incident photon. A few fibrillar structures in biological tissues like fibers of collagen, myosin and tublin fulfill the non-centrosymmetric structural condition and produce enough SHG photons to create an intensity based detectable SHG signal [1–6]. SHG signals are produced by endogenous structural pro-
J. Biophotonics 8, No. 9 (2015)
teins in tissue [4]. However, different molecular sources of SHG signals cannot be distinguished in a standard SHG intensity image. In several medical diagnostic applications, differentiation among the constituents of the tissue is an essential requirement. For example, during the progression of osteoarthritis, hyaline cartilage (mainly collagen type II) is being replaced by fibrocartilage (mainly type I collagen) [7]. In such case, it is important to detect any early stage modifications and identify different types of collagen fiber in the extracellular matrix of articular cartilage. By incorporation of polarization sensitivity in SHG images, it is possible to identify the different molecular sources of SHG scatters (e.g. collagen I, collagen II, myosin etc.) based on their susceptibility value (χ) as a quantitative parameter [8]. Incorporation of polarization information in SHG images can be accomplished by excitation polarization measurement in polarization second harmonic generation (p-SHG) microscopy. This technique provides an image of the sample based on the values of the susceptibility components of tissue constituents. The image obtained from p-SHG technique, which shows the morphology of the tissue based on susceptibility values, is referred to as a χ-image. In a χ-image, susceptibility values can be visualized using a false color map. A localized molecular inhomogeneities in the tissue can change susceptibility values. Therefore a change in local environment may appear as different colors in the χ-image. Such differentiation of different tissue constituents is not possible from an SHG intensity image alone [9–11]. Therefore p-SHG can potentially be useful in medical diagnosis. In last few years, p-SHG technique have been used to investigate a few biomedical problems e.g. breast cancer detection [12, 13] and pathological skin states [14]. Characterization of muscle [11] and engineered cartilage tissue has also been performed [14]. Collagen in extracellular matrix is highly altered in several pathological cases like connective tissue diseases, cardiovascular diseases, autoimmune diseases, ovarian diseases etc. [15]. p-SHG microscopy may quantitatively detect modifications at the molecular level and differentiate types of fibrillar collagen inside the extracellular matrix of the tissue. However, the application of p-SHG technique has been very limited to date and has not explored in many other tissue organs. It has to be noted that the value of the χ parameter for a biomolecule may change in different tissue types. Hence it is essential to obtain the prior knowledge of the χ parameter value for a specific tissue type, before applying it as an optical diagnostic parameter in pathological analysis. In this report, to the knowledge of the authors, p-SHG analysis for measuring χ-parameters value of heart muscle, ovary tissue and osteoarthritic cartilage, was performed for the first time. We successfully demonstrate that pSHG technique can detect modifications in human ar-
www.biophotonics-journal.org
731
ticular cartilage in the early stages of osteoarthritis, which in histology generally appears at the advanced stages of the disease. Our results indicate that pSHG has the capability of differentiating various sources of SHG at the molecular level and show potential for a new diagnostic tool in pathological analysis. Moreover, all our SHG measurements were performed in the backscatter imaging mode. This demonstrates that the technique is compatible with in vivo investigation by the use of an endoscopic probe, which necessarily uses the backscattered signal. Miniaturized multiphoton probes for endoscopic applications have been demonstrated by several groups [16–18]. We anticipate that the proof-of-concept results reported in this paper may provide a basis to apply p-SHG technique in diagnosis of various diseases relevant with cardiovascular, ovary and cartilage disorders. Moreover, in this report, we described the issues responsible for image artifacts in p-SHG analysis and provided the relevant solutions in terms of calibrations, which were not discussed previously.
2. Materials and methods 2.1 Background The physical model associated with collagen fibers has previously been described [8, 19–21]. Briefly, under the assumption of cylindrical symmetry for collagen fibers, that the sample lies in xz-plane (Figure 1) and, the propagation direction of the excitation laser beam is along the y direction (out-of-plane); the incoming SHG signal intensity I from the harmonophore can be modeled as ( 2 χzzz 2 2 I¼c sin ðθu θv Þ þ cos ðθu θv Þ χzxx ) 2 χxzx 2 þ sin 2ðθu θv Þ ð1Þ χzxx
Figure 1 Schematic diagram of coordinate system. The direction of the polarization of the laser beam and the orientation of collagen fibers are indicated.
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
732
R. Kumar et al.: Second harmonic generation microscopy for optical diagnosis
The above equation can be used to fit the signal intensity I as a function of difference in angle between the polarization of the incoming light (θu) and the orientation of the collagen fiber (θv), as well as the second-order susceptibility tensor ratios χzzz/χzxx χ1 and χxzx/χzxx χ2. By applying input parameter I and θu to the physical model, the numerical value of other variables can be found. The obtained values χ1 and χ2, provide a quantitative parameter which can be used for the purpose of biomedical diagnosis.
2.2 Experimental The experimental analysis is based on the excitation polarization measurement technique [8, 22, 23]. A standard Zeiss microscope was modified to achieve required experimental measurements. Although a few reports relevant with p-SHG microscopy are available [22–25], the difficulties involved in modification of a standard commercial microscope for development of a p-SHG optical system, have not been discussed previously. If ignored, polarization dependent effects introduced by the microscope can create artifact in image, which ultimately may lead to misleading results. In commercial microscope, there are several optical elements present along the path of the excitation beam, which are not specially designed for optical polarization measurements. Due to presence of such optical elements, χ-image obtained from p-SHG analysis is prone to artifacts. Such problems can be resolved by the implementation of appropriate calibration procedures (described in next subsections). Another issue is that if a commercial system is modified for p-SHG application, then, due to insertion of new optical elements (e.g., polarizers, waveplates etc.) in the optical path, other types of microscopy measurements can be affected. To overcome this problem, a portable optical module (Figure 2) was incorporated in a standard commercial system in such a way that it can be easily removed from the optical path, after the p-SHG measurements, and, hence the complete microscope systems
Figure 2 A photograph of the portable polarization optics module. LP: Linear polarizer, λ/4-plate: Quarter wave plate, λ/2-plate: Half wave plate.
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
again available for other kind of experiments without any modification of the optical system.
2.2.1 Excitation resolved p-SHG A commercial Zeiss LSM 510 META microscope was modified by the insertion of the polarization optics module in the path of the Ti : Sapphire (Coherent Mira) laser to perform excitation polarization measurements. A simplified schematic diagram of the experimental setup is shown in Figure 3. For the excitation of the tissue sample, 200 fs laser pulses with a repetition rate of 76 MHz was tuned to 790 nm. A band pass filter (395 ± 25 nm) was used to select SHG signal, while a combination of dichroic mirror and short pass filter was used to reject backscattered laser light. Inside the polarization optics module, the collimated laser beam was allowed to pass through the linear polarizer (Thorlabs, LPNIR050) to create a pure linearly polarized beam suitable for excitation. Sample was then illuminated by different angle of excitation ranging from 0° to 360°. The rotation of linearly polarized light was performed by the rotation of half wave plate (Thorlabs, λ/2 : 690–1200 nm). A pre-compensation in ellipticity (as described in next subsection) was achieved by a quarter wave plate (Thorlabs, λ/4 : 690–1200 nm). There are a few important issues, discussed below, which have to be considered during excitation resolved p-SHG measurements to avoid potential errors.
2.2.2 Calibration of pre-compensation for each excitation polarization angle One of the most important requirements for excitation polarization measurement is that the sample has to be illuminated by a perfect linearly polarized light. A high quality linearly polarized beam can be achieved by the use of a linear polarizer (LP). Even though high quality linearly polarize beam is created by the use of a LP, ellipticity is introduced during the propagation of light through scanning optics and a dichroic mirror in the optical path. In general, the optical elements present in commercial microscope are not optimize to maintain the polarization of the light beam. Hence, light arrived at sample plane is no longer a linearly polarized light. It is important to mention that the introduced ellipticity in the optical beam is different for a different angle of linear excitation. Hence to avoid any artifact in a χ-image, appropriate pre-compensation is required for each angle of linear excitation. The required pre-compensation can be achieved by a quarter wave plate in the optical path for each angle of linear excitation. In www.biophotonics-journal.org
J. Biophotonics 8, No. 9 (2015)
733
Figure 3 A simplified schematic diagram of the experimental setup for p-SHG microscopy. A χ-image is created by performing excitation polarization measurements with the use of the polarization optics module, which is shown inside the dottedline box. LP: Linear polarizer, LA: Linear analyzer, SP: Short pass, λ/4-plate: Quarter wave plate, λ/2-plate: Half wave plate, SHG: Second harmonic generation, TPEF: Two-photon excited fluorescence.
order to confirm that the light arrived at sample plane is linearly polarized light, additional experimental measurements can be performed, which is described in appendix A. Hence, by the introduction of an additional negative pre-compensation in order to compensate the introduced ellipticity for each corresponding angle of linear excitation, a calibration data for the specific microscope is obtained, which can be used during the excitation polarization measurements.
2.2.3 Optical power calibration Another important issue arises due to polarization dependent reflectivity of the dichroic mirror present in the optical path of the microscope. The actual excitation power of laser light arriving at the sample plane may change at a different angle of linear exci-
www.biophotonics-journal.org
tations because the presence of dichroic mirror. Any fluctuation in the laser power due to involved optics in the optical path of microscope may, cause additional errors in the result. Hence, in order to compensate the change in optical power arriving at sample plane, an adjustment of optical power for each corresponding angle of linear excitation provides another calibration, which should be employed during the excitation polarization measurements.
2.2.4 Polarization scrambling To perform SHG analysis high numerical aperture (NA) microscope objective is preferred, because, it provides high density of photons at focal point, which is an essential requirement to achieve nonlinear optical phenomena. On the other hand, high NA objective can scramble the polarization state of the excita-
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
734
R. Kumar et al.: Second harmonic generation microscopy for optical diagnosis
tion light. In a theoretical analysis it is mentioned that a lower NA microscope objective is desired to maintain the polarization property of light at focal volume of microscope objective [26]. Hence it is suggested to use a moderate NA objective for p-SHG measurements. To the knowledge of the authors there is no quantitative result available regarding the scrambling of linear polarization by high NA microscope objectives. A measurement was performed to quantify the scrambling of linear polarization, only due to the involvement of microscope objective in the optical path. Linearly polarized light is passed through a linear analyzer (LA) in trans-illumination configuration (Figure 3) through the microscope objective of different NA. The ratio of intensity count during parallel and cross position of LA, with respect to the linearly polarized excitation light, indicates the amount of scrambling caused by the microscope objective. The results from this measurement is shown in appendix B. Moderate NA microscope objective (e.g., 0.8 NA) is advisable to perform excitation resolved p-SHG measurements [8, 27].
2.3 Sample preparation The use of human tissue (osteoarthritic cartilage) in this study was approved by the regional committee for medical research ethics (2013/265 REK, Norway). Slides of mouse tissue (ovary and heart) was purchased from Abgent Europe Ltd. Paraffin embedded tissue sections (thickness 5 μm) was deparaffinized by dipping slides in two separate xylene solution, each dip for 5 minutes. Dehydration of the slides was performed through a series of graded alcohol 96%, 86% and 70% for a duration of 5 minutes, 4 minutes and 3 minutes, respectively. Slides were finally rehydrated in water for a duration of 1 minute. Tissue sections were then covered by a standard cover slip. Edges of the cover slip were sealed with vaseline to avoid any dehydration of the tissue sections. All slides were stored at 4 °C between measurements. In order to demonstrate the potential of the technique, in this proof-of-concept study, one tissue slide of each tissue type (ovary, heart and cartilage) was analyzed. In every tissue slide each measurement was performed three times; additionally, at two different locations. Similar results were observed.
2.4 Image acquisition and data analysis In order to minimize the effect of birefringence due to tissue sample, all SHG images were acquired at 3 μm below the surface of tissue. The sample was © 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
excited by equally spaced, 11 different angles of excitations, ranging from 0° to 360°. Each angle of excitation provides an image. Hence, in total, we have 11 images per sample, which are polarizationresolved SHG images. Data analysis was performed in Matlab (The Mathworks, 2014). By fitting Eq. (1) to the measurements, the χ-tensor element ratios can be determined at every pixel of the images. The known value is intensity (I), and excitation angle (θu), while other variables can be found performing least square fit to the model. The peak value of Gaussian fit into the histogram plot of susceptibility tensor components provides the value of χ1 and χ2, which is a characteristic value of the tissue. All Gaussian fit have been expressed with 95% confidence bounds.
3. Result and discussion 3.1 Ovary Ovary tissue is considered as quite dynamic tissue where remodeling process inside the matrix of tissue occurs continuously [28]. In general, collagen I, III and IV are major types of collagen present inside the tissue matrix. The relative quantification of different types of collagen is very important in understanding of various reproductive disorders [28, 29]. It is worth to mention that collagen type IV is a nonfibrillar collagen, and, hence it does not produce SHG signal. Therefore, p-SHG technique cannot image collagen IV. In this section, we are presenting p-SHG as a potential method for relative quantification of the content of different types of fibrillar collagen in ovary tissue. In Figure 4a, the SHG signal (red color) shows the distribution of the collagen fibers, while- autofluorescence (green color) represents other components present inside the ovary tissue. Although the intensity based SHG image can clearly distinguish collagen fibers from other tissue components (as shown in the Figure 4a), it cannot differentiate the types of collagen molecule present in extracellular matrix of the tissue. To identify collagen types in ovary tissue, p-SHG analysis was performed. The distribution of the histogram for the value of χ1 and χ2, are shown in the Figure 4b and 4c respectively. The Gaussian fit to the histogram plot of χ1 results in two peaks, as shown in Figure 4b. We identify that the larger peak at χ1 = 1.20 (1.999, 1.200) is associated with collagen I, as this value is close to the earlier reported value of collagen I in tendon [8]. The other peak is at χ1 = 0.83 (0.835, 0.836) is very close to the value of collagen III reported earlier [8]. Hence we identify that the smaller peak, in Figure 4b represents an-
www.biophotonics-journal.org
J. Biophotonics 8, No. 9 (2015)
735
Figure 4 Ovary tissue (a) SHG (red) and autofluoroscence (green) image (b) Histogram plot of χ1 value (c) Histogram plot of χ2 value (d) χ1 image.
other major component of fibrillar collagen found in ovary i.e., collagen III. The Gaussian fit to the distribution of χ2, as shown in Figure 4c, shows one peak at χ2 = 0.38 (0.384, 0.389). The appearance of only one peak indicates that χ2 is not able to create additional contrast in the χ-image. The reason is that parameter χ2 is more sensitive to SHG intensity fluctuations, as we can see in Eq. (1), so less accurate in value. Hence χ1 has a better diagnostic capability. However, to find the comparative suitability between χ1 and χ2 in order to determine the diagnostic capability, in general, a separate analysis is recommended for a specific disease associated with a tissue. The χ1 image of the ovary tissue, which is obtained by providing the false color to the value of χ tensor element ratio, is shown in the Figure 4d. The image indicates the capability of χ image in visualizing any relevant modification in ECM of tissue, which may appear by the variation in color. Moreover, by calculation of the ratio of Gaussian area (i.e., collagen I/collagen III) from the Figure 4b, as described in previous report [8], it is possible to find a relative estimate for the different types of collagen content quantitatively inside the ovary tissue. Evaluation of such estimate may play an important role in revealing hidden features of several biophysical dynamic process associated with ovary tissue [28, 29].
3.2 Heart (cardiac muscle) One of the main constituent of muscle tissue is myosin. It is endowed with extraordinarily long α-helices www.biophotonics-journal.org
which are highly aligned, especially in the tail portion and hence responsible for coherent SHG emission [30]. Structural modificationin cardiac muscle fiber can be an early biomarker in several cardiac dysfunctions [31, 32]. Due to any such alterations in cardiovascular tissue the characteristic value of χ1 may change, and, hence p-SHG microscopy potentially able to detect subtle structural differences at the level of myosin architecture, which would not be possible to observe directly by typical histology due to limited resolution. Although χ1 value of myosin (from chicken wing muscle) can be found in Ref. [8], it was not analyzed and reported for heart muscle yet. Figure 5a shows the SHG intensity image of the cardiac muscle fiber, which was obtained from a mouse heart. Based on p-SHG analysis, we obtained the χ tensor element ratio, χ1 and χ2, of cardiac muscle, which are shown in the Figure 5b and 5c. respectively. The χ1-image of heart muscle, which is obtained by providing the false color to the value of χ tensor element ratio, is shown in the Figure 5d. This image indicates that the value of χ tensor element ratio is capable in visualizing the morphology of tissue and hence any alteration in structure may appear by the variation in color. The Gaussian fit to the histogram distribution provides χ1 = 0.86 (0.859, 0.861) and χ2 = 0.50 (0.500, 0.514), as shown in the Figure 5b and 5c respectively. It is worth to note that the value of χ1 obtained for the heart muscle is slightly different from the earlier reported [8] value of myosin as 1.04 ± 0.08 for the chicken wing muscle. We hypothesized that the difference in value is caused by structural differences in the myosin assembly. Hence this analysis shows the importance of prior knowledge of χ-value, which is © 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
736
R. Kumar et al.: Second harmonic generation microscopy for optical diagnosis
by International Cartilage Repair Society (ICRS) [39, 40], was chosen for the p-SHG analysis. Figure 6a shows intensity based SHG image of osteoarthritic cartilage. The characteristic value of χ1 was evaluated for three diffrent regions of interst ROI 1, ROI 2 and ROI 3, which represent fibrocartilage, hyaline cartilage and composite cartilage respectively. The Gaussian fit to the histogram provides χ1 = 1.20 (1.201, 1.206) and χ1 = 1.07 (1.069, 1.072) for ROI 1 (Figure 6b) and ROI 2 (Figure 6c) respectively. The histogram plot for the value of χ1 associated with the ROI 3 (Figure 6d) shows two different Gaussian peaks, as expected, at 1.21 (1.210, 1.227) and 1.10 (1.100, 1.106), which represents fibrocartilage and hyaline cartilage, respectively. The two different values of χ1 in ICRS Grade I cartilage sample, illustrate the molecular modification in the cartilage matrix, from hyaline to fibrocartilage, even in early stage of OA. This is believed to be visible only at advance stages of disease in histological assessment. This result shows the capability of p-SHG technique in early diagnosis of OA. Figure 5 Heart muscle (a) SHG image (b) Histogram plot of χ1 value (c) Histogram plot of χ2 value (d) χ1 image.
necessarily required for a specific tissue type, before applying it as an optical diagnostic parameter in pathological analysis.
3.3 Osteoarthritic cartilage Collagen is one of the major constituents of the extracellular matrix of cartilage [33]. Since p-SHG is capable in differentiation of different types of collagen fiber, we applied this technique in a clinically relevant study of osteoarthritis (OA). It is believed that during progress of OA the quality of cartilage continuously degrades. It has been reported that during the progression of the disease, replacement of collagen fibers takes place in the extra cellular matrix of cartilage [34, 35]. The type of cartilage in repair tissue is not generally hyaline cartilage but fibrocartilage [7, 36]. It is important to differentiate types of cartilage because fibrocartilage is not as good as hyaline cartilage in terms of mechanical strength, stability and adaptability. Hyaline cartilage is composed mainly of collagen II, while, fibrocartilage mainly of collagen I. Hence, in general, fibrocartilage is expected to yield relatively more intense SHG image in comparision with hyaline cartilage [36, 37], as collagen I fibrils are generally thicker and therefore emits stronger SHG signal. We applied p-SHG method on osteoarthritic human articular cartilage obtained from the knee. Early stage osteoarthritic cartilage, Grade-I, defined
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
4. Conclusion In this study, we explored the application of p-SHG microscopy for probing the molecular differences in the the extracellular matrix of tissues that was not previously analyzed by p-SHG microscopy. We obtained χ tensor element ratio values associated with the relevant constituents in tissues. We also found that even for the same biomolecule, there is a difference in the value of χ tensor element ratio for different tissue types. Hence it is important to have a prior knowledge of the χ-value for a biomolecule in a specific tissue, before using it as an optical diagnostic parameter in pathological analysis. These values of χ tensor element ratios obtained from p-SHG microscopy may appear as potential bio-marker to detect any alterations or modification in response to a pathological condition. We successfully demonstrated that early stage modification due to osteoarthritis in the extracellular matrix of cartilage may be detected by p-SHG microscopy, which would typically appear only at advance stage of disease in histological assessment. We performed the tensor susceptibility ratio analysis in backscattered image mode. This demonstrates the potential of the technique for in vivo diagnosis by the application of a miniaturized multiphoton microscopy probe in various clinical applications e.g., fibrosis, wound healing, cancer detections etc. Moreover, the physical model of collagen fiber can be extended further to find the information about any change in helical pitch angle of the fiber, a feature which provides the information beyond the limit of optical resolution [41,
www.biophotonics-journal.org
J. Biophotonics 8, No. 9 (2015)
737
Figure 6 Osteoarthritic cartilage tissue (a) SHG image [38] (b) Histogram plot of χ1 value for ROI 1 (c) Histogram plot of χ1 value for ROI 2 (d) Histogram plot of χ1 value for ROI 3.
42]. In this report, we also discussed the issues that can create artifact in susceptibility (χ) image, and, provided remedies to overcome the problems by applying relevant calibrations during experimental measurements, applicable for a commerical multiphoton system. We have demonstrated that the p-SHG technique has the potential to become a biomedical diagnostic tool, which can further enhance the specificity of standard SHG technique and provide minimally invasive and label free detection of SHG harmonophores in the extracellular matrix of ovary, heart and musculoskeletal tissues. Acknowledgements We are pleased to acknowledge Kristin G. Sæterbø and Astrid Bjørkøy for their assistance in the laboratory.
References [1] F. Helmchen and W. Denk, Nature methods 2(12), 932–940 (2005). [2] P. Friedl, K. Wolf, G. Harms, and U. H. von Andrian, Biological Second and Third Harmonic Generation Microscopy (John Wiley and Sons, Inc., 2001). [3] T. Abraham, J. A. Hirota, S. Wadsworth, and D. A. Knight, Pulmonary Pharmacology and Therapeutics 24(5), 487–496 (2011).
www.biophotonics-journal.org
[4] W. Mohler, A. C. Millard, and P. J. Campagnola, Methods 29(1), 97–109 (2003). [5] K. Schenke-Layland, Journal of Biophotonics 1(6), 451–462 (2008). [6] M. Rivard, K. Popov, C. A. Couture, M. Laliberte, A. Bertrand-Grenier, F. Martin, H. Pepin, C. P. Pfeffer, C. Brown, L. Ramunno, and F. Legare, Journal of Biophotonics 7(8), 638–646 (2014). [7] A. T. Yeh, M. J. Hammer-Wilson, D. C. VanSickle, H. P. Benton, A. Zoumi, B. J. Tromberg, and G. M. Peavy, Osteoarthritis and Cartilage 13(4), 345–352 (2005). [8] W. L. Chen, T. H. Li, P. J. Su, C. K. Chou, P. T. Fwu, S. J. Lin, D. Kim, P. T. C. So, and C. Y. Dong, Applied Physics Letters 94(18), 183902 (2009). [9] S. W. Chu, S. Y. Chen, G. W. Chern, T. H. Tsai, Y. C. Chen, B. L. Lin, and C. K. Sun, Biophysical Journal 86 (6), 3914– 3922 (2004). [10] P. J. Su, W. L. Chen, T. H. Li, C. K. Chou, T. H. Chen, Y. Y. Ho, C. H. Huang, S. J. Chang, Y. Y. Huang, H. S. Lee, and C. Y. Dong, Biomaterials 31(36), 9415–9421 (2010). [11] S. V. Plotnikov, A. C. Millard, P. J. Campagnola, and W. A. Mohler, Biophysical Journal 90(2), 693–703 (2006). [12] T. Hompland, A. Erikson, M. Lindgren, T. Lindmo, and C. de Lange Davies, Journal of Biomedical Optics 13(5), 054050-1–054050-11 (2008).
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
738
R. Kumar et al.: Second harmonic generation microscopy for optical diagnosis
[13] R. Ambekar, T. Y. Lau, M. Walsh, R. Bhargava, and K. C. Toussaint, Biomed. Opt. Express 3(9), 2021–2035 (2012). [14] P. S. Hu, C. M. Hsueh, P. J. Su, W. L. Chen, V. A. Hovhannisyan, S. J. Chen, T. H. Tsai, and C. Y. Dong, IEEE Journal of Selected Topics in Quantum Electronics 18(4), 1326–1334 (2012). [15] P. Campagnola, Analytical Chemistry 83(9), 3224– 3231 (2011). [16] S. Tang, W. Jung, D. McCormick, T. Xie, J. Su, Y.-C. Ahn, B. J. Tromberg, and Z. Chen, Journal of Biomedical Optics 14(3), 034005-1–034005-7 (2009). [17] L. Fu, A. Jain, C. Cranfield, H. Xie, and M. Gu, Journal of Biomedical Optics 12(4), 040501-1–040501-3 (2007). [18] J. C. Jung and M. J. Schnitzer, Opt. Lett. 28(11), 902– 904 (2003). [19] I. A. Roldan, S. Psilodimitrakopoulos, P. L. Alvarez, and D. Artigas, Opt. Express 18(16), 17209–17219 (2010). [20] S. Psilodimitrakopoulos, D. Artigas, G. Soria, I. A. Roldan, A. M. Planas, and P. L. Alvarez, Opt. Express 17(12), 10168–10176 (2009). [21] S. Psilodimitrakopoulos, S. Santos, I. Amat-Roldan, A. K. N. Thayil, D. Artigas, and P. Loza-Alvarez, J. Biomedical Optics 14(1), 014001-1–014001-11 (2009). [22] P. S. Hu, A. Ghazaryan, V. Hovhannisyan, S. J. Chen, Y. F. Chen, C. S. Kim, T. H. Tsai, and C. Y. Dong, Journal of Biomedical Optics 18(3), 031102 (2012). [23] X. Chen, O. Nadiarynkh, S. Plotnikov, and P. J. Campagnola, Nature Protocols 7, 654–669 (2012). [24] T. Yasui, Y. Takahashi, S. Fukushima, Y. 0gura, T. Yamashita, T. Kuwahara, T. Hirao, and T. Araki, Opt. Express 17(2), 912–923 (2009). [25] C. K. Chou, W. L. Chen, P. T. Fwu, S. J. Lin, H. S. Lee, and C. Y. Dong, Journal of Biomedical Optics 13(1), 014005-1–014005-7 (2008). [26] B. Richards and E. Wolf, Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences 253(1274), 358–379 (1959). [27] E. Yew and C. Sheppard, Optics Communications 275 (2), 453–457 (2007). [28] S. Saha, P. Ghosh, D. Mitra, S. Mukherjee, S. Bhattacharya, and S. S. Roy, Cell Physiol Biochem. 19(1–4), 67–76 (2007). [29] C. Berkholtz, B. Lai, T. K. Woodruff, and L. D. Shea, Histochemistry and Cell Biology 126(5), 583–592 (2006). [30] F. Vanzi, L. Sacconi, R. Cicchi, and F. S. Pavone, Journal of Biomedical Optics 17(6), 060901-1–060901-8, (2012). [31] B. Lopez, A. Gonzalez, and J. Diez, Circulation 121 (14), 1645–1654 (2010). [32] B. L. Salazar, S. R. Albniz, T. A. Guedn, A. G. Miqueo, R. Querejeta, and J. D. Martnez, Rev. Esp. Cardiol. 59(10), 1047–1057 (2006). [33] H. A. Wieland, M. Michaelis, B. J. Kirschbaum, and K. A. Rudolphi, Nature Reviews Drug Discovery 4, 331–344 (2005). [34] A. Fraser, U. Fearon, R. C. Billinghurst, M. lonescu, R. Reece, T. Barwick, P. Emery, A. R. Poole, and D. J. Veale, Arthritis & Rheumatism 48(11), 3085–3095 (2003).
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[35] A. R. Poole, M. Kobayashi, T. Yasuda, S. Laverty, F. Mwale, T. Kojima, T. Sakai, C. Wahl, S. El-Maadawy, G. Webb, E. Tchetina, and W. Wu, Annals of the Rheumatic Diseases 61(2), ii78–ii81 (2002). [36] M. R. Tsai, C. H. Chen, and C. K. Sun, Proc. SPIE 7183, 71831V-1–71831V-10 (2009). [37] T. A. Theodossiou, C. Thrasivoulou, C. Ekwobi, and D. L. Becker, Biophysical Journal 91(12), 4665–4677 (2006). [38] R. Kumar, K. M. Grønhaug, E. l. Romijn, J. O. Drogset, and M. B. Lilledahl, Proc. SPIE 9129, 91292Z-1– 91292Z-9 (2014). [39] R. Kleemann, D. Krocker, A. Cedraro, J. Tuischer, and G. Duda, Osteoarthritis and Cartilage, 13(11), 958– 963 (2005). [40] R. E. Outerbridge, Journal of Bone & Joint Surgery, British Volume 43(B4), 752–757 (1961). [41] I. Amat-Roldan, S. Psilodimitrakopoulos, E. Eixarch, I. Torre, B. Wotjas, F. Crispi, F. Figueras, D. Artigas, P. Loza-Alvarez, and E. Gratacos, Proc. SPIE 7367, 73670O-1–73670O-7 (2009). [42] J. Bella, M. Eaton, B. Brodsky, and H. M. Berman, Science 266( 5182), 75–81 (1994).
Appendix A Linear excitation polarization measurement In order to perform linear excitation polarization measurement to create χ-image, sample should be excited by linearly polarized (LP) light. Due to presence of dichroic mirror and other scan optics along the optical path in the microscope, certain ellipticity is introduced during the propagation, and, hence light is no longer linearly polarized at sample plane. To compensate this ellipticity, an additional precompensation introduced by insertion of a quarter wave (λ/4) plate in the optical path. λ/4-plate produces certain negative ellipticity to compensate the introduced ellipticity during propagation of beam along the optical path, and hence, beam arriving at sample plane becomes LP light. To confirm that light arrived at sample plane is LP light, following two measurements can be performed:
Measurement 1 Polarization beam splitter (PBS) is kept at focus and power is measured by powermeter along the two arms of the output of PBS (Figure 7). In general, PBS splits randomly polarized light into two orthogonal linearly polarized components. In this specific case, since incoming beam is already linearly polar-
www.biophotonics-journal.org
J. Biophotonics 8, No. 9 (2015)
739
Figure 7 A schematic diagram of experimental setup for the confirmation of light beam arriving at sample plane is linearly polarized by the use of polarization beam splitter (PBS). LP: Linear polarizer, λ/4-plate: Quarter wave plate, λ/2-plate: Half wave plate.
ized light, and hence, output power along one of the arm of the PBS is expected near to all incident beam power (100%), while, output power along the other arm is expected near to zero. Such output result displayed by the powermeter (Figure 7), confirms that incoming beam is linearly polarized at sample plane and ellipticity compensation is perfect.
Once linearly polarized beam at sample plane is confirmed; beam can be rotated then by the use of λ/2plate and several polarization images at different angles of excitations can be captured.
Appendix B Measurement 2 Another cross-check can be performed to confirm whether the light beam arriving at sample plane is linearly polarized or not. It can be done by the use of another linear analyzer. An analyzer is placed in the sample plane and the output power passing through the analyzer, is measured by powermeter (Figure 8). During the rotation of analyzer, at a certain position, it will reach at cross position with respect to incoming linearly polarized beam. In this case the output power detected by powermeter should be zero. Such measurement, again, confirms that incoming beam is linearly polarized in sample plane.
Table 1 A measurement for the scrambling of linear polarization due to presence of microscope objective. The ratio of intensity count during parallel and cross position of LA, with respect to the linearly polarized excitation light (Figure 3), indicates that how much linearly polarized light is scrambled due to presence of microscope objective. NA of microscope objective
Scrambling of polarization (%)
0.8 (40X) 1.2 (40X) 1.4 (63X)
25.96 34.54 38.18
Figure 8 A schematic diagram of experimental setup for the confirmation of light beam arriving at sample plane is linearly polarized by the use of linear analyzer. LP: Linear polarizer, λ/4-plate: Quarter wave plate, λ/2-plate: Half wave plate.
www.biophotonics-journal.org
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
FULL ARTICLE
Polarization second harmonic generation microscopy provides quantitative enhanced molecular specificity for tissue diagnostics Rajesh Kumar*, 1, Kirsten M. Grønhaug2, Elisabeth I. Romijn1, Andreas Finnøy1, Catharina L. Davies1, Jon O. Drogset3, and Magnus B. Lilledahl1 1
Department of Physics, Norwegian University of Science and Technology (NTNU), N-7491, Trondheim, Norway Levanger Hospital, Norway 3 Trondheim University Hospital, Norway 2
Received 30 July 2014, revised 29 September 2014, accepted 6 October 2014 Published online 4 November 2014
Key words: polarization second harmonic generation (p-SHG) microscopy, collagen fiber, susceptibility (χ) tensor imaging, nonlinear optical imaging, multiphoton microscopy, osteoarthritis, human knee articular cartilage, biomedical optical diagnosis
Due to specific structural organization at the molecular level, several biomolecules (e.g., collagen, myosin etc.) which are strong generators of second harmonic generation (SHG) signals, exhibit unique responses depending on the polarization of the excitation light. By using the polarization second harmonic generation (p-SHG) technique, the values of the second order susceptibility components can be used to differentiate the types of molecule, which cannot be done by the use of a standard SHG intensity image. In this report we discuss how to implement p-SHG on a commercial multiphoton microscope and overcome potential artifacts in susceptibility (χ) image. Furthermore we explore the potential of p-SHG microscopy by applying the technique to different types of tissue in order to determine corresponding reference values of the ratio of second-order χ tensor elements. These values may be used as a bio-marker to detect any structural alterations in pathological tissue for diagnostic purposes.
1. Introduction Second harmonic generation (SHG) is a nonlinear optical process, in which two photons of the same frequency interact with a nonlinear material and generate another photon having twice the energy of * Corresponding author: e-mail: [email protected] © 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
The SHG intensity image (red) in (a) shows the distribution of collagen fibers in ovary tissue but cannot determine the type of collagen fiber. However, the histogram distribution (b) for the values of the χ tensor element ratio can be used to quantitatively identify the types of collagen fibers.
the incident photon. A few fibrillar structures in biological tissues like fibers of collagen, myosin and tublin fulfill the non-centrosymmetric structural condition and produce enough SHG photons to create an intensity based detectable SHG signal [1–6]. SHG signals are produced by endogenous structural pro-
J. Biophotonics 8, No. 9 (2015)
teins in tissue [4]. However, different molecular sources of SHG signals cannot be distinguished in a standard SHG intensity image. In several medical diagnostic applications, differentiation among the constituents of the tissue is an essential requirement. For example, during the progression of osteoarthritis, hyaline cartilage (mainly collagen type II) is being replaced by fibrocartilage (mainly type I collagen) [7]. In such case, it is important to detect any early stage modifications and identify different types of collagen fiber in the extracellular matrix of articular cartilage. By incorporation of polarization sensitivity in SHG images, it is possible to identify the different molecular sources of SHG scatters (e.g. collagen I, collagen II, myosin etc.) based on their susceptibility value (χ) as a quantitative parameter [8]. Incorporation of polarization information in SHG images can be accomplished by excitation polarization measurement in polarization second harmonic generation (p-SHG) microscopy. This technique provides an image of the sample based on the values of the susceptibility components of tissue constituents. The image obtained from p-SHG technique, which shows the morphology of the tissue based on susceptibility values, is referred to as a χ-image. In a χ-image, susceptibility values can be visualized using a false color map. A localized molecular inhomogeneities in the tissue can change susceptibility values. Therefore a change in local environment may appear as different colors in the χ-image. Such differentiation of different tissue constituents is not possible from an SHG intensity image alone [9–11]. Therefore p-SHG can potentially be useful in medical diagnosis. In last few years, p-SHG technique have been used to investigate a few biomedical problems e.g. breast cancer detection [12, 13] and pathological skin states [14]. Characterization of muscle [11] and engineered cartilage tissue has also been performed [14]. Collagen in extracellular matrix is highly altered in several pathological cases like connective tissue diseases, cardiovascular diseases, autoimmune diseases, ovarian diseases etc. [15]. p-SHG microscopy may quantitatively detect modifications at the molecular level and differentiate types of fibrillar collagen inside the extracellular matrix of the tissue. However, the application of p-SHG technique has been very limited to date and has not explored in many other tissue organs. It has to be noted that the value of the χ parameter for a biomolecule may change in different tissue types. Hence it is essential to obtain the prior knowledge of the χ parameter value for a specific tissue type, before applying it as an optical diagnostic parameter in pathological analysis. In this report, to the knowledge of the authors, p-SHG analysis for measuring χ-parameters value of heart muscle, ovary tissue and osteoarthritic cartilage, was performed for the first time. We successfully demonstrate that pSHG technique can detect modifications in human ar-
www.biophotonics-journal.org
731
ticular cartilage in the early stages of osteoarthritis, which in histology generally appears at the advanced stages of the disease. Our results indicate that pSHG has the capability of differentiating various sources of SHG at the molecular level and show potential for a new diagnostic tool in pathological analysis. Moreover, all our SHG measurements were performed in the backscatter imaging mode. This demonstrates that the technique is compatible with in vivo investigation by the use of an endoscopic probe, which necessarily uses the backscattered signal. Miniaturized multiphoton probes for endoscopic applications have been demonstrated by several groups [16–18]. We anticipate that the proof-of-concept results reported in this paper may provide a basis to apply p-SHG technique in diagnosis of various diseases relevant with cardiovascular, ovary and cartilage disorders. Moreover, in this report, we described the issues responsible for image artifacts in p-SHG analysis and provided the relevant solutions in terms of calibrations, which were not discussed previously.
2. Materials and methods 2.1 Background The physical model associated with collagen fibers has previously been described [8, 19–21]. Briefly, under the assumption of cylindrical symmetry for collagen fibers, that the sample lies in xz-plane (Figure 1) and, the propagation direction of the excitation laser beam is along the y direction (out-of-plane); the incoming SHG signal intensity I from the harmonophore can be modeled as ( 2 χzzz 2 2 I¼c sin ðθu θv Þ þ cos ðθu θv Þ χzxx ) 2 χxzx 2 þ sin 2ðθu θv Þ ð1Þ χzxx
Figure 1 Schematic diagram of coordinate system. The direction of the polarization of the laser beam and the orientation of collagen fibers are indicated.
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
732
R. Kumar et al.: Second harmonic generation microscopy for optical diagnosis
The above equation can be used to fit the signal intensity I as a function of difference in angle between the polarization of the incoming light (θu) and the orientation of the collagen fiber (θv), as well as the second-order susceptibility tensor ratios χzzz/χzxx χ1 and χxzx/χzxx χ2. By applying input parameter I and θu to the physical model, the numerical value of other variables can be found. The obtained values χ1 and χ2, provide a quantitative parameter which can be used for the purpose of biomedical diagnosis.
2.2 Experimental The experimental analysis is based on the excitation polarization measurement technique [8, 22, 23]. A standard Zeiss microscope was modified to achieve required experimental measurements. Although a few reports relevant with p-SHG microscopy are available [22–25], the difficulties involved in modification of a standard commercial microscope for development of a p-SHG optical system, have not been discussed previously. If ignored, polarization dependent effects introduced by the microscope can create artifact in image, which ultimately may lead to misleading results. In commercial microscope, there are several optical elements present along the path of the excitation beam, which are not specially designed for optical polarization measurements. Due to presence of such optical elements, χ-image obtained from p-SHG analysis is prone to artifacts. Such problems can be resolved by the implementation of appropriate calibration procedures (described in next subsections). Another issue is that if a commercial system is modified for p-SHG application, then, due to insertion of new optical elements (e.g., polarizers, waveplates etc.) in the optical path, other types of microscopy measurements can be affected. To overcome this problem, a portable optical module (Figure 2) was incorporated in a standard commercial system in such a way that it can be easily removed from the optical path, after the p-SHG measurements, and, hence the complete microscope systems
Figure 2 A photograph of the portable polarization optics module. LP: Linear polarizer, λ/4-plate: Quarter wave plate, λ/2-plate: Half wave plate.
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
again available for other kind of experiments without any modification of the optical system.
2.2.1 Excitation resolved p-SHG A commercial Zeiss LSM 510 META microscope was modified by the insertion of the polarization optics module in the path of the Ti : Sapphire (Coherent Mira) laser to perform excitation polarization measurements. A simplified schematic diagram of the experimental setup is shown in Figure 3. For the excitation of the tissue sample, 200 fs laser pulses with a repetition rate of 76 MHz was tuned to 790 nm. A band pass filter (395 ± 25 nm) was used to select SHG signal, while a combination of dichroic mirror and short pass filter was used to reject backscattered laser light. Inside the polarization optics module, the collimated laser beam was allowed to pass through the linear polarizer (Thorlabs, LPNIR050) to create a pure linearly polarized beam suitable for excitation. Sample was then illuminated by different angle of excitation ranging from 0° to 360°. The rotation of linearly polarized light was performed by the rotation of half wave plate (Thorlabs, λ/2 : 690–1200 nm). A pre-compensation in ellipticity (as described in next subsection) was achieved by a quarter wave plate (Thorlabs, λ/4 : 690–1200 nm). There are a few important issues, discussed below, which have to be considered during excitation resolved p-SHG measurements to avoid potential errors.
2.2.2 Calibration of pre-compensation for each excitation polarization angle One of the most important requirements for excitation polarization measurement is that the sample has to be illuminated by a perfect linearly polarized light. A high quality linearly polarized beam can be achieved by the use of a linear polarizer (LP). Even though high quality linearly polarize beam is created by the use of a LP, ellipticity is introduced during the propagation of light through scanning optics and a dichroic mirror in the optical path. In general, the optical elements present in commercial microscope are not optimize to maintain the polarization of the light beam. Hence, light arrived at sample plane is no longer a linearly polarized light. It is important to mention that the introduced ellipticity in the optical beam is different for a different angle of linear excitation. Hence to avoid any artifact in a χ-image, appropriate pre-compensation is required for each angle of linear excitation. The required pre-compensation can be achieved by a quarter wave plate in the optical path for each angle of linear excitation. In www.biophotonics-journal.org
J. Biophotonics 8, No. 9 (2015)
733
Figure 3 A simplified schematic diagram of the experimental setup for p-SHG microscopy. A χ-image is created by performing excitation polarization measurements with the use of the polarization optics module, which is shown inside the dottedline box. LP: Linear polarizer, LA: Linear analyzer, SP: Short pass, λ/4-plate: Quarter wave plate, λ/2-plate: Half wave plate, SHG: Second harmonic generation, TPEF: Two-photon excited fluorescence.
order to confirm that the light arrived at sample plane is linearly polarized light, additional experimental measurements can be performed, which is described in appendix A. Hence, by the introduction of an additional negative pre-compensation in order to compensate the introduced ellipticity for each corresponding angle of linear excitation, a calibration data for the specific microscope is obtained, which can be used during the excitation polarization measurements.
2.2.3 Optical power calibration Another important issue arises due to polarization dependent reflectivity of the dichroic mirror present in the optical path of the microscope. The actual excitation power of laser light arriving at the sample plane may change at a different angle of linear exci-
www.biophotonics-journal.org
tations because the presence of dichroic mirror. Any fluctuation in the laser power due to involved optics in the optical path of microscope may, cause additional errors in the result. Hence, in order to compensate the change in optical power arriving at sample plane, an adjustment of optical power for each corresponding angle of linear excitation provides another calibration, which should be employed during the excitation polarization measurements.
2.2.4 Polarization scrambling To perform SHG analysis high numerical aperture (NA) microscope objective is preferred, because, it provides high density of photons at focal point, which is an essential requirement to achieve nonlinear optical phenomena. On the other hand, high NA objective can scramble the polarization state of the excita-
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
734
R. Kumar et al.: Second harmonic generation microscopy for optical diagnosis
tion light. In a theoretical analysis it is mentioned that a lower NA microscope objective is desired to maintain the polarization property of light at focal volume of microscope objective [26]. Hence it is suggested to use a moderate NA objective for p-SHG measurements. To the knowledge of the authors there is no quantitative result available regarding the scrambling of linear polarization by high NA microscope objectives. A measurement was performed to quantify the scrambling of linear polarization, only due to the involvement of microscope objective in the optical path. Linearly polarized light is passed through a linear analyzer (LA) in trans-illumination configuration (Figure 3) through the microscope objective of different NA. The ratio of intensity count during parallel and cross position of LA, with respect to the linearly polarized excitation light, indicates the amount of scrambling caused by the microscope objective. The results from this measurement is shown in appendix B. Moderate NA microscope objective (e.g., 0.8 NA) is advisable to perform excitation resolved p-SHG measurements [8, 27].
2.3 Sample preparation The use of human tissue (osteoarthritic cartilage) in this study was approved by the regional committee for medical research ethics (2013/265 REK, Norway). Slides of mouse tissue (ovary and heart) was purchased from Abgent Europe Ltd. Paraffin embedded tissue sections (thickness 5 μm) was deparaffinized by dipping slides in two separate xylene solution, each dip for 5 minutes. Dehydration of the slides was performed through a series of graded alcohol 96%, 86% and 70% for a duration of 5 minutes, 4 minutes and 3 minutes, respectively. Slides were finally rehydrated in water for a duration of 1 minute. Tissue sections were then covered by a standard cover slip. Edges of the cover slip were sealed with vaseline to avoid any dehydration of the tissue sections. All slides were stored at 4 °C between measurements. In order to demonstrate the potential of the technique, in this proof-of-concept study, one tissue slide of each tissue type (ovary, heart and cartilage) was analyzed. In every tissue slide each measurement was performed three times; additionally, at two different locations. Similar results were observed.
2.4 Image acquisition and data analysis In order to minimize the effect of birefringence due to tissue sample, all SHG images were acquired at 3 μm below the surface of tissue. The sample was © 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
excited by equally spaced, 11 different angles of excitations, ranging from 0° to 360°. Each angle of excitation provides an image. Hence, in total, we have 11 images per sample, which are polarizationresolved SHG images. Data analysis was performed in Matlab (The Mathworks, 2014). By fitting Eq. (1) to the measurements, the χ-tensor element ratios can be determined at every pixel of the images. The known value is intensity (I), and excitation angle (θu), while other variables can be found performing least square fit to the model. The peak value of Gaussian fit into the histogram plot of susceptibility tensor components provides the value of χ1 and χ2, which is a characteristic value of the tissue. All Gaussian fit have been expressed with 95% confidence bounds.
3. Result and discussion 3.1 Ovary Ovary tissue is considered as quite dynamic tissue where remodeling process inside the matrix of tissue occurs continuously [28]. In general, collagen I, III and IV are major types of collagen present inside the tissue matrix. The relative quantification of different types of collagen is very important in understanding of various reproductive disorders [28, 29]. It is worth to mention that collagen type IV is a nonfibrillar collagen, and, hence it does not produce SHG signal. Therefore, p-SHG technique cannot image collagen IV. In this section, we are presenting p-SHG as a potential method for relative quantification of the content of different types of fibrillar collagen in ovary tissue. In Figure 4a, the SHG signal (red color) shows the distribution of the collagen fibers, while- autofluorescence (green color) represents other components present inside the ovary tissue. Although the intensity based SHG image can clearly distinguish collagen fibers from other tissue components (as shown in the Figure 4a), it cannot differentiate the types of collagen molecule present in extracellular matrix of the tissue. To identify collagen types in ovary tissue, p-SHG analysis was performed. The distribution of the histogram for the value of χ1 and χ2, are shown in the Figure 4b and 4c respectively. The Gaussian fit to the histogram plot of χ1 results in two peaks, as shown in Figure 4b. We identify that the larger peak at χ1 = 1.20 (1.999, 1.200) is associated with collagen I, as this value is close to the earlier reported value of collagen I in tendon [8]. The other peak is at χ1 = 0.83 (0.835, 0.836) is very close to the value of collagen III reported earlier [8]. Hence we identify that the smaller peak, in Figure 4b represents an-
www.biophotonics-journal.org
J. Biophotonics 8, No. 9 (2015)
735
Figure 4 Ovary tissue (a) SHG (red) and autofluoroscence (green) image (b) Histogram plot of χ1 value (c) Histogram plot of χ2 value (d) χ1 image.
other major component of fibrillar collagen found in ovary i.e., collagen III. The Gaussian fit to the distribution of χ2, as shown in Figure 4c, shows one peak at χ2 = 0.38 (0.384, 0.389). The appearance of only one peak indicates that χ2 is not able to create additional contrast in the χ-image. The reason is that parameter χ2 is more sensitive to SHG intensity fluctuations, as we can see in Eq. (1), so less accurate in value. Hence χ1 has a better diagnostic capability. However, to find the comparative suitability between χ1 and χ2 in order to determine the diagnostic capability, in general, a separate analysis is recommended for a specific disease associated with a tissue. The χ1 image of the ovary tissue, which is obtained by providing the false color to the value of χ tensor element ratio, is shown in the Figure 4d. The image indicates the capability of χ image in visualizing any relevant modification in ECM of tissue, which may appear by the variation in color. Moreover, by calculation of the ratio of Gaussian area (i.e., collagen I/collagen III) from the Figure 4b, as described in previous report [8], it is possible to find a relative estimate for the different types of collagen content quantitatively inside the ovary tissue. Evaluation of such estimate may play an important role in revealing hidden features of several biophysical dynamic process associated with ovary tissue [28, 29].
3.2 Heart (cardiac muscle) One of the main constituent of muscle tissue is myosin. It is endowed with extraordinarily long α-helices www.biophotonics-journal.org
which are highly aligned, especially in the tail portion and hence responsible for coherent SHG emission [30]. Structural modificationin cardiac muscle fiber can be an early biomarker in several cardiac dysfunctions [31, 32]. Due to any such alterations in cardiovascular tissue the characteristic value of χ1 may change, and, hence p-SHG microscopy potentially able to detect subtle structural differences at the level of myosin architecture, which would not be possible to observe directly by typical histology due to limited resolution. Although χ1 value of myosin (from chicken wing muscle) can be found in Ref. [8], it was not analyzed and reported for heart muscle yet. Figure 5a shows the SHG intensity image of the cardiac muscle fiber, which was obtained from a mouse heart. Based on p-SHG analysis, we obtained the χ tensor element ratio, χ1 and χ2, of cardiac muscle, which are shown in the Figure 5b and 5c. respectively. The χ1-image of heart muscle, which is obtained by providing the false color to the value of χ tensor element ratio, is shown in the Figure 5d. This image indicates that the value of χ tensor element ratio is capable in visualizing the morphology of tissue and hence any alteration in structure may appear by the variation in color. The Gaussian fit to the histogram distribution provides χ1 = 0.86 (0.859, 0.861) and χ2 = 0.50 (0.500, 0.514), as shown in the Figure 5b and 5c respectively. It is worth to note that the value of χ1 obtained for the heart muscle is slightly different from the earlier reported [8] value of myosin as 1.04 ± 0.08 for the chicken wing muscle. We hypothesized that the difference in value is caused by structural differences in the myosin assembly. Hence this analysis shows the importance of prior knowledge of χ-value, which is © 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
736
R. Kumar et al.: Second harmonic generation microscopy for optical diagnosis
by International Cartilage Repair Society (ICRS) [39, 40], was chosen for the p-SHG analysis. Figure 6a shows intensity based SHG image of osteoarthritic cartilage. The characteristic value of χ1 was evaluated for three diffrent regions of interst ROI 1, ROI 2 and ROI 3, which represent fibrocartilage, hyaline cartilage and composite cartilage respectively. The Gaussian fit to the histogram provides χ1 = 1.20 (1.201, 1.206) and χ1 = 1.07 (1.069, 1.072) for ROI 1 (Figure 6b) and ROI 2 (Figure 6c) respectively. The histogram plot for the value of χ1 associated with the ROI 3 (Figure 6d) shows two different Gaussian peaks, as expected, at 1.21 (1.210, 1.227) and 1.10 (1.100, 1.106), which represents fibrocartilage and hyaline cartilage, respectively. The two different values of χ1 in ICRS Grade I cartilage sample, illustrate the molecular modification in the cartilage matrix, from hyaline to fibrocartilage, even in early stage of OA. This is believed to be visible only at advance stages of disease in histological assessment. This result shows the capability of p-SHG technique in early diagnosis of OA. Figure 5 Heart muscle (a) SHG image (b) Histogram plot of χ1 value (c) Histogram plot of χ2 value (d) χ1 image.
necessarily required for a specific tissue type, before applying it as an optical diagnostic parameter in pathological analysis.
3.3 Osteoarthritic cartilage Collagen is one of the major constituents of the extracellular matrix of cartilage [33]. Since p-SHG is capable in differentiation of different types of collagen fiber, we applied this technique in a clinically relevant study of osteoarthritis (OA). It is believed that during progress of OA the quality of cartilage continuously degrades. It has been reported that during the progression of the disease, replacement of collagen fibers takes place in the extra cellular matrix of cartilage [34, 35]. The type of cartilage in repair tissue is not generally hyaline cartilage but fibrocartilage [7, 36]. It is important to differentiate types of cartilage because fibrocartilage is not as good as hyaline cartilage in terms of mechanical strength, stability and adaptability. Hyaline cartilage is composed mainly of collagen II, while, fibrocartilage mainly of collagen I. Hence, in general, fibrocartilage is expected to yield relatively more intense SHG image in comparision with hyaline cartilage [36, 37], as collagen I fibrils are generally thicker and therefore emits stronger SHG signal. We applied p-SHG method on osteoarthritic human articular cartilage obtained from the knee. Early stage osteoarthritic cartilage, Grade-I, defined
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
4. Conclusion In this study, we explored the application of p-SHG microscopy for probing the molecular differences in the the extracellular matrix of tissues that was not previously analyzed by p-SHG microscopy. We obtained χ tensor element ratio values associated with the relevant constituents in tissues. We also found that even for the same biomolecule, there is a difference in the value of χ tensor element ratio for different tissue types. Hence it is important to have a prior knowledge of the χ-value for a biomolecule in a specific tissue, before using it as an optical diagnostic parameter in pathological analysis. These values of χ tensor element ratios obtained from p-SHG microscopy may appear as potential bio-marker to detect any alterations or modification in response to a pathological condition. We successfully demonstrated that early stage modification due to osteoarthritis in the extracellular matrix of cartilage may be detected by p-SHG microscopy, which would typically appear only at advance stage of disease in histological assessment. We performed the tensor susceptibility ratio analysis in backscattered image mode. This demonstrates the potential of the technique for in vivo diagnosis by the application of a miniaturized multiphoton microscopy probe in various clinical applications e.g., fibrosis, wound healing, cancer detections etc. Moreover, the physical model of collagen fiber can be extended further to find the information about any change in helical pitch angle of the fiber, a feature which provides the information beyond the limit of optical resolution [41,
www.biophotonics-journal.org
J. Biophotonics 8, No. 9 (2015)
737
Figure 6 Osteoarthritic cartilage tissue (a) SHG image [38] (b) Histogram plot of χ1 value for ROI 1 (c) Histogram plot of χ1 value for ROI 2 (d) Histogram plot of χ1 value for ROI 3.
42]. In this report, we also discussed the issues that can create artifact in susceptibility (χ) image, and, provided remedies to overcome the problems by applying relevant calibrations during experimental measurements, applicable for a commerical multiphoton system. We have demonstrated that the p-SHG technique has the potential to become a biomedical diagnostic tool, which can further enhance the specificity of standard SHG technique and provide minimally invasive and label free detection of SHG harmonophores in the extracellular matrix of ovary, heart and musculoskeletal tissues. Acknowledgements We are pleased to acknowledge Kristin G. Sæterbø and Astrid Bjørkøy for their assistance in the laboratory.
References [1] F. Helmchen and W. Denk, Nature methods 2(12), 932–940 (2005). [2] P. Friedl, K. Wolf, G. Harms, and U. H. von Andrian, Biological Second and Third Harmonic Generation Microscopy (John Wiley and Sons, Inc., 2001). [3] T. Abraham, J. A. Hirota, S. Wadsworth, and D. A. Knight, Pulmonary Pharmacology and Therapeutics 24(5), 487–496 (2011).
www.biophotonics-journal.org
[4] W. Mohler, A. C. Millard, and P. J. Campagnola, Methods 29(1), 97–109 (2003). [5] K. Schenke-Layland, Journal of Biophotonics 1(6), 451–462 (2008). [6] M. Rivard, K. Popov, C. A. Couture, M. Laliberte, A. Bertrand-Grenier, F. Martin, H. Pepin, C. P. Pfeffer, C. Brown, L. Ramunno, and F. Legare, Journal of Biophotonics 7(8), 638–646 (2014). [7] A. T. Yeh, M. J. Hammer-Wilson, D. C. VanSickle, H. P. Benton, A. Zoumi, B. J. Tromberg, and G. M. Peavy, Osteoarthritis and Cartilage 13(4), 345–352 (2005). [8] W. L. Chen, T. H. Li, P. J. Su, C. K. Chou, P. T. Fwu, S. J. Lin, D. Kim, P. T. C. So, and C. Y. Dong, Applied Physics Letters 94(18), 183902 (2009). [9] S. W. Chu, S. Y. Chen, G. W. Chern, T. H. Tsai, Y. C. Chen, B. L. Lin, and C. K. Sun, Biophysical Journal 86 (6), 3914– 3922 (2004). [10] P. J. Su, W. L. Chen, T. H. Li, C. K. Chou, T. H. Chen, Y. Y. Ho, C. H. Huang, S. J. Chang, Y. Y. Huang, H. S. Lee, and C. Y. Dong, Biomaterials 31(36), 9415–9421 (2010). [11] S. V. Plotnikov, A. C. Millard, P. J. Campagnola, and W. A. Mohler, Biophysical Journal 90(2), 693–703 (2006). [12] T. Hompland, A. Erikson, M. Lindgren, T. Lindmo, and C. de Lange Davies, Journal of Biomedical Optics 13(5), 054050-1–054050-11 (2008).
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
738
R. Kumar et al.: Second harmonic generation microscopy for optical diagnosis
[13] R. Ambekar, T. Y. Lau, M. Walsh, R. Bhargava, and K. C. Toussaint, Biomed. Opt. Express 3(9), 2021–2035 (2012). [14] P. S. Hu, C. M. Hsueh, P. J. Su, W. L. Chen, V. A. Hovhannisyan, S. J. Chen, T. H. Tsai, and C. Y. Dong, IEEE Journal of Selected Topics in Quantum Electronics 18(4), 1326–1334 (2012). [15] P. Campagnola, Analytical Chemistry 83(9), 3224– 3231 (2011). [16] S. Tang, W. Jung, D. McCormick, T. Xie, J. Su, Y.-C. Ahn, B. J. Tromberg, and Z. Chen, Journal of Biomedical Optics 14(3), 034005-1–034005-7 (2009). [17] L. Fu, A. Jain, C. Cranfield, H. Xie, and M. Gu, Journal of Biomedical Optics 12(4), 040501-1–040501-3 (2007). [18] J. C. Jung and M. J. Schnitzer, Opt. Lett. 28(11), 902– 904 (2003). [19] I. A. Roldan, S. Psilodimitrakopoulos, P. L. Alvarez, and D. Artigas, Opt. Express 18(16), 17209–17219 (2010). [20] S. Psilodimitrakopoulos, D. Artigas, G. Soria, I. A. Roldan, A. M. Planas, and P. L. Alvarez, Opt. Express 17(12), 10168–10176 (2009). [21] S. Psilodimitrakopoulos, S. Santos, I. Amat-Roldan, A. K. N. Thayil, D. Artigas, and P. Loza-Alvarez, J. Biomedical Optics 14(1), 014001-1–014001-11 (2009). [22] P. S. Hu, A. Ghazaryan, V. Hovhannisyan, S. J. Chen, Y. F. Chen, C. S. Kim, T. H. Tsai, and C. Y. Dong, Journal of Biomedical Optics 18(3), 031102 (2012). [23] X. Chen, O. Nadiarynkh, S. Plotnikov, and P. J. Campagnola, Nature Protocols 7, 654–669 (2012). [24] T. Yasui, Y. Takahashi, S. Fukushima, Y. 0gura, T. Yamashita, T. Kuwahara, T. Hirao, and T. Araki, Opt. Express 17(2), 912–923 (2009). [25] C. K. Chou, W. L. Chen, P. T. Fwu, S. J. Lin, H. S. Lee, and C. Y. Dong, Journal of Biomedical Optics 13(1), 014005-1–014005-7 (2008). [26] B. Richards and E. Wolf, Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences 253(1274), 358–379 (1959). [27] E. Yew and C. Sheppard, Optics Communications 275 (2), 453–457 (2007). [28] S. Saha, P. Ghosh, D. Mitra, S. Mukherjee, S. Bhattacharya, and S. S. Roy, Cell Physiol Biochem. 19(1–4), 67–76 (2007). [29] C. Berkholtz, B. Lai, T. K. Woodruff, and L. D. Shea, Histochemistry and Cell Biology 126(5), 583–592 (2006). [30] F. Vanzi, L. Sacconi, R. Cicchi, and F. S. Pavone, Journal of Biomedical Optics 17(6), 060901-1–060901-8, (2012). [31] B. Lopez, A. Gonzalez, and J. Diez, Circulation 121 (14), 1645–1654 (2010). [32] B. L. Salazar, S. R. Albniz, T. A. Guedn, A. G. Miqueo, R. Querejeta, and J. D. Martnez, Rev. Esp. Cardiol. 59(10), 1047–1057 (2006). [33] H. A. Wieland, M. Michaelis, B. J. Kirschbaum, and K. A. Rudolphi, Nature Reviews Drug Discovery 4, 331–344 (2005). [34] A. Fraser, U. Fearon, R. C. Billinghurst, M. lonescu, R. Reece, T. Barwick, P. Emery, A. R. Poole, and D. J. Veale, Arthritis & Rheumatism 48(11), 3085–3095 (2003).
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
[35] A. R. Poole, M. Kobayashi, T. Yasuda, S. Laverty, F. Mwale, T. Kojima, T. Sakai, C. Wahl, S. El-Maadawy, G. Webb, E. Tchetina, and W. Wu, Annals of the Rheumatic Diseases 61(2), ii78–ii81 (2002). [36] M. R. Tsai, C. H. Chen, and C. K. Sun, Proc. SPIE 7183, 71831V-1–71831V-10 (2009). [37] T. A. Theodossiou, C. Thrasivoulou, C. Ekwobi, and D. L. Becker, Biophysical Journal 91(12), 4665–4677 (2006). [38] R. Kumar, K. M. Grønhaug, E. l. Romijn, J. O. Drogset, and M. B. Lilledahl, Proc. SPIE 9129, 91292Z-1– 91292Z-9 (2014). [39] R. Kleemann, D. Krocker, A. Cedraro, J. Tuischer, and G. Duda, Osteoarthritis and Cartilage, 13(11), 958– 963 (2005). [40] R. E. Outerbridge, Journal of Bone & Joint Surgery, British Volume 43(B4), 752–757 (1961). [41] I. Amat-Roldan, S. Psilodimitrakopoulos, E. Eixarch, I. Torre, B. Wotjas, F. Crispi, F. Figueras, D. Artigas, P. Loza-Alvarez, and E. Gratacos, Proc. SPIE 7367, 73670O-1–73670O-7 (2009). [42] J. Bella, M. Eaton, B. Brodsky, and H. M. Berman, Science 266( 5182), 75–81 (1994).
Appendix A Linear excitation polarization measurement In order to perform linear excitation polarization measurement to create χ-image, sample should be excited by linearly polarized (LP) light. Due to presence of dichroic mirror and other scan optics along the optical path in the microscope, certain ellipticity is introduced during the propagation, and, hence light is no longer linearly polarized at sample plane. To compensate this ellipticity, an additional precompensation introduced by insertion of a quarter wave (λ/4) plate in the optical path. λ/4-plate produces certain negative ellipticity to compensate the introduced ellipticity during propagation of beam along the optical path, and hence, beam arriving at sample plane becomes LP light. To confirm that light arrived at sample plane is LP light, following two measurements can be performed:
Measurement 1 Polarization beam splitter (PBS) is kept at focus and power is measured by powermeter along the two arms of the output of PBS (Figure 7). In general, PBS splits randomly polarized light into two orthogonal linearly polarized components. In this specific case, since incoming beam is already linearly polar-
www.biophotonics-journal.org
J. Biophotonics 8, No. 9 (2015)
739
Figure 7 A schematic diagram of experimental setup for the confirmation of light beam arriving at sample plane is linearly polarized by the use of polarization beam splitter (PBS). LP: Linear polarizer, λ/4-plate: Quarter wave plate, λ/2-plate: Half wave plate.
ized light, and hence, output power along one of the arm of the PBS is expected near to all incident beam power (100%), while, output power along the other arm is expected near to zero. Such output result displayed by the powermeter (Figure 7), confirms that incoming beam is linearly polarized at sample plane and ellipticity compensation is perfect.
Once linearly polarized beam at sample plane is confirmed; beam can be rotated then by the use of λ/2plate and several polarization images at different angles of excitations can be captured.
Appendix B Measurement 2 Another cross-check can be performed to confirm whether the light beam arriving at sample plane is linearly polarized or not. It can be done by the use of another linear analyzer. An analyzer is placed in the sample plane and the output power passing through the analyzer, is measured by powermeter (Figure 8). During the rotation of analyzer, at a certain position, it will reach at cross position with respect to incoming linearly polarized beam. In this case the output power detected by powermeter should be zero. Such measurement, again, confirms that incoming beam is linearly polarized in sample plane.
Table 1 A measurement for the scrambling of linear polarization due to presence of microscope objective. The ratio of intensity count during parallel and cross position of LA, with respect to the linearly polarized excitation light (Figure 3), indicates that how much linearly polarized light is scrambled due to presence of microscope objective. NA of microscope objective
Scrambling of polarization (%)
0.8 (40X) 1.2 (40X) 1.4 (63X)
25.96 34.54 38.18
Figure 8 A schematic diagram of experimental setup for the confirmation of light beam arriving at sample plane is linearly polarized by the use of linear analyzer. LP: Linear polarizer, λ/4-plate: Quarter wave plate, λ/2-plate: Half wave plate.
www.biophotonics-journal.org
© 2014 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
