We present three dimensional images of backscattered intensity, and to the best of our knowledge, the first 3D-images of retardation and fast axis orientation of human skin in vivo. The images were recorded with a phase resolved, polarization sensitive optical coherence tomography (OCT) system which is based on a fast transversal scanning of the sample. The three dimensional data sets were obtained by recording several en face images at different depths within the sample. Intensity and retardation images are combined to a 4 dimensional animation to enhance the visualization of the three dimensional data set. The three dimensional information enables a more accurate interpretation of the structural and birefringence information as compared to 2 dimensional B-scans. Birefringence properties of different skin regions are presented and discussed.
©2004 Optical Society of America
Optical coherence tomography (OCT) has developed to a powerful technique to image biological samples with a resolution of a few micrometers . Overviews of this method can be found in references [2, 3]. Conventional OCT is based on intensity measurements and provides structural information of tissue. Polarization sensitive OCT (PS-OCT) extends the concept of OCT and uses the polarization properties of light to gather additional information on the birefringence properties of the sample [4, 5]. Whereas first PS-OCT systems were able to measure backscattered intensity and retardation of a sample by using a polarization sensitive two channel detection unit, a phase sensitive method is required to obtain additional information on the sample (e.g., Stokes vectors, Mueller and Jones matrix elements and birefringent axis orientation) [6, 7, 8, 9]. The birefringence of tissue is often caused by form birefringence due to narrow fibrous structures which cannot be resolved with standard OCT systems. These fibrous structures are present in several kinds of human tissue (e.g. skin, tendon, cornea, retina) [10, 11, 12]. A change in the structural composition can result in a loss of birefringence, which can be used for diagnostic purposes. E.g., the estimation of burn depth of thermally damaged skin has been suggested as an application of this technology [5, 13, 14]. Of special importance to pharmacology of transdermal drugs and cosmetic agents is the stratum corneum which constitutes the major barrier for these agents. Therefore, PS-OCT is an interesting technology for studying this layer and might be of use to gain insight into its behavior under drug application.
In vivo PS-OCT imaging has yet been limited to acquiring of 2 dimensional tomograms. However, birefringence measurements of the cornea  suggested that it might be necessary to collect a 3 dimensional data set to interpret and understand the measured retardation and axis orientation images correctly. Conventional OCT images are obtained by recording several adjacent depth scans (A-scans) resulting in a huge amount of data if a three dimensional phase resolved measurement is performed. Although, fast A-scan based systems have been developed , which can handle this data stream, a reduction of the data volume is preferable to reduce the computational effort and storage space. Alternative scanning schemes based on a transversal scanning of the sample have been introduced to OCT [15, 16, 17], but these techniques were limited to intensity measurements, thus preventing the acquisition of phase information which is necessary to measure the birefringent axis orientation of the sample. Recently, we extended the concept of transversal scanning with a phase resolved, polarization sensitive method and applied it to imaging of human ocular tissue .
In this paper we use this technique to investigate polarization properties of different regions of human skin in vivo. The technique greatly reduces the amount of data compared to other phase resolved techniques. To the best of our knowledge, the first three dimensional images of retardation and birefringent fast axis orientation of human skin in vivo are presented. The three dimensional images enable a more detailed analysis of the local spatial distribution of birefringent structures of human skin and result in new insights into the polarizing properties of skin, especially of the stratum corneum, which are somewhat contradictory to the interpretation reported previously.
The experimental set-up is based on a Mach Zehnder interferometer. A detailed description can be found in reference . A light source (AFC Technologies, Canada) emitting at λ0=1310 nm (Δλ≈55nm, output power=24 mW) is vertically linear polarized before illuminating the interferometer. A quarter wave plate placed in the sample arm with a fast axis orientation of 45° to the horizontal converts the linear polarized light into circular polarized light. The circular polarized light is directed to the sample via an x-y galvo-scanner. The optical power on the sample was 1 mW which is well below the maximum permissible exposure for human skin at 1300nm. Two accousto optic modulators in the reference arm introduce a net frequency shift of 100kHz to the reference beam, which causes an interference signal on the detector of 100kHz in case of path length matching of the sample and reference arms (i.e. the path length difference is within the coherence gate). A translation stage placed in the reference arm enables a change in the reference arm length, which corresponds to a depth shift of the coherence gate within the sample. The light from the sample and reference arms is recombined at the interferometer exit. One advantage of the Mach Zehnder interferometer is the access to two interferometer exits which enables dual balanced detection very easily. The light from both exits is directed to a two channel polarization sensitive detection unit each. The signals of the corresponding polarization channels (horizontal channel, vertical channel) are subtracted, pre-amplified and band pass filtered centered at the carrier frequency. Two lock in amplifiers are used to demodulate each signal separately by multiplying each signal with a reference cosine wave and low pass filtering (corresponding to the demodulated real part of the signal) and multiplying each signal with a reference sine wave and low pass filtering (corresponding to a reconstructed demodulated imaginary part of the signal). Both, real and imaginary parts of each polarization channel were recorded by a data acquisition board. This technique enables a fast and phase sensitive demodulation of the interferometric signal and reduces the amount of data points as compared to other phase sensitive methods. A 3 dimensional data set is obtained by collecting en face images (xy-image plane) consisting of transversal lines at different depth positions in the sample. Figure 1 illustrates the scanning scheme.
From the measured data we can calculate the sample reflectivity R(z), which is given by 
where A1 denotes the amplitude of the horizontal channel, and A2 denotes the amplitude of the vertical channel, respectively. The retardation δ is given by the quotient of the amplitudes 
The fast axis orientation θ is encoded in the phase difference ΔΦ=Φ2-Φ1 between the two channels 
This method allows to measure the cumulative axis orientation of the sample with an unambiguous range of -90° to +90°.
Three skin regions were imaged in healthy volunteers in vivo. For each measurement a 3 dimensional data set consisting of 250×250×100 pixels (x-y-z-direction) and covering a volume of 1.8×2×1mm3 was recorded. The total recording time for each 3D data set was approximately 150sec. Once a 3D data set is recorded, 2D sections of arbitrary geometry can be derived by software.
Figure 2 shows B-scans (x-z) obtained from a 3 dimensional data set of a human fingertip in vivo. For index matching, the finger was pressed against a glass plate. The intensity image, shown in Fig. 2(a), reveals several structures of this skin region (indicated in the image). Figure 2(b) shows the measured retardation of this skin region. Areas where the signal intensity is below a certain threshold (set slightly above the mean noise value) are displayed in gray (because birefringence data from noisy areas are unreliable ). Light backscattered from within the stratum corneum is depolarized which can be clearly seen by the random retardation values in Fig. 2(b) and the random values of the axis orientation in Fig. 2(c). Interesting is the fact that the retardation and fast axis orientation of layers beneath the stratum corneum (consisting of other layers of the epidermis and of the dermis) are clearly defined and remain almost unchanged with depth (the wavy appearance in the lower part of each image is caused by motion artifacts). This specific pattern of the retardation in this skin region is similar to previously reported results . However, an increase of the threshold level (Fig. 2(d) and Fg. 2(e)) removes the backscattered light from within the stratum corneum and reveals retardation and fast axis orientation of small strongly backscattering features (e.g., eccrine sweat glands) that have been obscured before.
On a close look at the sweat gland, one might get the impression of an increasing retardation with depth and a well defined axis orientation at the eccrine sweat gland. However, the B-scan image shows too few image points of the sweat gland (because of the inclination and the spiral structure of the gland) to be sure.
To get a better impression of the birefringence properties of the stratum corneum by investigating the polarization state of light backscattered at the narrow, inclined spiraling sweat glands, a 3D volume rendered view of the 3D data set was constructed by volume rendering software .
Figure 3 shows a 3D animated movie of the volume rendered data set. The movie combines information on the backscattered intensity (corresponding to the opacity) and on the retardation (corresponding to the color coding) of the sample into one 4 dimensional animation. For better visualization, areas below the intensity threshold mentioned above are totally transparent. The increase of retardation with depth within the stratum corneum can be observed along several eccrine sweat glands.
Furthermore, Fig. 3 shows that the region beneath the stratum corneum does not further increase the retardation, as can be seen from the constant color with depth (as mentioned above, the wavy appearance of color boundaries with depth are caused by motion artifacts). Contrary to the light backscattered from within the stratum corneum, the backscattering from within other layers of the epidermis does not depolarize the light (the rather large areas of contiguous constant color indicate that there is no scrambling of polarization states).
To quantify the retardation caused by the stratum corneum we set a threshold (as mentioned above) to obtain only retardation values from strongly backscattering features (eccrine sweat ducts). Because of the spiral structure of the sweat ducts we averaged the retardation values, in an area (220×240µm2) containing a sweat duct, over both lateral directions resulting in a mean phase retardation depth profile of each sweat duct (a total of 6 sweat ducts were investigated). We applied a linear least square fit to the linear part (which corresponded to the length of a sweat duct) of each depth profile and assumed a mean refractive index of skin of 1.4. The measured single pass phase retardation rates were ranging from 0.20deg/µm to 0.30deg/µm with a fitting error of each fit of 0.02 deg/µm.
Figure 4 shows a 3D data set of another fingertip. A movie of transversal (en face) sections corresponding to increasingly deeper positions within the tissue was obtained from the 3D data set. In this measurement, the glass plate was omitted to preserve more clearly the structure of the surface fingerprint (spiral pattern).
In the first few frames the coherence gate was positioned in air outside the skin, hence the first frames are dark. The first bright structure appearing when the coherence gate is moved further in z-direction is the surface relief of the fingerprint followed by the relatively low backscattering stratum corneum. The signal from this layer is randomly polarized as already observed in Fig. 2, the axis orientation varies from speckle to speckle. When the coherence gate is moved deeper into the skin the fingerprint recurs on the interface between the stratum corneum and deeper layers of the epidermis which are highly backscattering. The retardation and fast axis orientation in these layers are well defined and constant in a region much larger than the size of a speckle indicating polarization preserving backscattering. This pattern remains unchanged with depth and a spiral pattern can still be recognized. As the coherence gate is moved further into the skin, upper layers of the dermis become visible and the images of the retardation and fast axis orientation start to blur because of the contribution of multiple scattered light to the signal.
To investigate a completely different skin region we imaged the backside of a human hand. A movie of en face sections was prepared in a similar way as in the case of Fig. 4. As shown in the upper left image of Fig. 5, the stratum corneum in this skin region is much thinner than at the fingertip region. The first frames of the movie appear dark again because the coherence gate is placed outside the tissue. A bright signal slightly above the surface of the skin results from hair follicle. As the coherence gate is moved through the epidermis (the stratum corneum can not be resolved in these images because of its thinness) we observe no or only weak retardation in the backscattered signal which corresponds to the previous measurement (the influence of the stratum corneum in this skin region can be neglected) in the way that layers of the epidermis beneath the stratum corneum are weakly birefringent. No or low retardation implies that the intensity of one polarization channel is very low and will be dominated by noise. Since the calculation of the axis orientation is based on the phase difference measured between the two orthogonal polarization channels (Eq. (3)), the axis orientation will be corrupted by noise in this case (corresponding to a random axis orientation).
As the coherence gate is moved deeper into the tissue we measure backscattered light from within the dermis and observe regions with increasing retardation thus providing also more reliable axis orientation values in these areas. Figure 6 shows B-scans (position of the scan is indicated with a black line in Fig. 5) derived from the three dimensional data set. Clearly visible are the low retardation of the whole epidermis and the increasing retardation in the deeper layers of the dermis which corresponds to previously published results . Noticeable are the rather large regions of constant axis orientation in Fig. 5 and Fig. 6, as compared to observations in the fingertip region, which will be discussed later. To quantify our results we averaged the retardation values of two small representative areas of Fig. 5 (one area with high retardation and one with low retardation) over both lateral directions (300×300µm2) for different depths. The result is a mean retardation depth profile for each area. We applied a linear least square fit to the linear part (~300µm in depth starting from ~100µm below the surface) of the depth profile. The mean single pass phase retardation rate and the corresponding fitting error were measured with (0.15±0.01) deg/µm in the area with high retardation and (0.06±0.01) deg/µm in the area of low retardation, respectively.
Different methods to investigate human skin by polarized light have been reported, previously. Cross polarized imaging was used by a multiphoton imaging technique  and in an early PS-OCT approach . More detailed PS-OCT studies measuring retardation quantitatively in healthy and burned skin were later reported [7, 24]. Recently, a detailed PS-OCT study of healthy skin at different areas of the human body was reported . Our results are generally in agreement with this latter study, however, there are discrepancies in the interpretation of the observed data. The overall impression of the retardation image of the finger tip region of Fig. 2 is similar to that of Ref. . A random polarization state of light backscattered by the stratum corneum, while light entering deeper layers of the epidermis does not further change its polarization state with depth. However, a careful analysis of our images leads to a different interpretation of the light behavior within the stratum corneum. The stratum corneum is birefringent, light traversing the stratum corneum undergoes a retardation that linearly increases with depth, but its polarization state is not scrambled. Only light directly backscattered from within the stratum corneum is depolarized, indicated by a random polarization state (c.f. Figs. 2(b), (c)). If we apply a higher threshold to select the light whose polarization state is measured and imaged, the light weakly backscattered from within the stratum corneum is rejected, and only the somewhat stronger backscattered signal from sweat glands is observed (c.f. Figs. 2(d), (e)). This light does not show a scrambled polarization state. Instead, a rather constant axis orientation and a steadily increasing retardation with depth are observed. Since the latter is hard to distinguish in 2D sections of the very thin, spiraling sweat glands, we used a 3D data set to demonstrate this effect in more detail. This increasing retardation can be observed along several sweat glands in Fig. 3. Once the light enters deeper layers of the skin, its polarization state remains preserved, indicating that these layers are non or only low birefringent. The cross sections of areas of constant polarization state within these layers are varying in shape (circular to elliptical) and spatial scales (90µm to 700 µm) which are much larger than the speckle size (similar to results obtained in ), again indicating that the light reaching deeper layers (at least to a depth of 800 µm) through the skin is not in an arbitrary polarization state, contrary to statements of previous work . The amount of birefringence of the stratum corneum and the orientation of its fast axis is closely related to the shape and orientation of the ridges of the fingerprint relief. This can be clearly observed in the movie of Fig. 4. The spiral structure of the fingerprint at the surface is preserved through the stratum corneum and can also be found in deeper layers of the epidermis. Retardation and fast axis orientation are determined by the stratum corneum indicated by the fact that these quantities also show a spiral structure closely related to the fingerprint relief. The polarization state is preserved in the deeper layers of the epidermis, until finally the contribution of multiply scattered light, which is in a random polarization state, to the image is dominating.
The cause of the polarizing behavior of the stratum corneum is not fully understood. The stratum corneum is known to contain membrane-like structures consisting of keratin filled corneocytes embedded in a lipid-enriched intercellular matrix . This structure was suggested to cause birefringence . The depolarized nature of light directly backscattered by the stratum corneum, on the other hand, might be caused by backscattering at large, non-spherical structures or particles . Further studies are required to understand this phenomenon in full detail. Figure 5 shows a different region of skin (back of the hand). It can be observed that layers of the epidermis beneath the stratum corneum are non birefringent which corresponds to previously published results of lower back and temple regions . Deeper layers which are part of the dermis consist of fibrous structures which can cause form birefringence. A considerable variability of skin birefringence with anatomical location has been reported previously. While these data are of value to establish a normative database for comparison with pathologies or burn lesions, care has to be taken with respect to the averaging area. As can be seen in the movies of Fig. 5, a considerable variation in averaged retardation can result: suppose, averaging is performed across the y direction. The values that would be obtained from a line centered in the left hand side of the image would considerably deviate from that obtained at the right hand side. This underlines the importance of 3D imaging: averaging should be performed not only across one lateral dimension, but across both, x and y directions.
We presented 3 dimensional OCT images of intensity and to the best of our knowledge, the first 3 dimensional images of retardation, and optical axis orientation of human skin in vivo. The images demonstrated a birefringent nature of the stratum corneum. The observation of a depolarized backscattered signal from within this layer, whereas signals from deeper layers have been found in a defined polarization state, can be regarded as a new phenomenon in PS-OCT imaging (similar images were interpreted as a random variation of the polarization state of light transmitted through this layer ). The importance of three dimensional imaging to measure birefringence properties of skin is underlined by the local (transversal) variations in retardation and optical axis orientation images. These variations could lead to misinterpretation of the observed data if only 2 dimensional images are available. Three dimensional PS-OCT has the potential to increase image contrast and to quantify retardation and orientation of birefringent structures within tissue.
The authors wish to thank H. Sattmann and L. Schachinger for technical assistance. Financial assistance from the Austrian Fonds zur Förderung der wissenschaftlichen Forschung (FWF-grant P14103-MED and P16766-MED) is acknowledged.
References and links
1. D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W. G. Stinson, W. Chang, M. R. Hee, T. Flotte, K. Gregory, C. A. Puliafito, and J. G. Fujimoto, “Optical coherence tomography,” Science 254, 1178–1181 (1991). [CrossRef] [PubMed]
2. B. E. Bouma and G. J. Tearney, Handbook of Optical Coherence Tomography, (Marcel Dekker, New York2002).
3. A. F. Fercher and C. K. Hitzenberger, “Optical coherence tomography,” Chapter 4 in Progress in Optics44, Elsevier Science B.V. (2002)
4. M. R. Hee, D. Huang, E. A. Swanson, and J. G. Fujimoto, “Polarization sensitive low coherence reflectometer for birefringence characterization and ranging,” J. Opt. Soc. Am. B 9, 903–908 (1992) [CrossRef]
5. J. F. De Boer, T. E. Milner, M. J. C. van Gemert, and J. S. Nelson, “Two-dimensional birefringence imaging in biological tissue by polarization sensitive optical coherence tomography,” Opt. Lett. 22, 934–936 (1997) [CrossRef] [PubMed]
6. J. F. De Boer, T. E. Milner, and J. S. Nelson, “Determination of the depth resolved Stokes parameters of light backscattered from turbid media by use of polarization sensitive optical coherence tomography,” Opt. Lett. 24, 300–302 (1999) [CrossRef]
7. C. E. Saxer, J. F. De Boer, B. H. Park, Y. Zhao, C. Chen, and J. S. Nelson, “High speed fiber based polarization sensitive optical coherence tomography of in vivo human skin,” Opt. Lett. 25, 1355–1357 (2000) [CrossRef]
9. C. K. Hitzenberger, E. Goetzinger, M. Sticker, M. Pircher, and A. F. Fercher, “Measurement and imaging of birefringence and optic axis orientation by phase resolved polarization sensitive optical coherence tomography,” Opt. Express 9, 780–790 (2001) http://www.opticsexpress.org/abstract.cfm?URI=OPEX-9-13-780 [CrossRef] [PubMed]
10. B. H. Park, M. C. Pierce, B. Cense, and J. F. de Boer, “Real-time multi-functional optical coherence tomography,” Opt. Express 11, 782–793 (2003) http://www.opticsexpress.org/abstract.cfm?URI=OPEX-11-7-782 [CrossRef] [PubMed]
11. E. Goetzinger, M. Pircher, M. Sticker, A. F. Fercher, and C. K. Hitzenberger, “Measurement and imaging of birefringent properties of the human cornea with phase-resolved polarization-sensitive optical coherence tomography,” J. Biomed. Opt. 9, 94–102 (2004) [CrossRef]
12. B. Cense, T. C. Chen, B. H. Park, M. C. Pierce, and J. F. de Boer, “In vivo depth resolved birefringence measurements of human retinal nerve fiber layer by polarization-sensitive optical coherence tomography,” Opt. Lett. 27, 1610–1612 (2002) [CrossRef]
13. K. Schoenberger, B. W. Colston Jr., D. J. Maitland, L. B. Da Silva, and M. J. Everett, “Mapping of birefringence and thermal damage in tissue by use of polarization sensitive optical coherence tomography,” Appl. Opt. 37, 6026–6036 (1998) [CrossRef]
14. J. F. de Boer, S. M. Srinivas, A. Malekafzali, Z. Chen, and J. S. Nelson, “Imaging thermally damaged tissue by polarization sensitive optical coherence tomography,” Opt. Expr. 3, 212–218 (1998) [CrossRef]
15. A. G. Podoleanu, G. M. Dobre, and D. A. Jackson, “En-face coherence imaging using galvanometer scanner modulation,” Opt. Lett. 23147–149 (1998) [CrossRef]
16. C. K. Hitzenberger, P. Trost, P. W. Lo, and Q. Zhou, “Three-dimensional imaging of the human retina by high-speed optical coherence tomography for retinal imaging,” Opt. Express 112753–2761 (2003) http://www.opticsexpress.org/abstract.cfm?URI=OPEX-11-21-2753 [CrossRef] [PubMed]
17. R. G. Cucu, A. G. Podoleanu, R. B. Rosen, A. B. Boxer, and D. A. Jackson, “En face polarization sensitive optical coherence tomography,” Proc. SPIE 5140, 113–119 (2003) [CrossRef]
18. M. Pircher, E. Goetzinger, R. Leitgeb, and C. K. Hitzenberger, “Transversal phase resolved polarization sensitive optical coherence tomography,” J. Phys. Med. Biol. 49, 1257–1263 (2004) [CrossRef]
19. M. C. Pierce, J. Strasswimmer, B. H. Park, B. Cense, and J. F. De Boer, “Birefringence measurements in human skin using polarization sensitive optical coherence tomography,” J. Biomed. Opt. 9, 287–291 (2004) [CrossRef] [PubMed]
20. P. J. Tadrous (2002), http://www.bialith.com/
21. B. A. I. van den Bergh, J. Vroom, H. Gerritsen, H. E. Junginger, and J. A. Bouwstra, “Interactions of elastic and rigid vesicles with human skin in vitro: electron microscopy and two-photon excitation microscopy,” Biochem. et. Biophys. Acta 1461, 155–173 (1999) [CrossRef]
22. Y. Sun, J.W. Su, W. Lo, S. J. Lin, S. H. Jee, and C. Y. Dong, “Multiphoton polarization imaging of the stratum corneum and the dermis in ex-vivo human skin,” Opt. Express 11, 3377–3384 (2003) http://www.opticsexpress.org/abstract.cfm?URI=OPEX-11-25-3377 [CrossRef] [PubMed]
23. J. M. Schmitt and S. H. Xiang, “Cross-polarized backscatter in optical coherence tomography of biological tissue,” Opt. Lett. 23, 1060–2062 (1998) [CrossRef]
24. B. H. Park, C. E. Saxer, S. M. Srinivas, J. S. Nelson, and J. F. de Boer, “In vivo burn depth determination by high-speed fiber-based polarization sensitive optical coherence tomography,” J. Biomed. Opt. 6, 474–479 (2001) [CrossRef] [PubMed]