INSTITUTE OF PHYSICS PUBLISHING
PHYSICS IN MEDICINE AND BIOLOGY
Phys. Med. Biol. 49 (2004) 1257–1263
PII: S0031-9155(04)74164-0
Transversal phase resolved polarization sensitive optical coherence tomography Michael Pircher, Erich Goetzinger, Rainer Leitgeb and Christoph K Hitzenberger Department of Medical Physics, University of Vienna, Waehringerstr 13, A-1090 Vienna, Austria E-mail:
[email protected]
Received 1 October 2003 Published 18 March 2004 Online at stacks.iop.org/PMB/49/1257 (DOI: 10.1088/0031-9155/49/7/013) Abstract We present a novel optical coherence tomography (OCT) method to measure backscattered intensity and birefringence properties (retardation and fast axis orientation) and apply it to imaging of human ocular tissue. The method is based on a Mach Zehnder interferometer, on transversal scanning, and on a polarization sensitive two-channel detection. A highly stable carrier frequency is generated by acousto-optic modulators (AOMs). This allows a phase sensitive demodulation by the lock-in technique. Since the recording of individual interference fringes is avoided by this method the amount of data to be recorded and processed is considerably reduced. We demonstrate this method on human cornea and anterior chamber angle and present, to the best of our knowledge, the first OCT images of retardation and fast axis orientation of the anterior chamber angle region in vivo.
1. Introduction Conventional optical coherence tomography (OCT) systems based on intensity measurements provide structural images of tissue (Huang et al 1991). Recent overviews of this technique can be found in Bouma and Tearney (2002), Fercher and Hitzenberger (2002). However, many samples show poor contrast, if imaged on intensity basis only. Polarization sensitive OCT (PS-OCT) obtains information on birefringence tissue properties and has developed into a promising technique to gather additional information on the sample (Hee et al 1992, De Boer et al 1997, Saxer et al 2000, Hitzenberger et al 2001, Jiao and Wang 2002). Conventional PS-OCT measures the envelope of the interference signal in two orthogonal polarization channels (Hee et al 1992, De Boer et al 1997). This technique allows measurement of reflectivity and retardation of a sample. If additional information is to be obtained, e.g. on the orientation of the birefringent axis, on Stokes vectors, on M¨uller and Jones matrix elements, phase sensitive recording is required (De Boer et al 1999, Saxer et al 2000, 0031-9155/04/071257+07$30.00 © 2004 IOP Publishing Ltd Printed in the UK
1257
1258
M Pircher et al
Hitzenberger et al 2001, Jiao and Wang 2002). Most of the presently used OCT techniques are based on longitudinal scanning of the sample. In this longitudinal OCT technique, the fast, or priority scan (A-scan), is in z direction, perpendicular to the sample surface. Several adjacent A-scans are combined to a cross sectional OCT tomogram. For phase resolved PS-OCT the full interferometric signal has to be recorded, i.e. each interference fringe has to be resolved, which results in a huge number of data points to fulfil the Nyquist criterion. A tissue whose polarizing properties are of great interest for medical diagnostics is the human cornea. Several diseases may change the birefringence of this tissue (Goetzinger et al 2003a, 2003b). Since 3D (three-dimensional) information is often required for a comprehensive diagnostic overview of the cornea (Goetzinger et al 2003b), scanning and data acquisition schemes that reduce the data volume are necessary. Although previous publications (Park et al 2003) presented polarization sensitive systems that do record the full interferometric signal in a very short time for phase resolved PS-OCT, these techniques have to handle a huge amount of data and therefore are limited to a relatively small number of A-scans compared to the number of A-scans needed for 3D imaging (>10 000 A-scans). An alternative OCT scanning scheme based on en-face imaging has been introduced (Podoleanu et al 1998a) and successfully applied for retinal imaging (Podoleanu et al 1998b). In this technique, the path length modulation introduced by x–y galvo scanning mirrors is used to generate a carrier frequency. While the advantage of this method of carrier frequency generation is that no additional electro-optic component is needed, its drawback is a varying carrier frequency which makes phase recovery by demodulation difficult. We extend the concept of en-face OCT to a phase resolved PS-OCT technique. To overcome the problem of varying carrier frequency, we generate a very stable carrier frequency by use of acousto-optic modulators (AOMs). This allows the extraction of amplitude and phase information by a lock-in demodulation technique. The advantages of this method are high speed and a largely reduced amount of data to be processed. We present this technology and show first results of PS-OCT images showing retardation and fast axis orientation in human ocular tissues, including the first in vivo PS-OCT images of human anterior chamber angle region. 2. Method The experimental set-up, which is based on a Mach Zehnder interferometer, is shown in ∼ 55 nm, figure 1. A light source (AFC Technologies, Canada) emitting at λ0 = 1310 nm (λ = output power = 24 mW) is used. Previous work already investigated the advantage of using a 1300 nm light source for imaging the anterior segment of the human eye (Radhakrishnan 2001). The emitted light is vertically linear polarized by a polarizer and coupled into the interferometer where it is split into a sample beam and a reference beam by the non-polarizing beam splitter NPBS1. The reference beam propagates through two AOMs (IntraAction Corp. AOM-40N) and through a half wave plate with a fast axis orientation of 22.5◦ to the horizontal (the beam is now linear polarized at 45◦ ). The first AOM shifts the frequency of the reference beam by −40 MHz, the second by +40.1 MHz, resulting in a net frequency shift of 100 kHz in comparison to the sample beam. It was necessary to use this configuration consisting of two AOMs to generate the net frequency shift of 100 kHz, because there are no AOMs available in the kHz frequency range. The used AOMs consist of dense flint glass as interaction medium and therefore introduce no additional birefringence to the reference beam. A path delay unit consisting of two mirrors mounted on a translation stage in the reference arm enables a change of the reference arm length (depth scanning). The sample beam propagates through a glass rod of the same material and length as the AOMs, for dispersion compensation. A non-polarizing beam splitter (NPBS2) directs the beam towards the sample. A quarter wave plate with a
Transversal phase resolved polarization sensitive optical coherence tomography
1259
Figure 1. Experimental set-up of the transversal phase resolved polarization sensitive optical coherence tomography technique (for the in vitro measurements the dual balanced detection unit was not implemented, only the signals from D1 and D2 were recorded). LS—light source, P—polarizer, NPBS—non-polarizing beam splitters, GS—galvo scanner, DC—dispersion compensation, AOM—acousto-optic modulators, λ/4—quarter wave plate, λ/2—half wave plate, PBS—polarizing beam splitters, D1, D2, D3, D4—detectors, PC—personal computer, DAQ—data acquisition board.
fast axis orientation of 45◦ to the horizontal converts the linear polarized light into circular polarized light. A mirror mounted on a galvo scanner and a lens (f = 50 mm) are used to scan the beam over the sample. The optical power on the sample was 1 mW which is well below the maximum permissible exposure of 15.4 mW for intrabeam viewing through a 7 mm pupil and exposure times of up to 8 h (American National Standards Institute 2000). The backscattered light from the sample is collimated by the lens and propagates back through the quarter wave plate (having now an elliptical polarization state depending on the birefringence properties of the sample), and through the non-polarizing beam splitter NPBS2. It is combined with the reference beam at the non-polarizing beam splitter NBPS3. Both beam components are directed towards two polarization sensitive two-channel detection units (Hee et al 1992). In case of path length matching, each detector measures an interference signal that has a beat frequency (carrier frequency) of 100 kHz due to the frequency shift introduced to the reference beam. The signals D1–D4 and D2–D3 (balanced detection) are pre-amplified, and a lock-in amplifier (Stanford Research Systems SR830DSP) demodulates each signal by multiplying the signal with a reference sine signal and low pass filtering (corresponding to the real part of the signal) and multiplying with a reference sine signal, which was phase shifted by π /2 and low pass filtering (corresponding to a reconstructed imaginary part of the signal). Each part of the signals was recorded by a data acquisition board. Our method is based on a transversal sample scanning. The reference path length is changed at low speed while the transverse scanning is performed at a repetition rate of 50 Hz. The main difference to other en-face OCT imaging techniques (Podoleanu et al 1998a, 1998b) is that the AOMs introduce a constant frequency shift to the reference beam which allows demodulation and a phase sensitive recording by the lock-in technique. This avoids the necessity of recording all individual interference fringes and thereby greatly reduces the number of data points. Our technique requires only four data points for each image pixel (one imaginary and one real value in each channel). Other phase resolved OCT techniques need at least two data points per fringe and channel. A typical image pixel has a depth extension
1260
M Pircher et al
of the order of the coherence length, ∼13 µm in case of our light source, consisting of ∼20 fringes. Therefore our technique reduces the number of necessary data points by at least a factor of 20. The method of calculating reflectivity, retardation and fast axis orientation from the real and imaginary parts of the signals recorded by the two detection channels is similar to that previously published (Hitzenberger et al 2001). Image acquisition time depends on transversal scanning speed and the number of transversal lines in the OCT image. If a transversal resolution equal to the focal spot diameter is to be achieved, the carrier frequency limits the transversal scanning speed. If we assume that two periods of the carrier signal are needed to resolve one transversal data point, the maximum transversal scanning speed Vmax can be calculated by x · F (1) Vmax = 2 with x the focal spot diameter and the carrier frequency F. In our system we used a carrier frequency of 100 kHz, a focal spot diameter of 16 µm which results in a Vmax ∼ 830 mm s−1. By placing neutral density filters into the sample arm we measured the weakest detectable intensity from the sample arm which defines system sensitivity. The sensitivity of our system was measured with 100 dB. 3. Results and discussion To compare our method with previously published polarization sensitive measurements of human cornea (Goetzinger et al 2003a, 2003b), which were recorded with a slow A-scan based interferometer, we recorded a 3D data set of a human donor cornea in vitro with the new technique. The cornea, which was not Hepatitis negative (and therefore could not be used for transplantation and would have otherwise been discarded), was received from the eye bank (Department of Ophthalmology, General Hospital, University of Vienna). The cornea was stored in Optisol (Chiron, CA) for approximately two weeks before the experiment. In these first measurements dual balanced detection was not implemented. The data set consists of 490 × 100 × 100 (x, y and z direction) pixels covering an area of 8 × 8 × 2.5 mm3, and the total measuring time was approximately 250 s. Figure 2 shows intensity (on a logarithmic scale), retardation and fast axis orientation of one slice (x–z plane) of the 3D data set. Only values above a certain intensity threshold were used in the calculations for the retardation and the fast axis orientation, values below this threshold are displayed in grey. (In case of low intensity birefringence data obtained by PS-OCT are not reliable (Everett et al 1998).) It should be mentioned that the interpretation of fast axis orientation (orientation of the fast axis with respect to the polarizing beam splitter of the detection unit) images deserves some care. Close to the surface of the sample, the total retardation is very low, ideally yielding no signal in channel 2 (because of crossed polarization states). This should cause a random variation of axis orientation measured at the surface. Contrary to that assumption, there is a predominant axis orientation along the surface layer. This is an artefact caused by the nonperfect polarization properties of the optic components used in the instrument. Furthermore, the axis orientation measured in each depth z is a cumulative axis orientation, i.e. the effective axis orientation of a single homogeneous linearly birefringent structure that would cause the same polarization changes to a double passing beam up to a depth of z. Figure 3 shows en-face images, which display the values of retardation (a) and fast axis orientation (b) at the posterior surface of the cornea of the recorded 3D data set. The circular symmetric retardation that increases towards the periphery of the cornea and the fast axis orientation, which is an approximately linear function of the azimuth angle, correspond well with previously published results (Goetzinger et al 2003a, 2003b).
Transversal phase resolved polarization sensitive optical coherence tomography
1261
(a)
(b)
(c)
Figure 2. (a) Intensity, (b) retardation δ (blue δ = 0, red δ = 90◦ ) and (c) fast axis orientation (blue = −90◦ , red = +90◦ ) images of human cornea in vitro. (Each image covers an area of 8 mm (transversal) times 2.5 mm (depth).)
(a)
(b)
Figure 3. En-face images of retardation (a) and fast axis orientation (b) at the posterior surface of a human cornea in vitro (each image covers an area of 8 × 8 mm2).
Figure 4 shows the anterior chamber angle region of a healthy human eye recorded in vivo with the new technique. Each image covers an area of 5 × 5 mm2 and consists of 500 (horizontal) × 100 (vertical) pixels. Image acquisition time was 2 s. The retardation image reveals the increasing birefringence at the periphery of the cornea which has already been shown in vitro. In figure 4(c) a change in the fast axis orientation near the limbus
1262
M Pircher et al
(a)
(b)
(c)
Figure 4. (a) Intensity, (b) retardation δ (blue δ = 0, red δ = 90◦ ) and (c) fast axis orientation (blue = –90◦ , red = +90◦ ) images of human anterior chamber angle in vivo. (The horizontal line in the upper part of the images is an artefact, i.e. a ghost image caused by the strong reflex from the pigment epithelium, each image covers an area of 5 × 5 mm2.)
(fusion region of cornea and sclera) is visible. We believe that this change results from the different fibril orientation of the cornea in this region (Newton and Meek). Remarkable is the non-birefringent nature of the anterior layers of the iris. The measured cumulative retardation and fast axis orientation in these layers represent the values obtained on the back surface of the cornea. On the other hand, the iris pigment epithelium (posterior part of iris) seems to be depolarizing which is clearly visible in figure 4(c), which shows a random orientation of the (cumulative) fast axis of the iris pigment epithelium.
4. Conclusion We have demonstrated a novel transversal phase resolved polarization sensitive optical coherence tomography technique for imaging human ocular tissue. The advantage of this method is a largely reduced number of data points as compared to other phase resolved OCT techniques. The en-face technique seems to be preferable if volumes with larger transversal extensions (x–y direction) than depth extensions (z direction) have to be imaged. Since the penetration depth in OCT is limited, the transversal extension in OCT images usually exceeds the depth extension. In this case the number of A-scans greatly exceeds the number of transversal lines. The current image acquisition speed is only limited by the maximum detectable frequency of our lock-in amplifiers (currently 100 kHz). A 10 to 100 times faster image acquisition speed seems to be possible by shifting the carrier frequency to 1–10 MHz (Hitzenberger et al 2003), which allows us to record a 3D data set within 1 s. The sensitivity of our technique is increased compared to other en-face techniques (Moreau et al 2003, Akiba et al 2003, Laubscher et al 2002).
Acknowledgments The authors wish to thank H Sattmann and L Schachinger for technical assistance and R Amon, General Hospital of Vienna, for providing the donor corneas. Financial assistance from the Austrian Fonds zur F¨orderung der wissenschaftlichen Forschung (grant no P14103-MED) is acknowledged.
Transversal phase resolved polarization sensitive optical coherence tomography
1263
References Akiba M, Chan K P and Tanno N 2003 Full-field optical coherence tomography by two dimensional heterodyne detection with a pair of CCD cameras Opt. Lett. 28 816–8 American National Standards Institute 2000 American National Standard for Safe Use of Lasers ANSI Z 136.1–2000 (Orlando: Laser Institute of America) pp 45–9 Bouma B E and Tearney G J 2002 Handbook of Optical Coherence Tomography (New York: Marcel Dekker) De Boer J F, Milner T E, van Gemert M J C and Nelson J S 1997 Two-dimensional birefringence imaging in biological tissue by polarization sensitive optical coherence tomography Opt. Lett. 22 934–6 De Boer J F, Milner T E and Nelson J S 1999 Determination of the depth-resolved Stokes parameters of light backscattered from turbid media by use of polarization-sensitive optical coherence tomography Opt. Lett. 24 300–2 Everett M J, Schoenberger K, Colston Jr B W and Da Silva L B 1998 Birefringence characterization of biological tissue by use of optical coherence tomography Opt. Lett. 23 228–30 Fercher A F and Hitzenberger C K 2002 Optical coherence tomography Prog. Opt. 44 215–302 Goetzinger E, Pircher M, Sticker M, Dejaco-Ruswurm I, Kaminski S, Skorpik C, Fercher A F and Hitzenberger C K 2003a Imaging of polarization properties of the human cornea with phase resolved polarization sensitive optical coherence tomography Proc. SPIE 4951 21–5 Goetzinger E, Pircher M, Sticker M, Dejaco-Ruhswurm I, Kaminski S, Findl O, Skorpik C, Fercher A F and Hitzenberger C K 2003b Three dimensional polarization sensitive optical coherence tomography of normal and pathologic human cornea Proc. SPIE 5140 22 Hee M R, Huang D, Swanson E A and Fujimoto J G 1992 Polarization sensitive low coherence reflectometer for birefringence characterization and ranging J. Opt. Soc. Am. B 9 903–8 Hitzenberger C K, Goetzinger E, Sticker M, Pircher M and Fercher A F 2001 Measurement and imaging of birefringence and optic axis orientation by phase resolved polarization sensitive optical coherence tomography Opt. Exp. 9 780–90 Hitzenberger C K, Trost P, Lo P W and Zhou Q 2003 Three-dimensional imaging of the human retina by high-speed optical coherence tomography for retinal imaging Opt. Exp. 11 2753–61 Huang D, Swanson E A, Lin C P, Schuman J S, Stinson W G, Chang W, Hee M R, Flotte T, Gregory K, Puliafito C A and Fujimoto J G 1991 Optical coherence tomography Science 254 1178–81 Jiao S and Wang L V 2002 Jones-matrix imaging of biological tissues with quadruple-channel optical coherence tomography J. Biomed. Opt. 7 350–8 Laubscher M, Ducros M, Karamata B and Lasser T 2002 Video-rate three-dimensional optical coherence tomography Opt. Exp. 10 429–35 Moreau J, Loriette V and Boccara A C 2003 Full-field birefringence imaging by thermal light polarization sensitive optical coherence tomography: II. Instruments and results Appl. Opt. OT 42 3811–8 Newton R H and Meek K M 1998 The integration of the corneal and limbal fibrils in the human eye Biophys. J. 75 2508–12 Park B H, Pierce M C, Cense B and De Boer J F 2003 Real-time multi-functional optical coherence tomography Opt. Exp. 11 782–93 Podoleanu A G, Dobre G M and Jackson D A 1998a En-face coherence imaging using galvanometer scanner modulation Opt. Lett. 23 147–9 Podoleanu A G, Seeger M, Dobre G M, Webb D J, Jackson D A and Fitzke F W 1998b Transversal and longitudinal images from the retina of the living eye using low coherence reflectometry J. Biomed. Opt. 3 12–20 Radhakrishnan S, Rollins A M, Roth J E, Yazdanfar S, Westphal V, Bardenstein D S and Izatt J A 2001 Real-time optical coherence tomography of the anterior segment at 1310 nm Arch. Ophthal. 119 1179–85 Saxer C E, De Boer J F, Park B H, Zhao Y, Chen C and Nelson J S 2000 High speed fiber based polarization sensitive optical coherence tomography of in vivo human skin Opt. Lett. 25 1355–7
