Characterization of dentin, enamel, and carious lesions by a polarization-sensitive optical coherence tomography system

Transcription

1 Characterization of dentin, enamel, and carious lesions by a polarization-sensitive optical coherence tomography system Yueli Chen, Linda Otis, Daqing Piao, and Quing Zhu Enamel and dentin are the primary components of human teeth. Both of them have a strong polarization effect. We designed a polarization-sensitive optical coherence tomography (PSOCT) system to study the spatially resolved scattering and polarization phenomena of teeth. The system is constructed in free space to avoid the complexity of polarization control in fiber-based PSOCT. The structural features of enamel were evaluated in five human teeth that had no visible evidence of caries. The teeth were subsequently sectioned in mesial distal orientation and coronal orientation. Then the structural aspects of dentin were evaluated. OCT images were made of the mantel dentin near the dentin enamel junction. Five teeth with interproximal and occlusal caries were also studied. With two channel and phase-retardation images, PSOCT provided better functional contrast and more detailed structural information than conventional OCT. For a better description of the measured PSOCT data, we classify these features by two types, i.e., the local textural features and the global structural features. This study indicates that PSOCT has the potential to be a powerful tool for research of dental formation and caries diagnosis Optical Society of America OCIS codes: , , Introduction Light-polarization effects are present in many biological tissues. For example, isotropic and anisotropic bands of the muscle are named after their different polarization properties. Although tissue polarization properties are important for understanding tissue structures, it is not an easy task to retrieve spatially resolved polarization information inside biological samples. The polarization state measured with conventional polarimetric and ellipsometric systems is usually the average or sum result of the bulk samples. Optical coherence tomography (OCT) is a new imaging modality, which can provide high-resolution imaging of subsurface structures Polarizationsensitive OCT (PSOCT) is a functional extension of OCT. 3,4,9,11 14 It can provide high-resolution images of Y. Chen, D. Piao, and Q. Zhu (zhu@engr.uconn.edu) are with the Department of Electrical and Computer Engineering, University of Connecticut, 371 Fairfield Road, Unit 1157, Storrs, Connecticut L. Otis is with the School of Dental Medicine, University of Pennsylvania, Philadelphia, Pennsylvania Received 10 August 2004; revised manuscript received 11 November 2004; accepted 15 November /05/ $15.00/ Optical Society of America tendons, teeth, vessels, skin, muscles, bones, nerve fibers, etc. 4,13,15 23 Polarization properties in tissues, such as birefringence, optical rotation, Stokes parameters and depolarization, can be uniquely studied with PSOCT. In addition, some of these properties are a consequence of the disease process and may also provide important diagnostic information. Teeth are primarily composites of enamel and dentin, each of which is also a composite. 24 Enamel, the hardest substance in the human body, is heavily mineralized with calcium salts. It is densely packed with hydroxyapatite crystals, and there is a regular alignment of the crystals that form enamel prisms. Dentin underlies the enamel and forms the bulk of a tooth. It is bonelike tissue that consists of collagen fibers and the same mineral salts of enamel, with micrometersized dentin tubules regularly aligned inside. The boundary between enamel and dentin is called the dentin enamel junction (DEJ). Studies of the tooth anatomy by conventional OCT have been reported in several papers Since the crystalline scatterers within the tooth matrix are anisotropic to light and the linearly ordered structures in dentin and enamel are birefringent, it is necessary to include polarization parameters for tooth imaging. Wang et al. 21 observed the birefringence effect in enamel. Baumgartner et al. 22 showed a larger phase-retardation change in the caries 10 April 2005 Vol. 44, No. 11 APPLIED OPTICS 2041

2 site, and a larger depth profile in the carious lesions compared with the conventional OCT image. In a study of simulated caries in bovine enamel, Fried et al. 23 used a fiber-based system to study induced carious lesions by fast and slow axes images. In this paper we designed a free-space dual-balanced PSOCT system and studied the structural aspect of enamel and dentin in five intact human teeth. PSOCT images of enamel have a particularly strong polarization effect in which signals deeper than 3 mm are well observed. Form birefringence and local polarization features are found in both enamel and dentin. Correlation of these features with the microstructures of enamel, dentin, and caries affection is discussed: Our results also demonstrat the potential of using PSOCT for dental research and caries diagnosis. Two major polarization mechanisms in biological samples were introduced by other investigations. 1,12 The first is the polarization anisotropy introduced by local scattering particles. The second is the structural anisotropy introduced by larger scale (i.e., global) linearly ordered structures. The structure anisotropy is also called form birefringence. In this paper we refer to imaging features introduced by the first mechanism as local features, which can be characterized by the rich textural pattern of spots smaller than 100 m. Other larger structures seen in PSOCT images are referred to as global features, such as DEJ and Hunter Schreger bands in enamel. More interesting, a periodical black-and-white band feature was observed uniquely from PSOCT images in both dentin and enamel. This global feature was the result of the form birefringence. The local feature is prominent when small PS particles exist randomly in the turbid medium. The global black-and-white band feature is an accumulative effect and requires long-range order in the tissue. These two features are related to the different polarization structures of tooth samples. 2. Method A. Experimental Configuration The schematic diagram of our PSOCT system is shown in Fig. 1. A 2 mw, 1310 nm super luminescent diode (SLD) provides the low-coherence illumination. Light beam from the SLD passes a vertical polarizer and is then seperated by a beam splitter. In the sample arm two configurations are used. The standard configuration translates the linearly polarized light into circularly polarized light with a quarterwaveplate (QWP) in which the fast axis is oriented at 45 with respect to the vertical direction [circularpolarization (CP) configuration]. In the second configration this QWP is removed such that the probing beam is kept in a linearly polarized state (LP configuration). This simpler configuration is an alternate choice for evaluating the scattering and polarization effects. More details of the two configurations are given in the subsection 2.B. In the reference arm another QWP is placed before a galvanometer. Instead of using a 22.5 azumth angle, as in most of the Fig. 1. Schematic of the PSOCT system: SLD, superluminescent diode; P, polarizer; BS, beam splitter; PBS, polarizing beam splitters; QWP1, QWP2, quarter-wave plates. The detected signals are filtered by two identical bandpass filters. After analog-to-digital conversion, the raw data are then processed by computer. PSOCT systems, we orient the QWP at an optimum angle to provide an equal amount of reference light for both of the orthogonal polarization channels. The galvanometer is implemented for calibration and phase modulation. The reference mirror is driven by a stepper motor to generate ranging depth. The lateral scan of the sample is controlled by another linear motor. The light beams from both the sample and the reference arm interfere at the beam splitter. The recombined light is separated by polarizing beam splitters before it enters the photodetectors. The vertically and horizontally polarized signals are detected independently by two detection channels (V and H channels). The two detection channels are identical, and the images from two separate channels can be formed simultaneously. A dual-balanced detection scheme is also implemented. This detection scheme can reject commonmode noise and improve the signal-to-noise ratio of the system. 25 Two polarizing beam splitters and two dual-balanced detectors are used for the balanced detection in our PSOCT system. B. Calculation of Polarization-Sensitive Optical Coherence Tomography Parameters The response of the two channels can be calculated based on Jone s matrix formalism. The complex light field is expressed as a vector: E Ẽ V Ẽ H, (1) where V and H indicate the two orthogonal polarization directions. The sample can be expressed in a general retarder form as 2042 APPLIED OPTICS Vol. 44, No April 2005

3 S(, ) cos2 sin 2 exp( j2 ) cos sin 1 exp( j2 ) cos sin 1 exp( j2 ) cos 2 exp( j2 ) sin 2, (2) where is the average phase retardation angle (single trip of backscattered photon) and is the average fast-axis orientation. In the standard CP configuration, we have E V A V exp ( z l) 2 sin, (3) E H A H exp ( z l) 2 cos, (4) where z is the optical path difference, l is the SLD coherence length, and E V and E H are the detected field amplitude. Since the two channels are identical, constants A V and A H can be considered to be the same. The phase-retardation can then be expressed by arctan(e V E H ), (5) and its value varies from 0 to 90 deg. The regular OCT image can be obtained by R E V 2 E H 2. (6) A detailed Jone s formalism calculation can be found in other papers. 12,14 In the LP configuration, in which the QWP is removed from the sample arm, the Jone s transfer matrix calculation of the light field is E Ẽ V Ẽ H cos 2 sin 2 exp( j ) (7) cos sin 1 exp( j ). It can be seen that and are coupled in Eq. (7). Consequently, Eq. (5) will not give the physical meaning of phase retardation. If the polarization effect is not prominent, the energy of the backscattered light from the biological sample will be preserved in one channel. However, the prevailing PS scatterers in a random fashion tend to scramble the energy distribution into two channels. 12 This type of scattering effect in the turbid tissue also reduces the degree of polarization Under this condition the LP configuration gave a local feature similar to the CP configuration. The energy ratio between two channels can be taken as the fundamental PS parameter to evaluate the sample properties. Thus Eq. (5) is simply a mapping algorithm without the physical meaning of phase retardation. C. Sample Preparation Five intact human teeth that had no visible evidence of caries were used to study the dental structure. Five tooth samples with caries that could be visually identified in occlusal and interproximal regions were imaged to characterize carious lesion properties. The structural features of enamel were evaluated from the facial aspect of the five intact human teeth with no visible caries evidence. OCT images of the enamel were made in a longitudinal orientation. For the assessment of dentin structural features, the five teeth were subsequently sectioned in a mesial-to-distal orientation and in a coronal orientation. OCT images were made of the mantel dentin, near the DEJ. 3. Results and Discussion A. Enamel, Dentin, and Dental Enamel Junction In this paper, without specification, the default image size is 2.5-mm long and 3.6-mm deep (the depth in air, with no index modification). The PS images are from the CP configuration, except for Fig. 5. Figure 2 shows the enamel structure and the well-delineated DEJ from the facial aspect. The DEJ boundary prolonged from the surface into the sample can be seen from the four images. From the V H channel and intensity images shown in Figs. 2(a), 2(b), and 2(d), the DEJ can be identified as a white interface owing to its strong reflection. The DEJ is differentiated as a line of texture pattern in the phase-retardation image shown in Fig. 2(c). The DEJ in conventional OCT images and in intraoral radiographs appears sharply demarcated; however, scanning electron microscopy reveals that it is an irregular, ruffled structure. This irregular, ruffled morphology is clearly visible in the PS OCT images. The penetration depth of dentin is generally found to be less than that of enamel. The images of enamel generally have richer information and deeper penetration, whereas images in dentin usually have a shallower depth profile. This has been attributed to the larger scattering coefficient of dentin reported by Fried et al. 23 Our result suggests that the deficiency of scattering particles or boundaries may be another explanation for the lesser depth penetration of dentin. The two tissues types can be clearly differentiated by their different polarization and scattering properties. B. Global Structures The global befringent properties of enamel arises from the regular orientation of the enamel rods. Enamel rods originate at the DEJ in a perpendicular orientation. After a short distance they turn in a plane horizontal to their long axis, first in one direction and then in the other until they assume a straight aligment perpendicular to the surface. 30 In longitudinal microscopic ground sections viewed with reflected light, this arrangement of the enamel rods gives rise to a light and dark pattern termed Hunter Schreger bands. Our results indicate that Hunter Schreger bands can be seen in the PSOCT images [Fig. 3(a)]. This morphology is consistent with that previously described in light 10 April 2005 Vol. 44, No. 11 APPLIED OPTICS 2043

4 Fig. 2. (a) V-channel image, (b) H-channel image, (c) phase-retardation image by Eq. (5), (d) intensity image of R E 2 v E 2 H. The bright reflection band inside is the DEJ. microscopic images. 30 It is also interesting to note that the curvature of the form birefrigent bands aligns normally with the structure of the enamel rods, which implies a relation between the form birefringence and the ordered spatial structures. The diameter of a single enamel rod is smallest at the DEJ and increases toward the surface. The average diameter of a single rod is approximately 4 m, whereas the PSOCT limit is 5 m/pixel. High-resolution OCT should provide further definition to the study human enamel rod microstructure. Incremental growth lines of enamel (lines of Retizius) have been described in light microscopic images of teeth. 30 When viewed in coronal sections, the lines of Retizius are seen as parallel concentric circles similar to the growth rings on a tree. In longitudinal sections they appear oblique to the surface. The lines of Retizius are typically less prominent at the cervical aspect of the crowns. In PSOCT images from the coronal aspect of enamel, we have observed alternate black-and-white bands that are consistent with the line of Retizius in most of the teeth imaged in this study [Fig. 3(b)]. The form birefringence from the coronal aspect is much stronger than that from the cervical aspect. Enamel surfaces that have sustained minimal abrasion frequently demonstrate longitudinal lamellae and cracks. Such lamellae were detected in some of the specimens in this study. In PSOCT images the lamellae boundaries tended to diffuse into the surrounding area; this may have arisen from the sur APPLIED OPTICS Vol. 44, No April 2005

5 it is a nondestructive method that can be used in vivo. It is likely that PSOCT will become an important tool in the structural assessment of dentin and enamel. C. Local Polarization-Sensitive Texture and Caries Both dentin and enamel show PS textures. These textural patterns are formed by PS particles about m in diameter. We put an emphasis on this feature because the textures revealed noninvasively very rich information underneath the tooth surface. Systematic analysis of these textures, such as particle size, polarization strength, orientation, and distribution of the patterns etc., may reveal a close relation to the mineralization status of the teeth. Dental caries results from a gradual demineralization of enamel and underlying dentin. The mineralization level is an important indicator of the extent of the carious lesions. Figure 4 shows images of the occlusal and interproximal carious lesions. The white arrow points toward the carious site, which can be visually identified as a black or brown spot from the surface. The carious lesions show richer texture information and higher contrast than the noncaries site. The pattern is characterized by randomly distributed spots in black and white, which exhibit the strong local PS scattering. Occlusal carious lesion is showed in Fig. 4(a), and the caries area is enlarged and shown in Fig. 4(b). The areas indicated by black arrows 1 and 2 show similar caries patterns, but caries were not clearly identified at the corresponding surfaces. The white bands in areas 3 and 4 indicate that the energy is preserved in the vertical channel along the scanning depth. These areas have textures composed of spots of similar size as in the caries site, but the state of polarization tends to be maintained; i.e., white always appears white and black always appears black. This type of pattern was seen frequently around the carious region. Similar features are also found in interproximal carious lesions [Fig. 4(c)]. Fig. 3. Horizontal black-and-white band structure is originated from the form birefringence. (a) Phase retardation image from the cervical side of the tooth near the DEJ. Vertical bands formed by enamel rods are aligned normal to the structure of the form birefringent bands. (b) Another image from the coronal aspect of a tooth after it was sectioned in mesial distal and coronal orientation. Strong form birefringence is observed. rounding PS structures. As enamel rods introduced form birefringence, dentin is expected to be birefringent owing to its ordered tubular structure. Some samples showed this periodic structure in dentin too but not as typical as in enamel. In summary, by using PSOCT we have detected ultrastructural characteristics of enamel and dentin that are consistent with light microscopic and scanning electron micsroscopic features. PSOCT has the advantage for the ultrastructural assessment of dentin and enamel because D. Comparison of Circular- and Linear-Polarization Configurations Images obtained from LP and CP configurations are shown in Fig. 5. The two configurations provide similar local features of dentin structures [longer arrows in Figs. 5(a) and 5(b)]. However, the global form birefringence patterns of enamel [shorter arrows in Figs. 5(a) and 5(b)] obtained from the CP configuration is degraded compared with the LP configuration. This is consistent with the theory of Eqs. (5) and (7), in which and are decoupled only in CP configuration and yield the global phaseangle variation, whereas the LP configuration can be used for local feature extraction. Resent literature reported that the depolarization rate of linear and circular polarized light were different It would also be interesting to compare in detail the difference between the two configurations. More studies on LP and CP configurations will be reported in the future April 2005 Vol. 44, No. 11 APPLIED OPTICS 2045

6 Fig. 4. (a) Phase-retardation image of a tooth with occlusal carious lesion. The image size is 5 mm in length and 3.6 mm in depth. The rich textures in the rectangular region may relate to the mineral composition, which is a sign of caries affection. (b) Amplified carious region. (c) Phase-retardation image with interproximal caries affection (left root, right crown) 4. Summary and Future Study In this paper, we report the use of a PSOCT system with improved system designs to characterize dentin and enamel and to study the special properties related to the structural orientation of enamel rods. As a general comparison with the conventional intensity OCT, the PSOCT images provide improved contrast and more detailed functional information. The results of this study indicate that PSOCT has the potential to provide higher-resolution informa APPLIED OPTICS Vol. 44, No April 2005

7 regarding the usefulness of this technology for the clinical diagnoses of dental caries. Since the different local texture patterns are most likely related to the mineral components within the tooth, it is interesting and promising to design dynamic processes with demineralization, remineralization, and acid control to track the relation between mineralization level and PSOCT features. The change of the local texture patterns before and after treatment can be potentially monitored and compared. Dynamic experiments may prove to be a sensitive way to quantitatively assess the microstructural components in the tooth. It is a priori knowledge of these features and their relations with mineralization level that will provide verification of the diagnostic power of PSOCT for caries and precarious lesions. Ultimately, clinical trials must be completed to study the sensitivity, specificity, and accuracy of PSOCT for dental caries diagnosis. This study is partially supported by National Institutes of Health (NIH) grant NIH 1R01-DE1154. We thank Puyun Guo of the Department of Electrical and Computer Engineering of the University of Connecticut for helping with electronics problems. We also appreciate the help of Joe Russi and Mark Pacelli from Haydon Switch and Instrument for the reference arm motion control. References Fig. 5. Images of one tooth sample in the mesial distal orientation obtained from (a) CP and (b) LP configurations. The images showed similarities in local features of dentin (longer arrows to the left of the images) and differences in global form birefringence patterns of enamel (shorter arrows to the right of the images). Image size is 4 mm long and 3.6 mm deep. tion related to the physiologic structure and pathogenic changes of enamel and dentin currently unavailable with any diagnostic or imaging method. PSOCT has the potential to be used for both dental research and clinic applications. As a relatively inexpensive and noninvasive system with high depth resolution, PSOCT also has the capability of threedimensional imaging for caries detection. Additional studies that correlate the changes in PSOCT ultrastructural features that occur in demineralization should provide important information 1. A. F. Fercher, W. Drexler, C. K. Hitzenberger, and T. Lasser, Optical coherence tomography principles and applications, Rep. Prog. Phys. 66, (2003). 2. 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, (1991). 3. 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, (1992). 4. J. F. de Boer, T. E. Milner, M. J. C. van Gemert, and J. S. Nelson, Two-dimensional birefringence imaging in biological tissue using polarization sensitive optical coherence tomography, Opt. Lett. 22, (1997). 5. X. J. Wang, T. E. Milner, and J. S. Nelson, Characterization of fluid flow velocity by optical Doppler tomography, Opt. Lett. 20, (1995). 6. G. J. Tearney, M. E. Brezinski, B. E. Bouma, S. A. Boppart, C. Pitris, J. F. Southern, and J. G. Fujimoto, In vivo endoscopic optical biopsy with optical coherence tomography, Science 276, (1997). 7. U. Morgner, W. Drexler, F. X. Krtner, X. D. Li, C. Pitris, E. P. Ippen, and J. G. Fujimoto, Spectroscopic optical coherence tomography, Opt. Lett. 25, (2000). 8. M. Wojtkowski, R. Leitgeb, A. Kowalczyk, T. Bajraszewski, and A. F. Fercher, In vivo human retinal imaging by Fourier domain optical coherence tomography, J. Biomed. Opt. 7, (2002). 9. H. Ren, Z. Ding, Y. Zhao, J. Miao, J. S. Nelson, and Z. Chen, Phase-resolved functional optical coherence tomography: simultaneous imaging of in situ tissue structure, blood flow velocity, standard deviation, birefringence, and Stokes vectors in human skin, Opt. Lett. 27, (2002). 10. S. H. Yun, G. J. Tearney, B. E. Bouma, B. H. Park, and J. F. de 10 April 2005 兾 Vol. 44, No. 11 兾 APPLIED OPTICS 2047

8 Boer, High-speed spectral-domain optical coherence tomography at 1.3 m wavelength, Opt. Express 11, (2003), M. J. Everett, K. Schoenenberger, B. W. Colston, Jr., and L. B. DaSilva, Birefringence characterization of biological tissue by use of optical coherence tomography, Opt. Lett. 23, (1998). 12. J. F. de Boer and T. E. Milner, Review of polarization sensitive optical tomography and Stokes vector determination, J. Biomed. Opt. 7, (2002). 13. J. F. de Boer, Z. Chen, J. S. Nelson, S. Srinivas, and A. Malekafzali, Imaging thermally damaged tissue by polarization sensitive optical coherence tomography, Opt. Express 3, (1998), C. Hitzenberger, E. Goetzinger, M. Sticker, M. Pircher, and A. F. Fercher, Measurement and imaging of birefringence and optical axis orientation by phase resolved polarizationsensitive optical coherence tomography, Opt. Express 9, (2001), B. Cense, T. C. Chen, B. H. Park, M. C. Pierce, and J. F. de Boer, In vivo depth-resolved birefringence measurements of the human retinal nerve fiber layer by polarization-sensitive optical coherence tomography, Opt. Lett. 27, (2002). 16. 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, (1999). 17. C. Saxer, J. de Boer, B. Park, Y. Zhao, Z. Chen, and J. Nelson, High-speed fiber based polarization-sensitive optical coherence tomography of in vivo human skin, Opt. Lett. 25, (2000). 18. B. W. Colston, U. S. Sathyam, L. B. Da Silva, M. J. Everett, P. Stroeve, and L. L. Otis, Dental OCT, Opt. Express 3, (1998), B. W. Colston, M. J. Everett, L. B. DaSilva, L. L. Otis, P. Stroeve, and H. Nathel, Imaging hard- and soft-tissue structure in the oral cavity by optical coherence tomography, Appl. Opt. 37, (1998). 20. A. Baumgartner, C. K. Hitzenberger, S. Dichtl, H. Sattmann, A. Moritz, W. Sperr, and A. F. Fercher, Optical coherence tomography of dental structures, in Lasers in Dentistry IV, J. D. B. Featherstone, P. Rechmann, and D. Fried, eds., Proc. SPIE 3248, (1998). 21. X.-J. Wang, T. E. Milner, J. F. de Boer, Y. Zhang, D. H. Pashley, and J. S. Nelson, Characterization of dentin and enamel by use of optical coherence tomography, Appl. Opt. 38, (1999). 22. A. Baumgartner, S. Dichtl, C. K. Hitzenberger, H. Sattmann, B. Robl, A. Moritz, A. F. Fercher and W. Sperr, Polarizationsensitive optical coherence tomography of dental structures, Caries Res. 34, (2000). 23. D. Fried, J. Xie, S. Shafi, J. D. B. Featherstone, T. M. Breunig, and C. Le, Imaging caries lesions and lesions progression with polarization-sensitive optical coherence tomography, J. Biomed. Opt. 7, (2002). 24. E. N. Marieb, Human Anatomy and Physiology, 5th ed. (Pearson Education, N.J., 2001), pp A. Gh. Podoleanu, Unbalanced versus balanced operation in a optical coherence tomography system, Appl. Opt. 39, (2000). 26. X. Wang and L. V. Wang, Propagation of polarized light in birefringent turbid media: a Monte Carlo study, J. Biomed. Opt. 7, (2002). 27. K. C. Hadley and A. Vitkin, Optical rotation and linear and circular depolarization rates in diffusively scattered light from chiral, racemic and achiral media, J. Biomed. Opt. 7, (2002). 28. J. Li, G. Yao, and L. V. Wang, Degree of polarization in laser speckles from turbid media: implication in tissue optics, J. Biomed. Opt. 7, (2002). 29. V. Sankaran, J. T. Walsh, and D. J. Maitland, Comparative study of polarized light propagation in biologic tissues, J. Biomed. Opt. 7, (2002). 30. H. Kitamura, M. Oda, and J. A Hess, Color Atlas of Human Oral Histology, 1st ed. (Ishiyaku EuroAmerica, St. Louis, 1992). 31. Y. Chen, L. Otis, and Q. Zhu, Study of tissue polarization with polarization-sensitive OCT by linearly and circularly polarized probing light, in Biomedical Optics, Proc. SPIE, (to be published) APPLIED OPTICS Vol. 44, No April 2005