Lasers in Surgery and Medicine 38: (2006)

Transcription

1 Lasers in Surgery and Medicine 38: (2006) Determination of Characteristics of Degenerative Joint Disease Using Optical Coherence Tomography and Polarization Sensitive Optical Coherence Tomography Tuqiang Xie, PhD, Shuguang Guo, PhD, Jun Zhang, PhD, Zhongping Chen, PhD,* and George M. Peavy, DVM, DABVP* Beckman Laser Institute, University of California, Irvine, California Background and Objectives: Previous studies have demonstrated that optical coherence tomography (OCT) could be used to delineate alterations in the microstructure of cartilage, and have suggested that changes in the polarization state of light as detected by OCT could provide information on the birefringence properties of articular cartilage as influenced by disease. In this study we have used both OCT and polarization sensitive optical coherence tomography (PS-OCT) technologies to evaluate normal and abnormal bovine articular cartilage according to established structural, organizational, and birefringent characteristics of degenerative joint disease (DJD) in order to determine if this technology can be used to differentiate various stages of DJD as a minimally invasive imaging tool. Materials and Methods: Fresh bovine femoral-tibial joints were obtained from an abattoir, and 45 cartilage specimens were harvested from 8 tibial plateaus. Whole ex vivo specimens of normal and degenerative articular cartilage were imaged by both OCT and PS-OCT, then fixed and processed for histological evaluation. OCT/PS- OCT images and corresponding histology sections of each specimen were scored according to a modified Mankin structural grading scale and compared. Results: OCT and PS-OCT imaging allowed structural evaluation of intact articular cartilage along a 6 mm surface length to a depth of 2 mm with a transverse resolution of 12 mm and an axial resolution of 10 mm. The OCT and PS-OCT images demonstrated characteristic alterations in the structure of articular cartilage with a high correlation to histological evaluation (k ¼ 0.776). The OCT images were able to demonstrate early to advanced structural changes of articular cartilage while the optical phase retardation images obtained by PS-OCT imaging were able to discriminate areas where disorganization of the cartilage matrix was present, however, these characteristics are much different than those reported where OCT images alone were used to characterize tissue birefringence. No evidence of differences in OCT or PS-OCT images were detected between specimens of similar structural characteristics where proteoglycan was judged present or absent by safranin-o Fast Green staining. Conclusions: The combined use of OCT and PS-OCT technologies to obtain images from a single system is able to demonstrate and discriminate between characteristics of very early stages of surface irregularities not previously reported for OCT imaging, to deep clefts and collagen matrix disorganization for tissue at depths of up to 2 mm with good correlation to histology. PS-OCT and accumulated optical phase retardation images of articular cartilage as constructed from alterations in Stokes vector parameters appear to give a valuable but different assessment of alterations in tissue birefringence and organization than have been reported for OCT images obtained with the use of polarized or non-polarized light sources. This is the first time that alterations in the polarization state of light reflected from within the tissue have been demonstrated to be consistent with changes observed in the orientation and organization of the collagen matrix in advanced stages of DJD. The degree of phase transformation of light reflected from within the tissue as determined by PS-OCT imaging does not appear to be altered by the presence or absence of proteoglycan. Lasers Surg. Med. 38: , ß 2006 Wiley-Liss, Inc. Key words: articular cartilage; birefringence; biomedical imaging; optical coherence tomography (OCT); optical imaging; osteoarthritis; phase retardation; polarization sensitive optical coherence tomography (PS-OCT) INTRODUCTION Degenerative joint disease (DJD) or osteoarthritis (OA) is one of the most frequent causes of physical disability among adults, affecting more than 20 million people in the United States [1]. The health and degeneration of the articular Contract grant sponsor: The U.S. Air Force Office of Scientific Research, Medical Free-Electron Laser Program; Contract grant number: F ; Contract grant sponsor: The National Center for Research Resources of the National Institutes of Health (Laser Microbeam and Medical Program); Contract grant number: RR-01192; Contract grant sponsor: National Institute of Health; Contract grant numbers: EB-00293, CA 91717; Contract grant sponsor: The Arnold and Mabel Beckman Foundation. *Correspondence to: Dr. George M. Peavy or Dr. Zhongping Chen, Beckman Laser Institute, University of California, 1002 Health Sciences Road East, Irvine, CA gpeavy@uci.edu or z.chen@uci.edu Accepted 12 July 2006 Published online 22 September 2006 in Wiley InterScience ( DOI /lsm ß 2006 Wiley-Liss, Inc.

2 DETERMINATION OF CHARACTERISTICS OF DEGENERATIVE JOINT DISEASE USING OCT AND PS-OCT 853 cartilage matrix is best understood when viewed as having two functional components, a solid matrix consisting of collagen and proteoglycans, and a fluid component consisting of water and ions. The fluid component provides the cartilage matrix with its visoelastic properties and load support, protecting the solid matrix from much of the functional load. Over time, shear and compressive forces coupled with the loss of proteoglycan content and the accompanying alterations in internal fluid dynamics result in breakdown of the cartilage matrix and failed repair [2]. Early and accurate diagnosis of injury, degeneration, and response of articular cartilage to treatment is essential for patient assessment and the development of successful strategies for early intervention. The structure of articular cartilage is recognized as being divided into zones that represent specific structural characteristics of collagen matrix and chondrocyte orientation. The collagen fibril network and the proteoglycans associated with it have similar refractive indices so are not easily differentiated by light microscopy, therefore, the zonal structure described for articular cartilage has been based largely on the morphology and distribution of chondrocytes. These zones are labeled as the superficial (tangential) zone, middle (transitional) zone, deep (radial) zone, and calcified cartilage or tidemark (Fig. 1). The collagen architecture is composed of Type II (hyaline) cartilage fibrils characterized by polarized light microscopy and electron microscopy studies as arising in the calcified cartilage zone, extending vertically toward the articular surface through the deep zone, arching toward a horizontal orientation in the middle zone, and are found in a horizontal orientation parallel to the articular surface in the superficial zone. Additionally, there is evidence of a separate but distinct 3 5 mm thick layer on the articular surface referred to as the lamina splendans [3 10]. In the early stages of DJD, focal lesions consisting of wrinkles, surface separations and erosions of the superficial zone, loss of proteoglycans and fibrillation of the collagen matrix result in the development of focal clefts that increase in width and depth until the full thickness of articular cartilage is destroyed. Once the subchondral bone and marrow become involved, local bleeding and blood clot formation may occur, providing a matrix and constituents for a localized spontaneous repair. The tissue of repair has a biochemical composition more like fibrocartilage than hyaline cartilage, and does not have either the organization of fibers or chondrocytes typical of normal articular cartilage [11 19]. A diagram of the progressive structural changes commonly observed in DJD is illustrated in Figure 1. Mankin proposed a histological-histochemical grading scale based on characteristics of structural alterations of the matrix, cellularity, safranin-o staining, and tidemark integrity that has been a common standard for the grading of sectioned specimens of articular cartilage [20]. Unfortunately, patient evaluation to the level of understanding provided by histological sectioning is limited by currently available imaging technology. Imaging with a conventional arthroscope and tactile probes can detect damage indicated by softness, fissuring, and advanced erosion of articular cartilage providing a general assessment of lesion size and textural changes that become evident with advanced DJD [21]. In addition to direct arthroscopic examination of articular surfaces, currently available morphologic imaging techniques including radiographs, computed tomography, magnetic resonance imaging, and ultrasound have been applied to the diagnosis of cartilage abnormalities [22 25]. However, the non-invasive diagnosis of early articular cartilage changes in clinical practice is limited by the Fig. 1. Schematic diagram of the structure of articular cartilage and structural changes characteristic of grading by a modified Manking grading scale defined by normal articular cartilage (DJD 0), superficial surface alterations (DJD 1), clefts into the superficial zone (DJD 2), clefts into the middle zone (DJD 3), clefts into the deep zone (DJD 4), clefts to the calcified zone (DJD 5), and complete disorganization of the collagen matrix (DJD 6). Note that in normal articular cartilage the hyaline cartilage fibers are highly organized and specifically oriented in each zone. The fibers are anchored in the calcified cartilage and extend upward through the deep zone in a direction perpendicular to the articular surface. At the middle zone the collagen fibers begin to transition in an arching pattern to a position that becomes parallel to the articular surface in the superficial zone.

3 854 XIE ET AL. inability to obtain images at a resolution of less than 100 mm with these technologies. Reliable, non-destructive techniques to image the microstructure of articular cartilage would be an important advancement in the ability to assess and monitor the management of cartilage abnormalities. Optical coherence tomography (OCT) is a relatively new biomedical imaging technique that enables non-invasive, in vivo cross-sectional imaging in turbid biological tissue at depths of up to 2 mm with resolution of up to 2 mm [26 29]. It is analogous to ultrasound, measuring the back reflection of infrared light rather than sound waves, and provides cross sectional images that have 10 times greater resolution. As a fiber optic-based technique OCT can be incorporated into or used through the operating channel of flexible or rigid scopes such as an arthroscope, to access small spaces for in vivo tissue imaging. Polarization sensitive optical coherence tomography (PS-OCT) imaging is a refinement of OCT technology that uses several different polarization states of incident light to obtain a series of depth-resolved Stokes parameter images from the backscattered light. The Stokes parameters are used to generate a depth-resolved optical phase retardation image that is characteristic of the birefringence of the tissue [30 33]. The measured optical phase retardation depends not only on the organization and alignment of the collagen matrix, but also on the relative orientations of tissue optical axis, incident light beam, and polarization directions. Recently, PS-OCT has been used to determine the orientation of the optical axis of biological tissue [31,34 37]. Because of its high resolution and sensitivity to alterations in tissue structure, OCT has been used for the imaging of joint cartilage [38 45]. These studies have demonstrated the ability to detect and characterize clefts (fissures) in articular cartilage that are consistent with a modified histological grading scale of structural alterations of articular cartilage in various stages of DJD. These studies have not attempted to detect early alterations at the articular surface that precede cleft formation into the superficial zone. With a resolution of mm, it was our expectation that OCT imaging should be able to demonstrate early alterations in the contour of the articular surface resulting from superficial laminar separations, wrinkles and fraying that can occur prior to the development of clefts. Interesting differences between healthy and degenerative articular cartilage specimens attributed to early changes in tissue birefringence are reported for OCT images obtained by the use of polarized or non-polarized incident light and a single channel OCT system [38,39,45]. The interpretation of birefringence and organization of articular cartilage in these reports, however, is substantially different than what has been demonstrated for normal articular cartilage where depth resolved optical phase retardation images obtained from a PS-OCT system were used to study equine articular cartilage [46]. The results of the latter study demonstrated no birefringence bands in optical phase retardation images of normal articular cartilage with the incident beam directed normal to the articular surface. In this study an optical fiber-based system is used to obtain co-registered OCT and PS-OCT images of normal articular cartilage and various stages of naturally occurring DJD in a bovine model to: (1) Determine if early disruptions of the articular surface that occur prior to cleft formation can be detected by OCT imaging and used as an early indicator of cartilage degeneration; (2) Determine if characteristics of articular cartilage birefringence in health and disease as reported for a single channel OCT system can be observed in accumulated optical phase retardation images obtained with a PS-OCT system where the phase retardation of backscattered light can be quantified and demonstrated; and (3) Determine if the presence or absence of proteoglycans in the matrix of articular cartilage has any discernable influence on PS-OCT optical phase retardation images. MATERIALS AND METHODS A schematic diagram of the single-mode fiber-based PS- OCT system used in this study is shown in Figure 2. Unpolarized low coherence light from a superluminescent diode (SLD) source (AFC, Toronto, Canada) with an output power of 10 mw, and a central wavelength of 1310 nm with a full-width-half maximum (FWHM) spectral bandwidth of 80 nm was coupled into a 22 fiber optic non-polarizing coupler which split the light into reference and sample arms. The axial resolution (Dz) of the OCT system is determined by the coherence length (l c ) of the light source [Dz ¼ l c /2 ¼ (2ln2/p) ( l 2 =Dl)], and is 10 mm for this system. The lateral resolution is determined by the diameter of the scanning light beam and is around 12 mm. An electro-optic (E-O) phase modulator was placed in the reference arm ahead of a rapid scanning optical delay (RSOD) and the offset of the scanning mirror was set to zero. The RSOD produced a group delay and the E-O modulator generated a stable carrier frequency at 500 KHz for heterodyne detection. Light returning from the reference RSOD passes through a polarizer that is mounted inside the phase modulator. The axis of the polarizer is oriented 458 with respect to the optical axis of the LiNO 3 crystal of the phase modulator. Four different polarization states of light of equal intensity are generated by the application of four different voltages (0, p/4, p/2, 3p/4) to the phase modulator. In the frame of the phase modulator, where the optical axis of the LiNO 3 crystal is chosen as the x axis, these four polarization states are separated by 458 angles in the UV plane of the Poincaré sphere. The different polarization states of the backscattered light collected in the sample arm are determined by coherence detection using the polarization states of light generated by the phase modulator in the reference arm. The choice of orthogonal polarization states in the Poincaré sphere for the reference arm is important because it ensures that the phase retardation measurements will be independent of the orientation of the optical axis in the sample. The polarization state of reflected light collected from the sample is altered by the birefringent properties of

4 DETERMINATION OF CHARACTERISTICS OF DEGENERATIVE JOINT DISEASE USING OCT AND PS-OCT 855 Fig. 2. Diagram of fiber-based PS-OCT system. Polarization modulator (Pol. Mod); polarization controllers (Pol. Control); phase modulator (Phase Mod); polarization beam splitter (PBS); rapid scanning optical delay line (RSOD); and detectors (D 1 &D 2 ). [Figure can be viewed in color online via the single mode optical fiber, thus we can only determine the changes in the orientation of optical axis, but not the absolute orientation of the tissue optical axis. Birefringence of the optical fiber results in a unitary transformation of the Poincaré sphere and thus does not affect the calculation of phase retardation [33]. For each reference polarization state, one A-line scan is performed, therefore, a total of four A-line scans are used to calculate the Stokes vectors. The polarization controller in the detection arm is turned to ensure that light from the RSOD is equally split into two detectors by the polarization beam splitter for each of the four polarization states. The two detected signals from the orthogonal polarization channels are filtered, amplified, digitized, and processed by computer to create a PS-OCT image. The four Stokes parameters I, Q, U, and V are calculated by: IðzÞ ¼a 2 x ðzþþa2 y ðzþ QðzÞ ¼a 2 x ðzþþa2 y ðzþ UðzÞ ¼2a x ðzþa y ðzþcos jðzþ VðzÞ ¼2a x ðzþa y ðzþsin jðzþ where a x and a y are amplitudes of two orthogonal components at depth z from two polarization detection channels, and j represents the phase difference between the two components. In the fiber-based PS-OCT system, the polarization state of light returning from within the tissue is compared to the polarization state at the tissue surface. In the Stokes vector images as shown in Figure 3b, the Fig. 3. OCT image (a) and Stokes vector images (b) of normal articular cartilage, its histological structure (c) and optical phase retardation image (d) which shows a very mild polarization change in the region mm immediately below the articular surface (d) (image size: 6mm 2.8 mm). [Figure can be viewed in color online via

5 856 XIE ET AL. vertical columns from left to right represent the four components of the Stokes vectors, I, Q, U, and V, respectively, and the horizontal rows from top to bottom represent four incident polarization states generated by the phase modulator. The axial orientation of the collagen fibrils in articular cartilage gradually changes by 908 as the fibrils extend from the deep zone through the middle zone to the superficial zone, and the depth of these zones may vary between joints and between locations within a single joint, which makes it very difficult to quantify local tissue birefringence as a function of depth by PS-OCT. For this study, we were interested in assessing the optical phase retardation changes in respect to disease states rather than quantifying alterations in tissue birefringence, therefore, we quantified the accumulative optical phase retardation using algorithms based on the assumption that the sample consists of only one birefringent layer of constant axis orientation. In this model, if the axis orientation changes with depth, an accumulated effect will be detected in the optical phase retardation image, and qualitative differences in tissue structure may be assessed by comparison of similarly acquired accumulated optical phase retardation images of specimens of the same tissue type. For a sample with an assumed linear birefringence of constant optical axis, there exist two eigenwaves that are polarized along the projected fast and slow axes of the sample normal to the propagation direction of incident light. The Stokes vectors of these eigenwaves determine a rotation axis in the equator plane of a Poincaré sphere. The effect of birefringence is to rotate the Stokes vector about this axis through an angle that is equal to the phase retardation of the sample. Conversely, the rotation axis can be determined from the known polarization states of incident and backscattered light at the sample location. The depth resolved accumulated optical phase retardation image is calculated by the rotation of the Stokes vectors in the Poincaré sphere [32,36,47]. Fresh bovine articular cartilage specimens were collected from the rear legs of adult (5 7 years old) Holstein dairy cows slaughtered within the prior 36 hours. Femoraltibial joints were isolated and dissected to expose the tibial plateau from which articular cartilage specimens were harvested from selected sites determined by visual examination to be either normal or undergoing some stage of degeneration. Full thickness samples that included subchondral bone were obtained for study by punch biopsy using a 1.0 cm diameter circular punch and mallet. A #11 scalpel blade was used to make two parallel full thickness cuts across the center of each cartilage specimen, 3 4 mm apart. The two outer semicircle shaped sections of cartilage were undercut and removed from the specimen, leaving the central rectangular section of cartilage attached to the bone. This configuration helped to maintain consistent orientation of the specimen between OCT/PS- OCT imaging and histological sectioning following fixation and tissue processing. Each prepared specimen was wrapped in a 0.9% saline soaked gauze sponge, placed in a sealed and marked plastic bag, and held at room temperature until imaged on the same day. At the time of imaging, each specimen was removed from its protective wrapping, the base set in a 2 cm2 cm piece of paraffin for positioning and stabilization during imaging, and then placed in a 2 cm round disposable plastic Petri dish. The Petri dish was positioned under the controlled motorized sample arm of the imaging system so that the path of the scanning beam, as evidenced by illumination from a superimposed 670 nm aiming beam, was directed along the surface of the cartilage specimen in a longitudinal direction at the specimen s center. The Petri dish was then filled with 0.9% saline solution up to just the specimen surface. After OCT imaging, each specimen was fixed in 10% buffered formalin. Following tissue fixation each specimen was decalcified by being placed into 8% formic acid and checked daily until soft enough to section. Samples were then dehydrated in progressive concentrations of ethanolwater, cleared and embedded in paraffin. The embedded specimens were cut in 6 mm thick serial sections made along the length of each specimen at its center where the OCT images had been obtained, and carefully placed on clean glass slides. The sections were deparaffinized, stained in Hematoxylin (Sigma-Aldrich, St. Louis, MO) and Eosin-y (H&E) (EM Science, Gibbstown, NJ), or safranin-o counter stained with Fast Green FCF (SOFG) (Sigma) and coverslipped. The histology slides were imaged using a microscope with a real-time digital color imaging camera (Microfire C, Optronics, Goleta, CA) coupled to a computer and recorded using image capture software (Picture Frame, Optronics, Goleta, CA). Because there can be substantial natural variability in the histologic features of degenerative articular cartilage as observed in adjacent tissue sections of the same specimen, it was important to select histology sections for evaluation that were representative of the section of the specimen imaged by OCT and PS-OCT without compromising the independent grading of the images. The H&E stained serial histology sections of each specimen were examined by light microscopy, and the section best matching the surface profile of the OCT image as evidenced by surface contour, and the presence, number and location of defects, was selected for separate evaluation under higher magnification and image capture at a later time. The OCT images were then grouped separately from the selected tissue sections and a second individual evaluated and graded each group independently. Without knowledge of the sample histology, OCT and PS- OCT images were reviewed and specimens sorted according to structural appearance. Based on the OCT image the specimen was assigned a grade DJD 0 5 based on a modification of the Mankin grading scale for structural alterations [20] as outlined in Table 1 and illustrated by diagram in Figure 1. Mankin s grading system lumps all surface irregularities into grade 1 or 2 depending on whether or not they are accompanied by vascular proliferations referred to as pannus. Mankin s grade 3 is the first grade of the system that includes the presence of a cleft (fissue) and the cleft is defined as extending into the middle

6 DETERMINATION OF CHARACTERISTICS OF DEGENERATIVE JOINT DISEASE USING OCT AND PS-OCT 857 TABLE 1. Modified Mankin Structural Grading Scale for Evaluation of Histology Sections and OCT/PS-OCT Images Histological characteristics Mankin grade [20] Modified Mankin grade Normal structure 0 DJD 0 Surface irregularities 1 Surface irregularities: wrinkles, superficial laminar separations, surface DJD 1 fraying, with no clefts Surface irregularities: clefts limited to the superficial zone DJD 2 Surface irregularities with pannus 2 Clefts into the middle (transitional) zone 3 DJD 3 Clefts into the deep (radial) zone 4 DJD 4 Clefts to the calcified zone (tidemark) 5 DJD 5 Complete disorganization 6 DJD 6 zone. Pannus was not observed in any of the joint specimens obtained for this study, so the grading scale was modified to eliminate this criterion. Because an objective of this study was to determine if OCT imaging could be used to detect early structural changes in articular cartilage, we further modified the Manking grading system to differentiate superficial wrinkles, separations and fraying (modified grade 1) from surface irregularities with clefts limited to the superficial zone (modified grade 2) from clefts that extended into the middle zone (modified grade 3). Characteristics used for grading of the OCT images used in this study were smooth curvilinear (DJD 0) or roughened (DJD 1) surface, and clefts by greatest apparent depth (DJD 2 5). The PS-OCT images were then used to sort the specimens into two subgroups according to either the presence or absence of alterations in optical phase retardation as evidenced by the presence or absence of banding in the PS-OCT Stokes vector and accumulated optical phase retardation images. The specimen was reassigned a grade of DJD 6 if the OCT image demonstrated any surface irregularities and multiple bands were observed in the PS- OCT Stokes vector and optical phase retardation images. Without knowledge of the combined OCT/PS-OCT grading of the specimen, each histology section was evaluated for surface irregularities, and clefts were characterized according to the depth and zone to which they extended into the cartilage specimen. Each specimen was evaluated for alterations of the organization of the cartilage matrix as evidenced by the orientation of collagen fibers and chodrocytes. Additional sections from 20 randomly selected specimens previously graded DJD 0 or DJD 1 by histology evaluation were stained with SOFG, evaluated and graded for proteoglycan content as being either positive (normal), reduced (25% 75%) or negative (minimal to no) as evidenced by presence and distribution of safranin-o staining. Agreement between the assigned histologic grade and the grade based upon the combined evaluation of the OCT and PS-OCT images was assessed by use of the kappa (k) statistic [48]. The statistical significance of k is determined by a z statistic with z ¼ k/se where SE is the standard error. A statistically significant z reflects greater agreement than expected by chance. A k > 0.75 is considered an indication of strong agreement. RESULTS A total of 45 specimens of articular cartilage from eight bovine legs were imaged by the OCT/PS-OCT system and processed for histology evaluation, grading and comparison to OCT/PS-OCT images as described. Histology grading of sections matched to OCT images was used as the standard by which the accuracy of initial OCT/PS-OCT grading was judged. The data and results of comparison of grading by the two techniques are presented in Table 2 and representative images are presented in Figures TABLE 2. Summary and Comparison of OCT Grading Compared to Histological Grading of Degenerative Joint Disease Characteristics Observed in 45 Specimens of Bovine Articular Cartilage Using a Modified Mankin Grading Scale Histology grade OCT grade DJD 0 DJD 1 DJD 2 DJD 3 DJD 4 DJD 5 DJD 6 DJD 0: n ¼ DJD 1: n ¼ DJD 2: n ¼ DJD 3: n ¼ 3 3 DJD 4: n ¼ DJD 5: n ¼ 1 1 DJD 6: n ¼ 3 3

7 858 XIE ET AL. Fig. 4. OCT image (a), Stokes vector images (b), a histology image (c), and optical phase retardation image (d) of articular cartilage graded DJD 1 because of irregularities to the surface caused by wrinkles, laminar separations, and superficial erosions without the formation of clefts (6 mm 2.8 mm). [Figure can be viewed in color online via wiley.com.] Histological evaluation determined seven specimens to be DJD 0 (normal), six (85.7%) of which had been judged to be DJD 0, and one had been judged to be DJD 1 by OCT. Eighteen histology specimens were found to have surface irregularities without clefts consistent with a grade of DJD 1, of these 14 (77.8%) had been graded DJD 1, while 3 had been graded DJD 0 and 1 had been graded DJD 2 from OCT images. Evaluation of histology sections found five specimens that had clefts that were limited to the superficial zone so were graded DJD 2, and of these four (80%) had been graded DJD 2, and one had been graded DJD 1 from OCT images. Additionally, three specimens were graded DJD 3 by histology examination because they demonstrated clefts to the middle zone, and all three (100%) had been graded DJD 3 from OCT images. A grade DJD 4 was assigned to sections from nine specimens that revealed clefts to the deep zone, eight (88.9%) of which had been judged to be DJD 4, and one had been judged to be DJD 5 from the OCT images. Only one histology section was found to have clefts to the calcified cartilage zone and graded DJD 5, and this specimen was incorrectly graded DJD 4 from the OCT image. Of all the specimens examined, three demonstrated alterations in organization in collagen matrix consistent with a grade DJD 6, and all three (100%) had been judged from PS-OCT images as having matrix alterations. Overall 38 (84.4%) of 45 specimens had been given the same grade by histology evaluation and OCT/PS-OCT imaging. When assessed by the kappa statistic, strong agreement between OCT and histology is noted with k ¼ (SE ¼ 0.072; z ¼ , P < 0.001). Due to limitations on the depth of imaging, the calcified zone could not be imaged by OCT/PS-OCT in articular cartilage specimens that were more than 2 mm in thickness. Figure 3 presents OCT (a), PS-OCT (b), and accumulated optical phase retardation (d) images of normal adult bovine articular cartilage (DJD 0). The OCT image demonstrates a smooth surface without apparent irregularities or defects (Fig. 3a) that is consistent with the morphology observed in the histology sections of the area imaged (Fig. 3c). Fig. 5. OCT image (a), Stokes vector images (b), histology image (c), and optical phase retardation image (d) of articular cartilage demonstrating clefts into the superficial zone and graded DJD 2 (6 mm 2.8 mm). [Figure can be viewed in color online via wiley.com.]

8 DETERMINATION OF CHARACTERISTICS OF DEGENERATIVE JOINT DISEASE USING OCT AND PS-OCT 859 Fig. 6. OCT image (a), Stokes vector images (b), histology image (c), and optical phase retardation image (d) of articular cartilage graded DJD 3 because clefts extending into the middle zone (6 mm 2.8 mm). [Figure can be viewed in color online via wiley.com.] This normal cartilage specimen demonstrates mild optical phase retardation down to approximately mm below the articular surface as observed in the PS-OCT images of Figure 3b and the optical phase retardation image in Figure 3d. Surface irregularities without obvious clefts can be seen in the OCT image of Figure 4a. These are consistent with wrinkles, separations, and superficial erosions of the surface laminar layers of the superficial zone as observed in the histology section. These structural changes give the OCT image a moth-eaten appearance, and was classified DJD 1 according to our modified Mankin structural grading scale. More advanced erosions or clefts into the superficial zone (Fig. 5) can be distinguished from superficial surface irregularities (Fig. 4) and are characteristic of a grade DJD 2. None of the specimens collected for imaging had vascular proliferations consistent with pannus, so we were unable to assess the ability to image this pathological change that Mankin included in the characterization of a structural grade 2. In Figure 6a, the OCT image shows a cartilage cleft to the middle zone consistent with a classification of DJD 3 that is confirmed by histology (Fig. 6c). A cleft extending into the deep zone is observed in the OCT and histology images of Figure 7 consistent with a DJD 4 grade, and the irregularity of surface erosion and fibrillation can be seen to the right of the cleft. The absence of banding in the PS- OCT images of this specimen (Fig. 7a,b) indicate that the collagen matrix in the area under the surface erosion and fibrillation of this specimen has not yet undergone reorganization as is observed in the specimen of Figure 9. The OCT image of articular cartilage in Figure 8 demonstrates a cleft that was found on histology evaluation to extend to the calcified zone consistent with a classification of DJD 5. The PS-OCT Stokes vector and accumulated optical phase retardation images of three specimens demonstrated banding typical of optical phase retardation observed in tissues with strong birefringence properties. When viewed on histological evaluation, all three of these specimens Fig. 7. OCT image (a), Stokes image (b), histology image (c), and optical phase retardation image (d) of articular cartilage graded DJD 4 demonstrating a cleft into the deep zone. The adjacent cartilage surface is eroded and fibrillated, but demonstrates only a mild polarization shift (6 mm 2.8 mm). [Figure can be viewed in color online via

9 860 XIE ET AL. Fig. 8. OCT image (a), Stokes vector image (b), histology image (c), and optical phase retardation image (d) of articular cartilage demonstrating a cleft into the calcified zone and graded DJD 5. Although the histology section demonstrates that the cleft extends to the calcified zone, the incomplete separation of the cleft margins in the area imaged by OCT only detected the cleft into the middle zone. Similar to other specimens graded DJD 0 1, only a mild polarization change is observed in this specimen (6 mm 2.8 mm). [Figure can be viewed in color online via demonstrated disorganization of the normal collagen matrix consistent with a structural classification of DJD 6. There were no specimens graded DJD 0 5 by histology evaluation that demonstrated such marked optical phase retardation changes on PS-OCT imaging. Exaggerated surface irregularities and some subsurface pockets of tissue separation are seen in the OCT image of the specimen in Figure 9. These changes are similar to the surface changes observed in the specimen of Figure 7 and represent erosion and fibrillation of the articular surface. In contrast to the specimen of Figure 7, however, the PS-OCT images of Figure 9 demonstrate a marked banding pattern where disorganization or, more likely, a re-orientation of the collagen matrix has altered the polarization state of light backscattered from the tissue (each band represents 1808 phase shift). The specimen presented in Figure 10 was graded DJD 6 on histological evaluation because the right half of the specimen demonstrates disorganization of the cartilage matrix that may be due to reorientation of the collagen matrix from erosion and compression or, more likely, because of fibrocartilage replacement of the hyaline cartilage matrix. The PS-OCT images clearly demonstrate a banding pattern consistent with two complete 3608 Fig. 9. OCT image (a), Stokes vector images (b), H&E stained histology images (c, e, f), optical phase retardation image (d), and SOFG stained histology image of articular cartilage (g) (image size: 6 mm2.8 mm). The Stokes vector and phase retardation images of this specimen show strong birefringence indicating a change in collagen matrix organization or orientation. The magnified (40) H&E stained histology images demonstrate the disorganization or reorientation of the collagen fibrilar network (e) compared to an area of more normal vertical fibrillar alignment (f) of the deep zone. [Figure can be viewed in color online via wiley.com.]

10 DETERMINATION OF CHARACTERISTICS OF DEGENERATIVE JOINT DISEASE USING OCT AND PS-OCT 861 Fig. 10. OCT image (a), Stokes vector image (b), H&E stained histology image (c), and optical phase retardation image of articular cartilage (image size 6 mm2.8 mm) (d). The specimen was graded DJD 6 on histological evaluation because the right half of the specimen demonstrates disorganization of the cartilage matrix due to fibrocartilage regeneration or phase shifts (four bands) on the right side of the specimen that correspond to the region in the histology specimen where the matrix is disorganized. To further investigate the influence of matrix degeneration on phase retardation, a graph of accumulated optical phase retardation as a function of depth was created for Fig. 11. Graph of phase retardation as a function of depth for articular cartilage graded by histology to be normal cartilage (B); DJD 1 (C); DJD 2 (D); DJD 3 (E); DJD 4 (F), DJD 5 (G); and DJD 6 (H). For all specimens DJD 0 through DJD 5 (B G) the phase retardation plots reveal a similar mild polarization shift demonstrated by an increasing slope of the plot to a depth of approximately 750 mm and then stabilizes. The DJD 6 specimen, on the other hand, reveals strong optical phase retardation (Fig. 9) manifest as a wide, undulating pattern on the graph (H) as the phase retardation plot reveals three complete 3608 polarization shifts of the incident light. [Figure can be viewed in color online via replacement of normal articular cartilage. The PS-OCT images demonstrate an alteration in the cartilage matrix that corresponds to the region in the histology specimens where normal matrix organization is disrupted. [Figure can be viewed in color online via specimens of each grade DJD 0 6 (Fig. 11). The phase retardation value for each depth was calculated from the PS-OCT imaging data as an average of the phase retardation values of 20 pixels adjacent to each other at the same depth in a sample area selected on an OCT image for structural orientation and then transferred to the PS-OCT image for data acquisition. The sample area used for optical phase retardation calculation was selected adjacent to a representative cleft since the polarization state of the light backscattering from a cleft surface is difficult to define with accuracy. The phase retardation plots for normal (DJD 0) and degenerative (DJD 1 5) specimens revealed a mild polarization shift (estimated to be approximately 10%) in the first 750 mm of depth from the articular surface and then stabilized (Fig. 11B G). The DJD 6 specimen manifesting banding in the PS-OCT and accumulated optical phase retardation images seen in Figure 9 demonstrates marked polarization shifts as seen by a wide undulating pattern on the phase retardation plot indicative of three complete 3608 polarization shifts of the incident polarized light (Fig. 11H). A similar graph (Fig. 12) was made to compare specimens where the matrix of the articular cartilage appeared minimally disrupted (DJD 0 1) but where proteoglycan content as evidenced by safranin-o staining was judged to be either positive (normal), reduced (25 75%), or negative (minimal to no staining). Similar phase retardation patterns were found for all of these specimens, suggesting that proteoglycan presence or absence has no apparent influence on the phase retardation as detected by PS-OCT imaging, and that phase retardation of the returning light as detected by PS-OCT is principally due to differences in the orientation or organization state of the collagen matrix of the specimen in relation to the axis of the incident beam of light. DISCUSSION As a new morphological imaging technology OCT using non-polarized and polarized light sources has been reported

11 862 XIE ET AL. Fig. 12. Graph of phase retardation as a function of depth (a) comparing cartilage specimens of an apparent normal matrix structure that contain proteoglycan (b) to structurally similar specimens that are proteoglycan reduced (c) or have little to no proteoglycan present (d). The phase retardation graph (a) shows very similar optical phase retardation slopes for proteoglycan positive (P), reduced (R), and negative (N) specimens suggesting that the presence or absence of proteoglycan has little to no influence on the optical phase retardation. [Figure can be viewed in color online via www. interscience.wiley.com.] to be a promising new tool for the imaging and assessment of DJD [38,39,44,45]. Based on results of polarization microscopy, previous studies in articular cartilage using conventional OCT techniques have indicated a potential for PS-OCT imaging to differentiate normal cartilage from early DJD by differences in their birefringence properties [39,45]. In this study we have attempted to further evaluate the imaging capabilities of this technology to detect characteristics of articular cartilage in various states of health and degeneration to a level of discrimination comparable to histological evaluation using a system that is able to concurrently acquire both an OCT image and depth resolved four vector PS-OCT data on optical phase retardation as influenced by tissue birefringence. Overall 38 (84.4%) of 45 specimens were given the same grade by histology evaluation and OCT/PS-OCT imaging, with strong agreement confirmed by k statistic. Specimens in this study that differed in grade between histology evaluation and evaluation of OCT/PS-OCT images were within a single grade of difference, and these variations were attributable to either the ability to detect subtle differences between grades (DJD 0 2), clefts that approached the boundary of morphologic zones that are not easily distinguished by OCT imaging (DJD 2 4), and clefts that extend beyond the image depth in specimens that are more than 2 mm thick (DJD 4 5). The ability to image and characterize cleft development in articular cartilage as demonstrated in this study is consistent with the findings of the one previous study that correlated OCT images to structural grading of histology sections based on a modified Mankin grading scale [44]. This study further showed, however, that early structural changes in the superficial zone can be detected by OCT imaging prior to the manifestation of structural clefts. Although the grading of OCT images and matched histology sections was done independently and blinded, the selection of histology sections for review based on similarity of surface contour and characteristics to OCT images could have added some positive bias to the agreement between the OCT image and histology evaluations. Because the characteristics of DJD may be expressed to various degrees within a small region of a joint and the inherent variables associated with histology tissue sectioning, we believe that the failure to match each OCT image to a histology section that demonstrated similar characteristics of surface contour and the presence, number and location of defects would have lead to a much stronger and misrepresentative negative bias to the study. For the methods employed to have introduced a significant positive bias into the study a substantial number of specimens would have had to have had adjacent tissue sections that were more representative of the OCT image than the tissue actually under the 12 mm wide OCT beam path. In contrast to the conclusions of another group [38,39,45], the PS-OCT Stokes vector images (Figs. 3 10) and accumulated optical phase retardation data (Fig. 11) obtained in this study demonstrated that normal (DJD 0) and abnormal articular cartilage specimens (DJD 1 5) had little polarization sensitivity and were not able to be differentiated by their birefringence properties until there was a major alteration in matrix orientation or composition in the later stages of the degenerative process. The specimens presented in Figures 9 and 10 demonstrate that the banding pattern typical of alterations in

12 DETERMINATION OF CHARACTERISTICS OF DEGENERATIVE JOINT DISEASE USING OCT AND PS-OCT 863 the polarization state of light as influenced by tissue birefringence is observed in the PS-OCT Stokes vector images (Figs. 9b and 10b) and in the accumulated optical phase retardation image (Figs. 9d and 10d), but are not demonstrated by the OCT imaging (Figs. 9a and 10a). These specimens illustrate the ability of PS-OCT imaging to demonstrate alterations in articular cartilage matrix orientation and organization where surface visualization or OCT imaging alone may not be able to appreciate these matrix alterations. The initial model of collagen fibril orientation of articular cartilage proposed an arching orientation in the middle zone with return of fibrils to the calcified cartilage zone and a separate layer of horizontal fibrils forming the superficial zone [3]. Organization of the collagen fibrils in the middle zone and a distinctly separate fibrillar network comprising the superficial zone has been controversial, however, work by Clark [6,8] and Jeffery et al. [9] has demonstrated that the structural matrix of articular cartilage is composed of collagen fibrils that are anchored in the calcified cartilage zone, travel vertically through the deep zone grouped in sheets that are separated by vertical rows of chondrocytes encased in fibrillar capsules, and that these sheets arch from their vertical orientation as they approach the surface, overlapping and continuing in a horizontal (tangential) orientation in the superficial zone. In the superficial zone the overlapping fibril layers are flattened into thin, tightly apposed lamellae 1 3 fibrils thick, and fibrils of adjacent lamellae cross one another at angles. In contrast to polarized light microscopy of transversely illuminated histology sections cut 908 to the articular surface, PS-OCT imaging is done in a reflection mode on intact tissues with light entering the cartilage at a 908 angle to the articular surface and traveling in an axial direction toward the subchondaral bone. In this orientation, the light may be expected to encounter some phase retardation in the superficial zone and perhaps the middle zone where it transects the collagen fibrillar network, but should not be expected to encounter much phase retardation when passing in an axial direction that is essentially parallel to the fibrillar orientation of the collagen matrix in the deep zone of articular cartilage unless that matrix has been altered by degeneration or disease. This hypothesis is consistent with studies that have used PS-OCT imaging to demonstrate changes in birefringence patterns relative to the incident angle of illumination of the collagen matrix in normal articular cartilage [46], as well as by the observations reported in this study. The optical phase retardation as observed by PS-OCT imaging in this study is consistent with the results anticipated from the model described above, but is in contrast to previous reports where polarized and nonpolarized light sources were used in conjunction with a single channel OCT system to assess alterations in articular cartilage birefringence in early stages of degeneration. An explanation for the differences in the observations of these reports is not yet clearly apparent, however, several factors could affect the results. First is location within the joint from which an image is obtained. Normal alterations in the collagen matrix orientation and structural composition that occur near the periphery of the joint surface [8,49] may alter the tissue birefringence or optical axis of the matrix in respect to the axis of the imaging beam, providing a different phase retardation image of normal cartilage from this region than is obtained from normal cartilage at the center of the joint surface. Second, cartilage specimens used in this study and judged as being normal might have sustained early alterations that were not evident on histologic evaluation but influenced a reduction in tissue birefringence. Third, there could be unidentified differences in articular cartilage related to species of origin (human vs. bovine) that have not been recognized. Fourth, position of the imaging beam is important as variation of the incident angle of the imaging beam away from the normal will increase the optical phase retardation of the image acquired [46]. Finally, imaging artifacts may occur when using a system that is not designed to compensate for such things as phase alteration caused by bending of the optical fiber during image acquisition. Fiber bending during image acquisition can result in a banding pattern in OCT images obtained with a single channel system even when imaging non-birefringent tissue. The PS-OCT system used in this study eliminates the possibility of this artifact by utilizing a two channel detector design and the optical phase retardation image generated is derived from multiple measurements that use multiple incident Stokes vectors. Because of the possibility that cartilage specimens used in this study and judged as being normal by histology might have had early alterations in matrix organization that were not evident in the histology but that might influence a reduction in tissue birefringence properties as has been suggested by other investigators [38,39,45], we screened our approach for this possibility. We randomly selected three specimens from our study group that had been judged by histological evaluation to be normal (DJD 0), and examined unstained, 6 mm thick vertically cut and mounted tissue sections from these specimens by polarized light microscopy in the manner described in the previous reports [38,39]. Tissue birefringence as manifest by phase retardation in these vertically cut specimens was observed in the superficial and deep zones of all three specimens consistent with the criteria used in the previous reports to classify these specimens as being normal articular cartilage. This study demonstrates an advantage to obtaining both OCT and PS-OCT images when making an assessment of articular cartilage. Because of the accuracy in the ability to define structural alterations but difficulty in determining boundaries of specific morphologic zones of articular cartilage by OCT/PS-OCT imaging, it seems that an independent grading system for OCT/PS-OCT imaging based on apparent alterations of the superficial zone, quantitative assessment of cleft number and depth, and evaluation of matrix organization would be appropriate. It is also realized that an OCT image is representative of only the narrow vertical plane of the imaged tissue, and that cleft depth or other structural alterations may vary