The Application of Optical Coherence Tomography in Musculoskeletal Disease

Many musculoskeletal disorders (MDs) are associated with irreversible bone and cartilage damage; this is particularly true for osteoarthritis (OA). Therefore, a clinical need exists for modalities which can detect OA and other MDs at early stages. Optical coherence tomography (OCT) is an infrared-based imaging, currently FDA approved in cardiology and ophthalmology, which has a resolution greater than 10 microns and acquisition rate of 120 frames/second. It has shown feasibility for imaging early OA, identifying changes prior to cartilage thinning both in vitro and in vivo in patients and in OA animal models. In addition, OCT has shown an ability to identify early rheumatoid arthritis (RA) and guide tendon repair, but has the potential for an even greater impact. Clinical trials in OA are currently underway, as well as in several other MDs.


Introduction
Musculoskeletal diseases are one of the leading causes of disability in the United States. Fiy million adults in USA have been diagnosed with arthritis, rheumatoid arthritis, gout, systemic lupus erythematosus, or �bromyalgia, and approximately 1 in 3 people between ages 18 to 64 with diagnosed arthritis have work limitations [1,2]. Similarly, disease of the periarticular structures, such as tendons and ligaments, contributes to additional disabilities and limitations. Fullor partial-thickness tears of rotator cuff tendons (RCT) are relatively common; they occur in approximately 30% of the population and represent around 4.5 million clinic visits and 40,000 surgeries in the USA per year [3]. Arthritis also affects the pediatric population, with an estimate 294,000 children under the age of 18, or 1 in every 250, having some form of arthritic or rheumatologic condition [4]. Studies estimate that by the year 2030, 67 million Americans older than 18 years will have doctor-diagnosed arthritis [5].
is paper examines the potential of the new micronscale imaging technology and optical coherence tomography (OCT), for the management of musculoskeletal disease. It focuses on the existing clinical need for a high-resolution micron-scale imaging system in the �eld of orthopedics, including osteoarthritis (OA), rheumatoid arthritis (RA), and rotator cuff repair (RCR). However, this is far from the full extent of potential applications. Early detection of disease, understanding early disease markers, and accurate assessment of tissue microstructure are necessary to increase success of treatment, reduce patient morbidity, and determine the progress of future therapeutics in hopes of improving patient outcomes.

The Developments and Advantages of OCT Nontransparent Tissue Imaging
Based on technology from telecommunications, optical coherence tomography (OCT) systems have been modi�ed to accommodate many medical specialties, ranging from ophthalmology and cardiology to orthopedics. e �rst two applications are FDA approved and are routinely used in patients. Since 1994 our lab has been developing OCT for nontransparent tissue, from engineering and physics to clinical trials.
Analogous to ultrasound, OCT is an imaging modality that utilizes infrared light rather than sound. As infrared light is generated, it is split into both a sample and reference arm. Light re�ecting back from the sample combines with light from the precisely controlled reference arm mirror. e intensity of interference is measured using a technique called low-coherence interferometry, with this intensity plotted as a function of tissue depth as the reference arm mirror is moved. As the beam is scanned across the surface of the tissue, two and three dimensional data sets are derived [14]. ese image sets can then be used to interpret the microstructure of the tissue.
Current clinical imaging modalities such as MRI, CT, and X-ray are critical to the diagnosis of disease. However, early disease diagnosis requires micron-scale, real-time imaging. OCT has shown many advantages over current imaging modalities such as a resolution 25x higher than any clinical subsurface imaging modality, a data acquisition speed of 120 high-resolution images per second, and the ability to perform imaging through a 0.017-inch �ber-optic endocatheter with automated probe pull-back [15][16][17]. ese characteristics make OCT an ideal system for the clinical setting in terms of speed and volumes of information gathered on both target and auxiliary structures. Additionally, the OCT engine is comparative in size to a portable ultrasound machine, which allows easy transport into procedure rooms ( Figure 1). ese advantages make OCT an ideal tool for both the clinical and research settings.
Finally, OCT can be combined with many adjuvant techniques: Doppler OCT, OCT elastography, OCT spectroscopy, second-order correlation (SOC) OCT, and polarization-sensitive OCT (PS-OCT). ese adjuvant technologies are described in further detail in the appendix. For the purposes of this paper, we will examine the ability of OCT, focusing primarily on structural OCT and PS-OCT, to identify and track early disease progression and its potential for the early diagnosis of OA, RA, and RCR. to increase both proliferation and synthetic and degradative efforts when disrupted [18][19][20][21][22][23]. e signaling cascade is started by damage to the cartilage, either through a single macrotraumatic event or several microtraumatic events, that fragment and release proteoglycans and type II collagen into the synovial �uid of the joint [24][25][26][27][28][29][30][31]. is may occur either at the surface or the cartilage-subchondral bone interface. Although the chondrocytes react at the site of articular damage and release degradative enzymes in hopes of selfrepair, such as metalloproteases, the effort is generally futile [22,[32][33][34]. ese chemical changes in the joint only amplify the amount of damage to the articular cartilage, which reacts abnormally. Under previously normal biomechanics and load characteristics, cartilage is more easily deformed, and �uid loss is increased by the loss of glycosaminoglycans [35,36]. Advanced osteoarthritis is visualized by thickening of subchondral bone and synovium, bone spurs at the joint periphery, and feathering/full thickness defects in the articular surface [37,38]. However, by this stage of OA advancement, reversal is unlikely. us, it is imperative to diagnose OA early. Predictive early changes in an arthritic joint consist of the loss of an organized collagen and glycosaminoglycan matrix, consistent with less water retention in articular cartilage with increased load [7,39]. Since the OA cascade is so detrimental to the cartilage, it is optimal to diagnose the disease in its early manifestations so that preventative therapies can be initiated.

OCT's Contribution
In the initial years, extensive analysis was performed to compare and match OCT imaging with histopathology of several tissue types to con�dently make distinctions between normal and diseased tissue states. With an increasing library of data and literacy of the OCT images, pathological Polarization-sensitive OCT (PS-OCT) imaging is particularly attractive in osteoarthritis detection. It is capable of assessing both the general structure and collagen organization of the most relevant tissue types, including cartilage and tendon. Birefringence, used to assess collagen organization, is a characteristic of tissues with highly organized structures, which decompose and modify the polarization state of incident light. is birefringence re�ects the degree of microstructure organization in tissue containing birefringent components such as collagen, as the disruption and loss of collagen in many pathological processes results in the loss of polarization sensitivity [6,12,14,[40][41][42][43][44]. is is crucial to assess the extent of OA, as the disruption of highly organized tissue, particularly collagen, is an initial marker of OA.
Normal human cartilage imaged by PS-OCT is presented in Figure 4 [6,12,40,41,45]. PS-OCT imaging (Figures 4(a)-4(c)) shows the positions of the bands change with the manipulation of the incident light polarization state, moving with a homogeneous rate across the sample without signal intensity dropout. is characteristic is attributed to the highly organized-collagen, which gives the cartilaginous tissue birefringent properties, and is veri�ed with histopathology ( Figure 4(e)). e bright yellow homogeneous picrosirius staining represents highly organized type II collagen-laden tissue, con�rming the absence of OA. In contrast to Figure 4, Figure 5 demonstrates the initial loss of organized collagen, indicating initial disease progression [40]. Comparing this PS-OCT imaging in a mildly diseased cartilage sample (Figures 5(a)-5(c)) with Figure 4, a banding pattern is seen that shis as the polarization state is manipulated, but does not move together in a linear fashion. ese areas of signal intensity dropout and banding pattern nonuniformity are one of the �rst markers for the loss of parallel structure, decreased striation, and irregular shape of the collagen structure. Again con�rmed with histopathology ( Figure 5(e)), areas of collagen loss are represented by the loss of the intense yellow hue present in the previous histological sample. In both Figures  2 and 3, the hematoxylin and eosin stains show a smooth, homogeneous surface with full thickness and without lacunar proliferation, representing a visually grossly normal articular surface cross-section. is ability of PS-OCT to produce high-resolution optical biopsies of the tissue microstructure makes this imaging modality tremendously promising in the assessment of early OA. It is also performed in real time. Figure 6 shows normal and pathologic polarization changes in vivo [8]. is image series of in vivo human knee cartilage shows a banding pattern discrepancy between the le and right sides; bands become less homogeneous with large areas of signal intensity dropout as you move right. e far right not only shows loss of homogeneous and tight banding patterns but also shows a nonuniform articular surface. is loss of birefringence, or polarization sensitivity, again represents an early marker of OA. Note that normal tissue and diseased tissue are directly adjacent to each other and how the scanning beam of PS-OCT immediately captures the discrepancy in a single imaging frame. is capability of assessing early disease has great potential for early interventions.
An OCT endocatheter has recently been developed, which can be introduced into the human knee joint while patients undergo knee arthroscopy through an 18-gauge spinal needle port [9,17]. e approach itself is minimally invasive and adds little additional risk to the patient. Figure  7 represents normal articular cartilage imaged with OCT, arthroscopy, and MR images [9]. e arthroscopic image (Figure 7(a)) of the tibial plateau displays smooth uniform articular cartilage despite being directly adjacent to a feathered femoral condyle. e arthroscopic image is veri�ed by MRI (Figure 7(b)), showing thick uniformly grey cartilage without signal dropout. e OCT image (Figure 7(c)) shows the endocatheter placed directly in the center (z plane) and shows three uniform tight banding patterns, indicative of highly organized collagen. Conversely, Figure  8 shows smooth uniform articular cartilage arthroscopically and shows a thick uniform band of grey cartilage without areas of dropout by MRI [9]. However, in the OCT image (Figure 8(c)) there is little evidence of a uniform banding pattern and a large area of dropout. is depicts the advantage of OCT's micron-scale resolution over other diagnostic imaging modalities; both the arthroscopic and MR images were unable to detect disorganization in the cartilage, as compared to the OCT scans, which showed early disruption.
In addition, this �gure shows that as the articular surface was being assessed, the meniscus could be visualized in the interim, also showing a tight uniform banding pattern indicative of highly organized tissue (birefringence).

OCT Imaging with Periarticular Structures.
OCT is also capable of assessing periarticular structures. One prominent example is rotator cuff issues, which affect approximately 30% of the US population, and which yields 4.5 million clinical visits and 40,000 corrective surgeries per year [3]. Acute insults result in pain with movement, reduced range of motion, and joint weakness [46][47][48][49]. Additionally, changes in the shoulder articular surface and bony spurs from the acromion can cause rotator cuff tendon (RCT) impingement. ese mechanisms cause collagen �bers to become abnormal through the loss of their organized, parallel structure, decreased striation, and irregular shape [10]. Impingement syndrome, in addition to traumatic mechanical insults to the shoulder, is also responsible for full-or partial-thickness tears of the rotator cuff tendons, commonly the supraspinatus, at the enthesis point with the greater humoral tuberosity [50][51][52]. Even with rotator cuff repairs, RCR, 25-60% of repaired shoulders rerupture within two years [53,54].
Currently, RCT injuries are repaired surgically by resection of the torn segment, �attening of the greater tuberosity, and securing of the healthy segment with anchored F 6: In vivo imaging of human knee cartilage. e le of the image is relatively normal cartilage, by the banding pattern, while the right is more signi�cantly diseased and has lost polarization sensitivity. Image courtesy Li et al. [8].
sutures [55]. is site becomes the new enthesis point of �brocartilage, depending heavily on local chondrocytes, �broblasts, and pluripotent mesenchymal cells [56][57][58]. Even though this anabolic process is integral to enthesis stability, the integrity of the repaired tendon cannot be underestimated in the formation of a strong enthesis. It is hypothesized that reattachment of a microstructurally damaged tendon can increase the probability of RCR failure [59]. Surgeons currently rely on experience and visual identi�cation of injured RCT for RCR surgeries. Knowing that the attachment of structurally unstable tendon with damaged collagen microstructure is detrimental to the success of RCR, surgical assistance to correctly identify and resect damaged tendon increases the hope of successful RCR by attachment of organized tendon to the enthesis. Similar to articular cartilage and meniscus, tendons and ligaments have a highly organized collagen structure and share similar properties of birefringence. Initial research in OCT imaging of periarticular structures was performed in postmortem and postsurgical resection samples [10]. Figures  9 and 10 represent periarticular tissue that shows normal structure by OCT and histopathology and has been used as a standard for normal tissue by OCT. It is important to visualize normal tissue as a control against which to assess diseased tissue, due to tissue preservation efforts common in surgical practice. Real-time high-resolution optical biopsies with OCT allow greater balance between tissue preservation and diseased tissue resection.
Biceps tendon is another extremely birefringent tissue, which serves as a prime example of periarticular OCT imaging ( Figure 9). OCT effectively demarcates the layer of the tendon, and the tight uniform banding pattern produced by the change in polarization state signi�es a dense concentration of organized collagen.e sensitivity of the imaging also illustrates the difference between the fascial layer surrounding the tendon, which shows no banding pattern. is is ver-i�ed by corresponding picrosirius red histopathology, which shows highly organized, thick bands of collagen [10]. Bovine meniscus, which nearly identically resembles human tissue, represents the completion of periarticular tissue ( Figure 10). e OCT images taken at two polarization states show a tight banding pattern without areas of dropout, again veri�ed with histopathology. Similarly, ligaments and other tendons have been shown to exhibit this same organized structure. Both healthy anterior cruciate ligaments and Achilles tendon display a uniform, heterogeneous banding pattern that is veri�able with histology [10,60,61]. Even though these varying tissues reside in different compartments within the body, they all have a similar composition and similar content of organized collagen, thus exhibiting similar birefringence and serving as controls against which to assess diseased tissue. e human rotator cuff de�nes a complex balance of muscle, tendon, and capsule to provide an outstanding range of motion. Once this unit is damaged, it presents as a difficult task to repair and regain an acceptable range of physiological mobility and strength. is difficulty is expressed in the staggering number of failed RCRs within two years of repair. e failure point is generally at the enthesis of the tendon and bone. Although partly dependent on the local chondrocytes, �broblasts, and mesenchymal cells, failure also depends on the quality of the organizational structure of the attached RCT [50,51,54]. A section of RCT taken from an RCR surgery that was deemed normal for reattachment by the surgeon can be seen in Figure 11. Comparing the OCT images with the previous �gures (Figures 9 and 10), it is evident that the tissue does not have the same dense uniform banding pattern, illustrating a weaker birefringence and organizational structure. However, this sample still shows some sensitivity to changes in polarization state, signifying only a mild disease, with the region that does not change with manipulations of polarization state representing a more advanced disease. Examining the picrosirius red histology (Figure 11(b)) a section of organized collagen dropout at the arrow is otherwise surrounded by healthy collagen-laden tendon, mirroring the OCT images. e trichrome stained histology shows areas of �brocartilage at the dropout area, further explaining the lack of organization by OCT [11]. is pilot data illustrates the feasibility for PS-OCT to be used as an interoperative imaging system to assess the borderline between tendon optimal for reattachment and disrupted tendon. F 11: is �gure demonstrates a very mildly diseased section of supraspinatus tendon. Even though this sample was considered to be "normal" by the surgeon, the picrosirius red-stained section shows a dropout of organized collagen (arrow in (b)) in otherwise healthy collagen. e trichrome section (c) shows the area is �brocartilage. In the OCT images, back-re�ection intensity changes dramatically between the two images, except the area with the arrow that essentially has no change (not birefringent) [11].
(a) (b) F 12: e �gure depicts preliminary data of a mouse tibiotalar joint without CIA induction (a) and one at 1 week aer RA induction (b). Both images have the capsule and so tissue intact (imaged from outside the joint) generated for this proposal. e skin was removed to simulate subcutaneous imaging as a potential approach. In (a), no distorted architecture is seen. e red arrow is the cartilage surface with the synovial space above (black, minimal signal). e black gap just below the cartilage is subchondral bone. e green circle is a large synovial space and the blue a tendon. e banding in the tendon is from the highly organized collagen. e orange arrow is synovium. In Figure 3(b), 1 wk aer CIA, the orange circle is synovial hypertrophy and pannus formation. Note the increased back-re�ection intensity relative to (a), consistent with cellular in�ltration. e red arrow is thickened cartilage (swollen in early arthritis). e subchondral bone (yellow arrow) is wider with more diffuse bone below it (edema). e green arrow is so tissue swelling outside the joint. (Unpublished results.)

OCT in Rheumatoid
Arthritis. Rheumatoid arthritis (RA) can be considered a more aggressive global in�ammatory arthritis due to the accelerated pathogenesis caused by genetic and immunological cofactors. RA affects 1% of the population, and even though some patients have mild symptoms of joint swelling/discomfort and live with little disability, more than 50% of patients with RA have aggressive disease presentation, causing great disability with substantial joint deformity and global panarticular damage aer only a few years [62][63][64][65][66]. Treatment for this disease was previously palliative, but with the onset and popularity of disease modifying antirheumatic drugs (DMARDs) which target TNF-alpha, T-cell costimulation, B-cells, or IL-6 pathways, there is a new push for aggressive early treatment to prevent articular and periarticular damage with these medications [65,66]. �owever, the �eld is plagued by the fact that there are no reliable markers for differentiating patient populations that need aggressive therapy at early stages from those that require less aggressive treatment for mild symptoms. Both adult and pediatric patients with RA are likely to bene�t from early diagnosis by a comprehensive understanding of RA pathogenesis/biomarkers. Early detection and enhanced understanding of disease states would assist practitioners with the ability to better choose therapies and limit the disease course. RA deserves investigation with OCT, since it has shown great potential in illustrating the early markers for OA. Preliminary data using OCT to visualize in�ammatory arthritis has been promising in animals. Preliminary images of a mouse tibiotalar joint one week aer induction of collagen-induced RA and a normal mouse tibiotalus joint have shown promise (Figure 12). Both images have the F 13: OCT imaging and corresponding histology of rat knee. Control knee (a)-(e) and diseased knee (f)-(j). Image courtesy of Adams et al. [12].
capsule and so tissue intact. e top image shows no distortion of the tissue architecture, with the articular surface and a thin synovial space indicated. e black gap just below the cartilage represents the subchondral bone, the green circle represents the synovial space, and blue arrows represent tendon, exhibiting a banding pattern signifying highly organized collagen. In contrast, the bottom image represents a mouse tibiotalus joint aer one week of collageninduced arthritis, showing synovial hypertrophy and pannus formation (orange circle) due to noted increase of back-re�ection intensity relative to the top image, consistent with cellular in�ltration. Cartilaginous thickening (red arrow), subchondral bone widening (yellow arrow), and more diffuse bone edema and so tissue swelling seen outside the joint (green arrow) all point to early signs of in�ammatory arthritis. is preliminary data shows the feasibility of OCT as a technology used to visualize in�ammatory arthritis in hopes of tracking the progress of future pharmaceutical management of the disease and to better the understanding of early signs�markers of in�ammatory arthritis.

OCT in Animal Models for
Research. Animal models are crucial for any research destined for human use. While many animal models exist for the investigation of OA, they show many limitations, including the need to sacri�ce the animals at certain time points for histology assessment. Since OCT has shown strong in vivo correlations with histology in birefringent tissue, it can be used to serially assess the same animal and follow sequential changes. OCT has recently been used in various animal models to assess normal and diseased tissue, focusing on OA, RCT, and RA. e ability to perform serial assessments of diseased tissue not only allows better understanding of the disease Supraspinatus tendon

Enthesis
Greater tuberosity F 14: e supraspinatus tendon, enthesis, and humerus are well de�ned. e banding in the tendon is due to polarization rotation caused by healthy collagen which would have been previously lost in diseased areas. For histological comparison, this sample was stained with both Masson's Trichrome, to expose any structural abnormalities, and picrosirius red, to determine collagen content using a polarization �lter. OCT studies in the musculoskeletal system of animals and in vivo in humans offer the potential of understanding mechanisms and identifying new therapeutic approaches, of which this image is an example. Image courtesy of Vercollone et al. [13].
state, but also establishes a method to track the efficacy of therapeutics. Following the same tissue sample at multiple time points with a higher resolution than other clinical imaging modalities allows the accuracy of OCT to have a more powerful use in research. OCT images of animals with and without induced OA in the femoral condyle illustrate this ability ( Figure 13) [12]. e most obvious difference between the images is the lack of cartilage on samples with induced OA. e respective histopathology con�rms the lack of cartilage in the diseased joint and the presence of organized collagen in the control knee. In four papers, our lab has taken serial images of the rat knee joint with mechanically induced OA and has been able to track the progression of the disease against therapeutic treatment. e progression of in�ammatory arthritis, RA, produces images different from OA, which is evident when comparing Figures 12 and 13. Much more evident are the global in�ammatory�edematous markers than loss of cartilage thickness and organization in focal areas. Even though tracking a disease state is integral to its understanding, an animal model to assess the potential surgical techniques is just as critical. In Figure 14, the resolution of OCT can be seen in coronal image of a rat rotator cuff with corresponding trichrome histopathology [62]. e supraspinatus tendon, greater humoral tuberosity, and enthesis are well de�ned, and the banding in the tendon is attributed to the organization of the tissue's collagen structure. is banding and birefringence would be lost or disrupted in a diseased or torn tendon. e histology shows an intact section and tendon-enthesis junction, once again verifying OCT �ndings. e addition of OCT to RCT, OA, and RA animal models will allow new therapeutic approaches, as a technique to visualize the progress and understand the underlying mechanisms of diseases.

Conclusion
OCT is a potentially powerful technology for assessing musculoskeletal disease at a micron scale resolutions and in real time. More clinical trials and the increased use of adjuvant approaches are needed for it to become a routine tool for clinical assessments.

A. Polarization-Sensitive OCT
PS-OCT measures the birefringence of samples by assessing the polarization state of back-re�ected light. Highly organized tissue, such as collagen, modi�es the polarization state of incident light in a homogeneous fashion, producing a uniform banding pattern on OCT imaging. is not only provides microstructural information from conventional OCT, but also allows the assessment of the sample's organization (in terms of degree of birefringence), which is indicative of disease progression [6,12,14,[40][41][42][43][44][45].

B. Doppler OCT
Doppler OCT combines traditional OCT with laser Doppler velocimetry. As the Doppler frequency shis do not produce a large effect on the conventional OCT image, they can be used concurrently. And, as the wavelength of light is shorter than that of sound, the resolution is far greater than that generated by traditional ultrasound Doppler imaging. Adding this high-resolution spatial imaging to the traditionally nonimaging technique of laser Doppler velocimetry allows for the assessment of clinical �ow measurements, such as blood �ow [14].

C. OCT Elastography
is technique measures the mechanical properties of tissue in a manner similar to ultrasound elastography. A mechanical force is applied to a tissue sample, and the resultant alteration in the tissue is recorded, producing information about the tissue's composition. Applications include phantoms and skin, with recent advances in the use of Air-Indentation OCT Elastography to assess mechanical properties of in vivo human skin in the extremities [67][68][69].

D. OCT Spectroscopy
In addition to OCT absorption spectroscopy, which functions by combining traditional OCT with spectroscopic analyses, recent work has shown promise in the development of a second-order correlation (SOC) measurement technique. is technique utilizes photon entanglement, a quantum phenomenon which gives insight into tissue composition [70,71].