Magnetic resonance elastography for musculoskeletal joint tissues

Abstract

Methods and devices for magnetic resonance imaging (MRI) and, more particularly, to a method and apparatus that is able to apply a mechanical load to the human knee in vivo within an MRI environment, and when combined with inverse constitutive modeling, enables the measurement of cartilage structure and mechanical function.

Description

[0001] MAGNETIC RESONANCE ELASTOGRAPHY FOR MUSCULOSKELETAL JOINT TISSUES

[0002] CROSS-REFERENCE TO RELATED APPLICATIONS

[0003] This application claims the benefit of U.S. Provisional Application No. 63 / 538,761, filed September 15, 2023.

[0004] STATEMENT OF GOVERNMENT INTEREST

[0005] This invention was made with government support under grant number AR063712 awarded by the National Institutes of Health. The government has certain rights in the invention.

[0006] FIELD OF INVENTION

[0007] This invention relates to methods and devices for magnetic resonance imaging (MRI) and, more particularly, to a method and apparatus that is able to apply a mechanical load to the human knee in vivo within an MRI environment, and when combined with inverse constitutive modeling, enables the measurement of cartilage structure and mechanical function.

[0008] BACKGROUND OF THE INVENTION

[0009] Osteoarthritis (OA) is one of the most common and costly orthopedic diseases that affects at least 32.5 million adults in the U.S. [Losina et al., 2015. Lifetime Medical Costs of Knee Osteoarthritis Management in the United States: Impact of Extending Indications for Total Knee Arthroplasty. Arthritis Care Res (Hoboken) 67, 203-215.; Safiri et al., 2020 Global, regional and national burden of osteoarthritis 1990-2017: a systematic analysis of the Global Burden of Disease Study 2017. Ann Rheum Dis 79, 819-828.]. OA is characterized by abnormal degeneration of the cartilage, loss and breakdown of major macromolecules, and joint pain and dysfunctionality [Buckwaiter et al., 2004. The Impact of Osteoarthritis: Implications for Research. Clin Orthop Relat Res 427, S6-S15], Structural changes to cartilage alter the biochemical and molecular composition, and consequently, the functional (biomechanical) response [Saarakkala et al., 2010. Depth-wise progression of osteoarthritis in human articular cartilage: investigation of composition, structure and biomechanics. Osteoarthritis Cartilage 18, 73-8], Although current methods to diagnose advanced OA have been established, the identification of OA in an early stage, where it can be prevented or decreased by intervention and correction of relevant factors, remains an unmet challenge [Hinton et al., 2002. Osteoarthritis: diagnosis and therapeutic considerations. Am Fam Physician 65, 841-848; Moskowitz et al., 2007. Osteoarthritis: diagnosis and medical / surgical management., 4th ed. Lippincott Williams & Wilkins; Sinusas, 2012. Osteoarthritis: diagnosis and treatment. Am Fam Physician 85, 49-56; Taruc-Uy and Lynch, 2013. Diagnosis and Treatment of Osteoarthritis. Primary Care: Clinics in Office Practice 40, 821-836; Heidari, 2011. Knee osteoarthritis diagnosis, treatment and associated factors of progression: part II. Caspian J Intern Med 2, 249-55],

[0010] Elastography is a non-invasive imaging approach that is commonly used to quantify the material properties of soft tissues of the body. Quantification of material stiffness or modulus can provide a powerful approach to understanding structure-function relationships in load-bearing tissues, specifically in cartilage. In osteoarthritis (OA) pathogenesis, cartilage softening is one functional hallmark of the disease due to structural changes to the collagen network and depletion of proteoglycans. Therefore, quantifying changes to cartilage material properties non-invasively in vivo is of substantial interest as a potential diagnostic tool.

[0011] SUMMARY OF THE INVENTION

[0012] Palpation is an ancient biomarker to assess illness for tissue scarring or cancer (hardening) or degeneration (softening). Noninvasive measurement of tissue stiffness or modulus using imagebased methods persists as a holy grail for disease diagnosis, but elastography methods applicable to stiff tissues like cartilage located deep within the human body remain technically challenging. We provide a novel, first-in-human elastography technique as a combination of displacement- encoded MRI and inverse modeling to reveal cartilage stiffness maps in knee joint cartilage susceptible to post-traumatic osteoarthritis. Following successful validation based upon forward models and Monte Carlo simulations, we applied the technique in a small cohort of participants at risk for the development of osteoarthritis due to prior ligament reconstruction. We show that cartilage stiffness heterogeneity is strongly correlated with altered loading patterns during gait, as well as patient-reported outcome scores of physical function and pain. This approach enables the study of in vivo cartilage biomechanics and identifies cartilage stiffness gradients as a possible biomarker for osteoarthritis pathogenesis, that could be used for patient assessment prior to development of radiographic osteoarthritis. Displacement-based elastography has the potential to quantify stiffness in joints and connective tissues susceptible to other degenerative arthropathies. Such imaging-based early diagnostics also have the potential to identify personalized therapies that optimize delivery and timing of emerging orthobiologics.

[0013] Our invention documents a first-in-human measurement of knee cartilage displacement and strain leading to stiffness, as well as shows the clinical relevance of stiffness measurements. We observed significant differences in femoral and tibial cartilage softening patterns post-injury, that shed new light on the early time-course changes of a healing knee. Our observations suggest that increased medial contact forces in the affected knee are strongly correlated with cartilage softening in the tibial cartilage and cartilage stiffening in the femoral cartilage, hinting at differential and location-specific patterns of post-traumatic tissue degeneration. Elastography allowed us to have a deeper understanding of the pathogenesis of post-traumatic osteoarthritis by tracking the degeneration of the cartilage and could assist in detection of such disease at an early stage.

[0014] Additionally, our invention overcomes decades-old challenges in imaging technology that have prevented stiffness measurements in joint tissues. Conventional MRI elastography was originally presented in 1995 in Science [Muthupillai, R. et al. Magnetic Resonance Elastography by Direct Visualization of Propagating Acoustic Strain Waves. Science (1979) 269, 1854-1857 (1995)] using high-frequency vibrational waves, but has advanced only for a few soft tissue diseases such as liver fibrosis. In contrast, propagation of waves in stiff musculoskeletal and connective tissues, deeply embedded between small bony spaces in the body, has completely precluded measurement in joint cartilage susceptible to osteoarthritis. Here, we overcome these challenges using displacement and strain mapping via displacement-encoded MRI, and relative stiffness mapping through novel displacement-based elastography algorithms. We apply mechanical loading using an apparatus at low-frequency, typically between 0.001 cycles / sec to 100 cycles / sec, and at high- force magnitude, typically between 0.1 to 10 times body weight (i.e., the weight of the individual person’s body), which mimics numerous normal physiological activities (e.g., walking, running, etc.). We also combine and co-register our elastography approach with widely-used quantitative MRI techniques (T2, T2* and Tip mapping) to directly compare the relative diagnostic power of state-of-the-art and modern imaging approaches.

[0015] The diagnosis of early-stage osteoarthritis remains as an unmet challenge in medicine and a roadblock to evaluating the efficacy of disease-modifying treatments. Because cartilage softens in early disease states, elastography (i.e., spatial maps of mechanical quantities like displacements, strain, or stiffness) provide an ideal approach to assessing early disease pathogenesis. Also, cartilage and musculoskeletal tissue softening may serve as a surrogate for overall joint degeneration and decreased health during OA progression. However, current conventional elastography methods, for example using ultrasound, lack the ability to penetrate deep within the musculoskeletal tissues of the joint space, due to the dissipation of high frequency shear waves. MRI is unique among imaging tools as it can characterize tissues with high spatial detail, has a deep penetration depth, and can assess tissues non-invasively. Unfortunately, MRI elastography also uses mechanical waves that are not able to capture cartilage mechanics in vivo, or the mechanics of other stiff musculoskeletal and stiff connective tissues, and other approaches are required. Displacement-encoded stimulated echo MRI provides pixel-level and full-field displacement maps of soft tissues under repetitive motion, and when combined with inverse modeling, may form the foundation of elastography methods suitable to image musculoskeletal tissues. The present invention provides an elastography approach that utilizes displacement maps as input to a biomechanical model-based elasticity imaging pipeline, providing capability to quantify cartilage stiffness maps non-invasively.

[0016] Recent studies demonstrate that unique patterns of intratissue cartilage deformation under cyclic loading can serve as potential biomarkers to detect early disease pathogenesis. However, a workflow to obtain deformation, strain, stiffness, and quantitative MRI metrics due to the loading of articular cartilage in vivo has not been fully developed. We further characterize and demonstrate an apparatus that is able to apply low-frequency, physiologically-relevant mechanical loads and forces to the human knee in vivo within an MRI environment to enable the measurement of cartilage structure and mechanical function. The apparatus was first tested in a lab environment. Then the functionality and utility of the apparatus were examined in a clinical 3T MRI system for human imaging. The apparatus developed according to the present invention enables quantitative MRI metrics for biomechanics and relaxometry data acquisition during mechanical loading leading to, for example, varus-valgus motion and compression of the medial or lateral knee compartment. Integration and timing of in vivo, physiologically-relevant mechanical loading with MRI displacement-encoded pulse sequences, for example spiral DENSE, during cyclic loading provided time-dependent displacement and strain maps within the tibiofemoral cartilage. Other MRI data acquisition is possible beyond spiral imaging, including but not limited to single-shot fast spin echo, gradient echo, and steady-state free precession imaging. The results from these procedures demonstrate that the performance of this loading apparatus meets the design criteria and enables a simple and practical workflow for studies of clinical cohorts and the identification of imaging-based biomechanical biomarkers.

[0017] ADD A PARAPHRASE OF THE CLAIMS

[0018] BRIEF DESCRIPTION OF THE DRAWINGS

[0019] For a fuller understanding of the invention, reference should be made to the following detailed description, taken in connection with the accompanying drawings, in which:

[0020] FIG. l is a set of images showing the validation of stiffness reconstructions in complex materials. Stiffness calculations results of a gradient with ideal displacements. For this figure, a 3T Siemens (Munich, Germany) MRI was used to collect DENSE (displacement encoding with stimulated echoes) MR images with a custom pneumatic loading device that applies a varus-valgus load at one half body weight to the cartilage. Imaging is not restricted to 3T or any particular vendor, and the acquisition of images can be attained over any field strength (e.g., 0.5T-14.7T) and with any commercial or custom MRI system. Multi-frame DENSE MR images including magnitude and phase are acquired during the loading time frame. Displacements at the final time step are then calculated from phase information and smoothed with a LOWESS filter.

[0021] FIG. 2 is a set of schematics and a graph showing the validation of stiffness reconstructions in complex materials. Stiffness calculation results of a gradient model with different levels of noise added. A 3T Siemens (Munich, Germany) MRI was used to collect DENSE MR images with a custom pneumatic loading device that applies a varus-valgus load at one half body weight to the cartilage. Multi-frame DENSE MR images including magnitude and phase are acquired during the loading time frame. Displacements at the final time step are then calculated from phase information and smoothed with a LOWESS filter.

[0022] FIG. 3 is a set of images, schematics, and graphs showing the validation of stiffness reconstructions in complex materials. Stiffness calculations are robust to soft inclusion representative of tissue defect. For this figure, a 3T MRI was used to collect DENSE MR images with a custom pneumatic loading device that applies a varus-valgus load at one half body weight to the cartilage. Multi -frame DENSE MR images including magnitude and phase are acquired during the loading time frame. Displacements at the final time step are then calculated from phase information and smoothed with a LOWESS filter.

[0023] FIG. 4 is a set of images, schematics, and graphs showing the validation of stiffness reconstructions in complex materials. Spiral DENSE MRI demonstrates in vivo calculation of cartilage displacements in a healthy participant while under a cyclic varus-valgus load. A 3T MRI was used to collect DENSE MR images with a custom pneumatic loading device that applies a varus-valgus load at one half body weight to the cartilage. Multi-frame DENSE MR images including magnitude and phase are acquired during the loading time frame. Displacements at the final time step are then calculated from phase information and smoothed with a LOWESS filter.

[0024] FIG. 5 is a pair of images showing the overall structure and control of the MRI-compatible apparatus, configured here for varus-valgus loading. (A) The loading apparatus is shown separately from the MRI system. (B) The loading apparatus is mounted on a 3T MRI scanner (1), and interfaces with a commercially-available (QED 15-channel) knee coil for volumetric imaging (3). The apparatus (10) consists of two parts: a loading section (10a) and a thigh constraint (10b). FIG. 6 is a pair of images showing the corresponding computer aided design renderings demonstrating (A) the patient positioning within the MRI scanner and (B) MRI-compatible apparatus, configured here for varus-valgus loading, including the loading section and a thigh constraint.

[0025] FIG. 7 is a schematic showing the overall structure and control of the MRI-compatible loading apparatus, configured here for varus-valgus loading. Control of mechanical actuation within the MRI scanner involves multiple components: Lab VIEW interface and control code on a computer (5) to integrate timing with compressed air from a gas tank (6), via an air regulator (7), to actuate the pneumatic cylinder at the ankle and foot of the patient.

[0026] FIG. 8 is a set of three images showing the detailed structure of the loading section of the MRI- compatible apparatus, configured here for varus-valgus loading. The loading section consists of the following parts: main frame (1), linear bearing system (2, 5, 6), rear support (3), foot cast assembly (4), pneumatic cylinder assembly (7, 8). B) Layout of the device in the MRI scanner. The entire apparatus is aligned using the existing rail system on the scanner. This configuration shown is for the left knee and can achieve both varus and valgus loading to either knee.

[0027] FIG. 9 is a set of four images showing the MRI-compatible apparatus, configured here for varus- valgus loading and affixed to an MRI scanner. The entire apparatus is aligned using the existing rail system on the scanner. This configuration shown is adapted for the left knee and can achieve both varus and valgus loading to either knee, depending upon setup.

[0028] FIG. 10 is a set of two graphs showing the force calibration of an MRI-compatible apparatus, configured here for varus-valgus loading. Calibration of input voltage (x-axis), and output forces (y-axis) provide a linear relationship (valgus: Pearson’s R=0.9999, p<0.001; varus: Pearson’s R=0.9998 p<0.001). The linearized relationship is as follows: varus: F=97.706V-4.319; valgus: F=112.051V-6.729. Here, V is the input voltage (Volt), and F is the output force (Newton).

[0029] FIG. 11 is an image and a schematics showing in-lab verification and validation process and results. (A) Layout of the experimental setup in the lab environment, with KEYENCE laser system (lower right arrow) attached to the main frame. (B) Scheme of varus loading: here, LI (the distance between the location of the load and the intercondylar eminence of the tibia, measured with tape measure), L2 (the distance between the intercondylar eminence of the tibia and the center of the contact region of the tibiofemoral cartilage, medial side for varus loading, lateral side for valgus loading, measured from DESS image), and reaction force are used to calculate load force using balance of moment. FIG. 12 is a set of graphs showing in-lab verification and validation of the process and results. (A) Repeatability was assessed using 3 consecutive loading durations that are 20 min each with 10 min rest time in between. The loading duration represents the actual loading time expected within MRI scanner (Lee et al., 2023), while the resting time is used for patient recovery. (B) Estimates for standard deviation used the ascending portion of cyclic loading. The mean and PSD of the standard deviation (C) throughout the ascending section of each patient is shown in the graphs in FIG. 13.

[0030] FIG. 13 is a set of graphs showing in-lab verification and validation of the process and results. To assess time-dependent effects, we compare the standard deviation of the entire 20 minutes versus the last 12 minutes of cyclic loading. We also calculated the estimated standard deviation of displacement at the location of the knee cartilage using LI and L2.

[0031] FIG. 14 is a set of graphs and images showing the relaxometry measurements - Tl, T2, T2*, and Tip - before and during mechanical loading. (A) The timing and integration of MRI and varus loading enables qMRI acquisition before and biomechanics-related changes in structure and biochemistry. (B) The spatial maps of relaxometric measures were acquired before and after a 20- minute 25N constant varus load (corresponding to 424N compressive load at the medial condyle), then the corresponding relaxometry maps within the manually selected cartilage ROI were calculated and overlay ed on the weighed image. The Tl, T2, T2*, and Tip value before / after the loading were 1026ms / 1028ms, 31.16ms / 34.04ms, 21.15ms / 26.13, and 37.26ms / 35.29ms.

[0032] FIG. 15 is a set of images and graphs showing the integration of mechanical loading with DENSE MRI image acquisition enables measurement of spatiotemporal maps of displacement results. (A) Scheme of spiral DENSE MR sequence (Lee et al., 2023) to measure spatial patterns of mechanical metrics, here shown for the medial knee compartment during varus loading. (B) A temporal sequence of images were acquired during varus loading, with highlighted images shown in green at frame 1 (t=Os), frame 10 (t=0.37s), frame 19 (t=0.7s), and frame 27 (t=ls). C) Spatiotemporal displacements in x and y direction at 4 time steps indicates that the medial tibiofemoral cartilage moved ~2.5mm in -x direction and ~0.8mm in +y direction from t=Os to t=ls.

[0033] FIG. 16 is a set of images showing the integration of mechanical loading with DENSE MRI image acquisition enables measurement of spatiotemporal maps of displacement results. Spatiotemporal displacements in x andy direction at 4 time steps indicates that the medial tibiofemoral cartilage moved ~2.5mm in -x direction and ~0.8mm in +y direction from t=Os to t=ls. FIG. 17 is a set of images showing spatiotemporal maps of Green-Lagrange strain during varus loading. Here we show the normal strain in x (Exx), y (Eyy), and shear strain (Exy) directions. From t=Os to t=ls, the medial side of medial tibiofemoral cartilage experiences compression gradually while the lateral side experiences expansion in y direction. On the contrary, in x direction, the medial tibiofemoral cartilage experiences lateral expansion gradually during varus. The shear strain is primarily compressive during varus loading, while the absolute strain values demonstrate that the cartilage is mainly under compression in they direction.

[0034] FIG. 18 is a set of schematics of an experimental overview and calculation of relative stiffness or modulus. (A) and (B) Experimental Overview. The experimental design involved acquiring qMRI relaxometry values (T2, T2*, Tip) during rest, along with spiral displacement encoding with stimulated echoes (DENSE) MRI images during varus loading. Initially, scout and 3D double echo steady state (DESS) sequences were conducted to visualize knee anatomy. A specific slice was chosen to maximize tibiofemoral contact in the medial condyle. After DESS, relaxometry measurements (T2, T2*, Tip) were obtained. A subset of participants underwent gait collection. An MRI-compatible varus loading device, comprising a thigh constraint, pneumatic cylinder, and boot section, applied cyclic varus loading (1 s load followed by 1 s unload). Preconditioning lasting 8 minutes before DENSE acquisition minimized cartilage viscoelasticity and motion artifacts. Post-DENSE acquisition, phase images in x and y were converted to displacement fields in selected cartilage regions of interest, smoothed, and used to calculate Green-Lagrange strains. All participants were supine inside the MRI bore for all sequences. (C) Calculation of relative modulus from displacements. Briefly, displacements were mapped onto a finite element mesh. The virtual fields method was employed to determine an initial moduli guess, followed by an interactive optimization routine to derive relative moduli maps. (D) The modulus gradient map and gradient index were calculated using a finite difference method to calculate modulus gradients in a localized manner perpendicular to the cartilage interface. Cumulative histogram thresholding was utilized to derive a gradient index value, representing the entire region of interest. DESS, Double echo steady state. DENSE, Displacement encoding with stimulated echoes. qMRI, quantitative magnetic resonance imaging.

[0035] FIG. 19 is a set of schematics and graphs. Synthetic models were used to validate the elastography pipeline. A. Five distinct finite element models were simulated using FEBio to assess the feasibility of the elasticity imaging framework in conjunction with the DENSE MRI sequence for imaging cartilage moduli These simulations varied in cartilage stiffness heterogeneity and geometries, increasing in complexity. All models assumed linear elastic cartilage material properties with depth-dependent moduli and a nearly incompressible Poisson ratio. The models, ordered from least to most complex, included: (B) a two-dimensional square with increasing moduli values under unconfined compression; (C) the same model with an additional soft inclusion at the top to simulate cartilage bulk softening; (D) the model with a narrow soft inclusion to mimic a cartilage fissure; (E) a two-dimensional thin cartilage-like half annulus under unconfined compression; and (F) a three-dimensional finite element knee model with three layers of depth-dependent articular cartilage moduli, subjected to medial compartmental compression. All average percentage errors are shown in Table 6 for B-F. (G) Monte Carlo simulations were used to assess the response of the elasticity imaging framework in response to noise.

[0036] FIG. 20 is a set of graphs. Modulus gradient metrics, that is, spatial patterns of cartilage modulus or stiffness, are significantly correlated with patient reported outcome severity. After controlling for variability introduced by covariate data correlations with KOOS (knee injury and osteoarthritis outcome score) overall score, KOOS quality of life subscore and KOOS pain score are shown. All unadjusted regressions are given in the supplement. Confidence intervals of 95% are shown in gray. (A) Relationship between KOOS overall score and modulus measurements. (B) Relationship between KOOS pain score and modulus measurements. (C) Relationship between KOOS quality of life score and modulus measurements.

[0037] FIG. 21 is a set of images and graphs. Modulus gradient metrics are significantly correlated with knee loading severity. After controlling for variability introduced by covariate data, correlations of modulus gradient measures with knee adduction moment and medial contact force are shown. All unadjusted regressions are given in the supplement. Confidence intervals of 95% are shown in gray. (A) Representative images of relative modulus and modulus gradient. (B) Relationship between normalized knee adduction moment and modulus measurements. (C) Relationship between medial contact force and modulus measurements.

[0038] FIG. 22 is a drawing providing perspective view of the loading apparatus and thigh constraint.

[0039] FIG. 23 is a drawing providing perspective view of the loading apparatus of the MRI-compatible, varus-valgus loading apparatus. The loading section consists of the following parts: main frame (1), linear bearing system (2, 5, 6), rear support (3), foot cast assembly (4), pneumatic cylinder assembly (7, 8). The entire apparatus is aligned using the existing rail system on the scanner.

[0040] FIG. 24 is an alternative perspective view of the MRI-compatible apparatus, configured for varus- valgus loading shown in FIG. 23. FIG. 25 is a perspective view of the loading apparatus mounted on a 3T MRI scanner (1), and interfaces with a QED 15-channel knee coil (3).

[0041] FIG. 26 is a schematic showing the overall structure and control of the MRI-compatible apparatus, configured for varus-valgus loading. Control of mechanical actuation within the MRI scanner involves multiple components: Lab VIEW interface and control code on a computer (5) to integrate timing with compressed air from a gas tank (6), via an air regulator (7), to actuate the pneumatic cylinder at the ankle and foot of the patient.

[0042] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

[0043] Elastography is a non-invasive imaging approach that is commonly used to quantify the material properties (e.g., stiffness, modulus) of soft tissues of the body. Quantification of material stiffness can provide a powerful approach to understanding structure-function relationships in load-bearing tissues, specifically in cartilage. In osteoarthritis (OA) pathogenesis, cartilage softening is a functional hallmark due to structural changes to the collagen network and depletion of proteoglycans. Therefore, quantifying changes to cartilage material properties non-invasively in vivo is of substantial interest as a potential diagnostic tool.

[0044] While elastography seems like an ideal approach to accomplish this, current conventional elastography methods, most commonly using ultrasound or MRI, lack the ability to penetrate deep within the musculoskeletal tissues of the joint space, due to the dissipation of high frequency shear waves. MRI is unique among imaging tools as it can characterize tissues with high spatial detail and contrast in image intensity, has a deep penetration depth, and can assess tissues non- invasively. Recently, spiral displacement encoding with stimulated echoes (spiral DENSE) MRI has been used to calculate pixel-level full-field displacement maps of soft tissues under repetitive motion [Lee, W., et al., 2023. High frame rate deformation analysis of knee cartilage by spiral dualMRI and relaxation mapping. Magn Reson Med 89, 694-709. https: / / doi.org / 10.1002 / mrm.29487], In this work, we first developed synthetic displacement maps of MRI phantoms and heterogenous cartilage under load. We then use these synthetic displacement maps as input to validate a biomechanical model-based imaging pipeline [Miller CE, Jordan JH, Thomas A, Weis JA. Developing a biomechanical model-based elasticity imaging method for assessing hormone receptor positive breast cancer treatment-related myocardial stiffness changes. J Med Imaging (Bellingham). 2021 Sep;8(5):056002. doi:

[0045] 10.1117 / 1. JMI.8.5.056002. Epub 2021 Sep 30. PMID: 34604442.], that has the capability to quantify cartilage stiffness maps non-invasively. In brief, DENSE MRI is used to calculate in vivo cartilage displacement and strain data, which are then used as input to the biomechanical elasticity imaging pipeline. For example, an initial elastic input may be calculated from input displacements using the virtual fields method [Mei Y et. Al., Comput Meeh. 2021], An objective-function defined by the sum of squared error mismatch between the observed and model-estimated displacement fields is then minimized via the adjoint method. An L-BFGS quasi-Newton algorithm is used to iteratively update the modulus estimates for each element until the overall error is minimized [Miller CE et al., J Med Imaging. 2021],

[0046] To validate the pipeline, we constructed forward simulations representative of tissue heterogeneity. Simulations of phantoms of increasing stiffness (500 to 1000 Pa) with and without a soft inclusion (200 mPA) under 10 percent unconfined compression (FIG. 19) were run in FEBio. Displacements for each node were used as input to the elasticity pipeline. To analyze error, we applied Monte Carlo simulations. Within each iteration of the 100 Monte Carlo simulations, random Gaussian noise was added to the ideal displacements of all nodes (FIG. 19). To analyze the effect of smoothing on stiffness reconstruction error, LOWESS smoothing was separately applied to the noisy displacements during Monte Carlo simulations, and percent error from the known stiffness was calculated. In preparation for in vivo work, spiral DENSE MRI was used to quantify in vivo cartilage displacements under a varus-valgus load, as previously described [Lee W et al., Magn Reson Med. 2023],

[0047] The elasticity pipeline successfully reconstructed the gradient pattern of stiffness defined in the forward simulation (FIG. 1A). Percent error of stiffness for each element remained under 2 percent. With the addition of LOWESS smoothing to the displacements prior to the calculation of stiffness, the stiffness calculation was robust to increasing noise levels present due to MRI acquisition (FIG. 2). When six rounds of LOWESS smoothing were applied to the displacements, the average percent error was minimized to less than 5 percent for all noise cases. The elasticity pipeline also successfully reconstructed the stiffness of a phantom with a soft inclusion, although error was higher in the inclusion area that included discontinuous material properties, and that reached up to 15 percent (FIG. 3).

[0048] This work provides a proof-of-concept and validation of an elasticity imaging framework. When combined with dualMRI to noninvasively extract full field displacements (FIG. 4), this method can quantify the cartilage stiffness map in vivo. Based upon ex vivo cartilage mechanical studies, changes in stiffness to the medial cartilage middle and deep zones are potential biomarkers for osteoarthritis pathogenesis. Ultimately, this framework provides techniques and systems to measure cartilage material properties in a clinical MRI setting and could be extended to other load- bearing musculoskeletal tissues, including ligament and intervertebral disc. The same framework can be applied more broadly to musculoskeletal and connective tissues that include cartilage, bone, meniscus, ligament, tendon, bone marrow, bursa, synovium, adipose, skin, muscle, and intervertebral disc. Additionally, the framework can be applied to these tissues that are healthy, degenerative, injured, diseased, repaired, or regenerated, and also tissues that are young, juvenile, or aging or old. The measurements computed for relative or absolute stiffness or material properties include elastic properties, tensile or compressive stiffness or modulus, shear stiffness or modulus, Young’s modulus, equilibrium modulus, viscoelastic constants (e.g., complex, storage, or loss modulus), multiphasic constants (e.g., fluid load fraction). One can also compute spatial distributions or spatial gradients of the relative or absolute stiffness or material properties. The framework is useful for the prediction of disease state or degeneration, and also for assessment of engineered materials, or materials or tissues that have been repaired or regenerated.

[0049] Advances in magnetic resonance imaging (MRI) technology has equipped us with unique non- invasive imaging methods to probe the altered structure of the cartilage in vivo. For example, quantitative MRI (qMRI) is commonly used to link physiological metrics to MRI-related physical parameters [Matzat, S.J., et al., 2013. Quantitative MRI techniques of cartilage composition. Quant Imaging Med Surg 3, 162-74. https: / / doi.Org / 10.3978 / j.issn.2223-4292.2013.06.04; Nieminen, M.T., et al., 2001. T 2 relaxation reveals spatial collagen architecture in articular cartilage: A comparative quantitative MRI and polarized light microscopic study. Magn Reson Med 46, 487-493. https: / / doi.org / 10.1002 / mrm.1218.], while displacement under applied loading MRI (dualMRI) is emerging as a means to determine the intratissue functionality and disease- related biomechanical changes of cartilage [Chan et al., 2016; Griebel et al., 2013], Both technologies could serve as useful and potentially complementary tools to study early-stage OA and aid in OA diagnostics and potential treatments.

[0050] More recently, the combination of mechanical loading with the timing of MR images has led to the development of “stress tests”, especially for understanding knee ligament and cartilage injuries [Paranjape, C.S., et al., 2019. A New Stress Test for Knee loint Cartilage. Sci Rep 9, 2283. https: / / doi.org / 10.1038 / s41598-018-38104-2; Schad, P., et al., 2020. Magnetic resonance imaging of human knee joint functionality under variable compressive in-situ loading and axis alignment. I Meeh Behav Biomed Mater 110, 103890. https: / / doi.org / 10.1016 / jjmbbm.2020.103890]. For such tests, there are existing devices that can apply axial compressive loading to the knee in the MRI scanner [Chan, D.D., et al., 2016. In vivo articular cartilage deformation: noninvasive quantification of intratissue strain during joint contact in the human knee. Sci Rep 6, 19220. https: / / doi.org / 10.1038 / srepl9220; Wang, H., et al., 2015. An MRI-compatible loading device to assess knee joint cartilage deformation: Effect of preloading and inter-test repeatability. JBiomech 48, 2934-2940. https: / / doi.org / 10.1016 / jjbiomech.2015.08.006], as well as the devices that adopt the idea of the varus-valgus stress test [Ciba, M., et al., 2022. Varus stress MRI in the refined assessment of the posterolateral corner of the knee joint. Sci Rep 12, 11858. https: / / doi.org / 10.1038 / s41598-022-15787-2; Said, O., et al., An MRI-compatible varus-valgus loading device for whole-knee joint functionality assessment based on compartmental compression: a proof-of-concept study. Magnetic Resonance Materials in Physics, Biology and Medicine 33, 839-854. https: / / doi.org / 10.1007 / sl0334-020-00844-6]. The varus-valgus stress test can be used clinically to assess knee integrity and laxity by inducing pressurization to the medial or lateral tibiofemoral cartilage while engaging the ligaments. This method provides a unique way to assess compartmental OA, and is easier to implement [Said et al., 2020] compared to axial compression alone [Chan et al., 2016], However, stress tests have not yet been developed to quantify intratissue biomechanics, despite evidence that tissue-specific and internal patterns of displacement and strain may serve as early biomarkers for disease progression [Chan et al., 2016; Griebel, A.J., et al., 2013. Noninvasive dualMRI-based strains vary by depth and region in human osteoarthritic articular cartilage. Osteoarthritis Cartilage 21, 394-400. https: / / doi.Org / 10.1016 / j.joca.2012. l l.009; Lee et al., 2023], Existing varus-valgus loading devices are either not designed for in vivo human knee [Siston, R.A., Maack, T.L., Hutter, E.E., Beal, M.D., Chaudhari, A.M.W., 2012. Design and Cadaveric Validation of a Novel Device to Quantify Knee Stability During Total Knee Arthroplasty. J Biomech Eng 134. https: / / doi.org / 10.1115 / L4007822], or not designed to easily accommodate a standard knee coil [Said, O., et al., 2020. An MRI-compatible varus-valgus loading device for whole-knee joint functionality assessment based on compartmental compression: a proof-of-concept study. Magnetic Resonance Materials in Physics, Biology and Medicine 33, 839-854. https: / / doi.org / 10.1007 / sl0334-020-00844-6].

[0051] Therefore, this invention exploits the concept of the stress test in an MRI environment along with qMRI methods to improve the functional assessment of cartilage and the knee joint for early-stage OA. In combination with dualMRI, we develop an MRI-compatible loading device to investigate the intratissue displacement and strain during varus-valgus loading. The disclosure for the present invention teaches the design, verification, and validation process of a loading apparatus that can be used in a clinical 3 Tesla (T) MRI scanner to enable unique biomechanical and load-dependent qMRI measurements. With this device, patient cohorts can be examined for the quantitative analysis of joint health and diagnosis of early-stage OA. EXAMPLE 1 - MATERIALS AND METHODS

[0052] The present disclosure provides the (1) general approach to calculate mechanical quantities like displacement, strain, and stiffness in cartilage using a loading device, MRI scanner, and inverse constitutive modeling, (2) general design and layout of the loading apparatus and its control unit, (3) in-lab calibration, verification, and validation of the loading apparatus and elastography method workflow, and (4) proof-of-concept application of the elastography method and loading apparatus in an MRI environment for human imaging of cartilage structure and function.

[0053] Design of the Loading Apparatus

[0054] The mechanism of the loading apparatus, configured for varus-valgus loading, contrasts with the clinical varus-valgus stress test (Said et al., 2020)(FIG. 5A). The loading apparatus consists of a thigh constraint section, which fixes the thigh to eliminate gross translational motion, and a loading section, which exerted a rotational (varus-valgus) torque on the participant’s tibiofemoral joint, leading to compressive loading to the medial (varus) or lateral (valgus) knee compartment and tibiofemoral cartilage. The loading apparatus (FIGS. 5, 6) interfaces with / affixes to an MRI scanner (e.g., 3T imaging system) and a knee coil (e.g., Tx / Rx Knee 15 channel coil) (See e.g., FIG. 5A) for combined and integrated actions and load-dependent image acquisitions.

[0055] The thigh constraint section (10b) includes a base (11) and two pillar blocks (12) that were wrapped in 1cm thick silicon rubber (Table 1) for the comfort of the patient (FIG. 22). The pillar block locations are laterally adjustable to compensate for individual anatomical differences of patients, as well as left or right leg choices. In the MRI scanner, the thigh constraint was secured by the MRI bore, the knee coil, and the patient’s own weight.

[0056] The loading section (FIGS. 22 and 23) consists of materials that are required to be both MRI- compatible (i.e., non-ferrous and non-metallic) and structurally robust (See Table 2 for parts). The mainframe of the loading section (22) attaches the patient table of the MRI machine, which supports the loading section. The force application is executed by a pneumatic cylinder (24) that performs linear reciprocating motion controlled by a customized control unit (FIG. 26). The cylinder (24), along with a customized connector, is positioned and affixed to the loading section to achieve mechanical loading of and interchangeability between the patient’s legs. The piston of the cylinder can be directly connected to the foot cast assembly (26). The foot cast assembly consists of a geometrically modified foot cast and a frame, allowing it to slide on the linear motion system. It could be reconfigured (26a) for 5 different flex angles of the knee (hyperextension, 5°, 10°, 15°, 20°). All these configuration changes can be made within minutes to save scan time. The piston rod length can be controlled and fixed for extension-contraction motion to restrict excessive piston displacement.

[0057] Design of the Control Unit

[0058] A control unit is also utilized in the varus-valgus loading apparatus (FIG. 7). The control unit consists of an air pump (52), an air regulator (54), a power source, a data acquisition (DAQ) system (56), and a customized control circuit (58) (See Table 3). Due to safety and performance concerns, these components can be located outside of the scanner room, sheltered from the strong magnetic field. A control algorithm (Lab VIEW 2022, National Instruments Corp., Austin, TX) ran on a computer (58) and communicated to the DAQ system (56), which transmitted the control signal to the air regulator. The air regulator (54) outputs compressed air via hoses with a desired pressure from the cylinder (52) to actuate piston (24) movement. In one embodiment, the piston applies movement to the foot, bends the lower leg and leads to compressive and shear loading in the knee joint. Compressive and shear loading in the knee is detected during synchronous actions of the MRI scanner, which utilize phase-contrast imaging to visualize displacement during low- frequency (e.g., 0.5-2.0 Hz) actuation at up to 2 times body weight. Displacements are used to calculate large (e.g. Green Lagrange) strain within the cartilage, and / or used to calculate stiffness within cartilage when combined with inverse modeling.

[0059] For inverse modeling, we provide minimization algorithms, virtual fields, or topology optimization in combination with constitutive linear (e.g., Hookean), nonlinear (e.g., Neo- Hookean or Mooney -Rivlin), or multiphasic or poroelastic models to extract mechanical quantities like stiffness or modulus. Finite element analysis framework can be utilized for calculations. In the finite element analysis framework, specification of displacement boundary conditions, often output from the displacement-encoded MRI data, enable the calculation of relative measures of material properties. If force boundary conditions are known or can be estimated, then absolute measures of material properties is possible. Importantly, to account for variation in joint geometry that might otherwise influence local forces (stresses) and strains among joints, we implement inverse modeling to reveal spatial patterns of material properties musculoskeletal tissues like cartilage. We are additionally able to estimate stress fields in cartilage using estimates of material properties. For example, for spatial mapping of material properties, the relative stiffness of each finite element may be varied to minimize the absolute difference of the displacements atN discrete points between the simulations and MRI displacement measurements using an iterative solver in a finite element analysis software.

[0060] Voltage-Force Calibration of the Pneumatic Cylinder To ensure consistent force actuation in varus or valgus, calibration of the control unit and the pneumatic cylinder was performed. The input was entered in the Lab VIEW code in Volts, while the output force was read by a factory-calibrated loadcell (LCM302-200N, Omega Engineering Inc., Norwalk, CT) in Newtons. Here, the input ranged from OV to IV with a 0. IV interval. With each input, we read three force outputs were read and their mean and standard deviation were calculated.

[0061] In-Lab Validation and Verification

[0062] The performance of the apparatus was validated and verified in the lab by tracking the displacement response of the knee and lower shank during varus loading. Three individuals (two male, one female, average age=28.7 years, range=25-35, healthy left knee) were recruited to perform repeatability and precision testing. The loading apparatus and a mock knee coil were first configured identically to mimic their arrangement within the MRI scanner (FIG. 9) and secured to the ground. Then, the participant was asked to lie down in supine position with their left leg in the apparatus (FIG. 11 A) and requested to stay still throughout the test. With each participant, three 20-minute individual tests with two 10-minute intervals in between were performed (FIG. 12). A 0.5Hz (Is loading session, 25N, followed by Is unloading session, 30N) cyclic varus load was applied to the ankle to mimic a walking cadence. Before each test, the sensor reading was set to 0. Displacement vs. time data throughout the tests was collected using a laser-based displacement sensor system (LJ-G5000, Keyence Corp., Itasca, II) at 60Hz. The sensor readout ascended for the varus loading section and descended for the unloading section.

[0063] MRI Validation and Verification for Relaxometry Measurements

[0064] The performance of the apparatus was additionally verified and validated within an MRI session. Here, one participant (female, age 25, healthy left knee) was recruited, and biomechanical and relaxometry data was acquired. We obtained proton density -weighted double echo steady state (DESS) images, along with T1-, T2-, T2*-, and Tip-weighted images before and after a constant 25N varus load that was applied for 20 minutes (FIG. 5 A).

[0065] The parameters for each measurement are listed in Table 4. Other than listed parameters, the Tip [Witschey, W.R.T., et al., 2008. Tip-prepared balanced gradient echo for rapid 3D Tip MRI. Journal of Magnetic Resonance Imaging 28, 744-754. https: / / doi.org / 10.1002 / jmri.21444] sequence has spin-lock frequency=500Hz along with spin-lock durations (TSL)=0ms, 50ms, 100ms, 150ms, and 200ms. The Tl, T2, and T2* maps were directly output from the scanner, while the Tip mapping was done by the monoexponential fitting of the pixel intensity values from the MR images at each TSL using customized software (MATLAB R2020a, MathWorks®, Natick, MA). After obtaining the Tl, T2, T2*, and Tip map, the medial side of the tibiofemoral cartilage was selected as a region of interest (ROI).

[0066] MRI-Based Measurement of Intratissue Displacement and Strain

[0067] The spiral Displacement ENcoded Stimulated Echoes (DENSE) MRI sequence (FIG. 15 A) was used to acquire knee images under a cyclic varus loading [Lee et al., 2023; Zhong, X., et al., 2010. Imaging three-dimensional myocardial mechanics using navigator-gated volumetric spiral cine DENSE MRI. Magn Reson Med 64, 1089-1097. https: / / doi.org / 10.1002 / mrm.22503]. Due to the nature of the DENSE sequence, we synchronized the cyclic load with the MRI data acquisition using an MRI-controlled trigger that fed a defined (synthetic) electrocardiographic signal to the scanner. The spiral DENSE MR images were collected only during the varus loading motion. Here, the 1-second loading session was divided evenly into 27 frames (FIG. 15B). Before the DENSE session, the geometry of the tibia (LI and L2)(FIG. 1 IB) was measured. With the measurements, the balance of moment was then applied to calculate the net load force, aiming to load half body weight on medial condyle. This calculated value plus the measured friction force induced by the linear bearing system was the total needed output force from the apparatus. The corresponding retraction force was 5N to 10N more than the output force to ensure the boot assembly could reach the hard stop each cycle. The loading scheme was identical to the in-lab validation and verification session. The parameters for spiral DENSE were as follows: temporal resolution=40ms, interleaves=10, TR / TE=20ms / 2.5ms, spatial resolution=360><360pm2, slice thickness= 1.7mm, displacement encoding gradient (ke)=0.64cycles / mm.

[0068] In-plane, 2D Green-Lagrange strains (£«, Eyy, Exy) were computed within tibiofemoral cartilage at each frame [Chan, D.D., et al., 2016. In vivo articular cartilage deformation: noninvasive quantification of intratissue strain during joint contact in the human knee. Sci Rep 6, 19220. https: / / doi.org / 10.1038 / srepl9220; Lee, W., et al., 2023. High frame rate deformation analysis of knee cartilage by spiral dualMRI and relaxation mapping. Magn Reson Med 89, 694-709. https: / / doi.org / 10.1002 / mrm.29487]. To calculate intratissue strain, the cartilage region was identified and manually selected from the DENSE magnitude image and isolated as an region of interest (ROI). Then, the raw displacements in both x and y directions within the selected ROI were calculated using the DENSE phase data. A locally weighted scatterplot smoothing (LOWESS) filter was applied to the raw displacements to obtain smoothed displacements (dxand dy). Finally, the in-plane mappings of the Green-Lagrange strains were calculated from the smoothed displacements. Data Analysis

[0069] We analyzed the linearity of the voltage-force relationship from the pneumatic cylinder using Pearson’s correlation analysis. The relationship between those two parameters was established by linear regression. The precision was quantified by the mean and pooled standard deviation (PSD) during the ascending section of the displacement and analyzed on a participant basis. We also separated the data into two groups, based on whether the eight-minute preconditioning time was included or not, to analyze the effect of preconditioning. The repeatability was also grouped and analyzed on a participant basis. The maximum displacement of each cycle was extracted, then quantified by the mean and PSD from each test. We also estimated the precision and repeatability at the tibiofemoral cartilage. The relaxometry results were reported as mean values within the selected regions of interest, along with the spatial maps. The DENSE displacement data was also reported as mean values within the selected regions of interest, along with the displacement and strain maps.

[0070] EXAMPLE 2 - RESULTS

[0071] Calibration of the Pneumatic Cylinder

[0072] The regression of the mean of the output force versus the input voltage confirmed a linear relationship for both varus and valgus loading scenarios (FIG. 10).

[0073] In-Lab Verification and Validation

[0074] The varus motion of the participant’s lower leg and the immobilization of the participant’s thigh were visually confirmed by the researchers. The precision data (FIG. 12) showed the averaged PSDs of the displacement during the ascending part of the cyclic loading of 3 participants were 270pm / 351pm (with / without preconditioning or preloading of the viscoelastic joint tissues), 300pm / 323pm, and 346pm / 380pm, which corresponded to 18pm / 23pm, 21pm / 22pm, and 20pm / 22pm at the tibiofemoral cartilage location. Meanwhile, the maximum PSDs were 363pm / 434pm, 349pm / 365pm, and 445pm / 470pm, which corresponded to 24pm / 29pm, 24pm / 25pm, and 26pm / 28pm at the location of tibiofemoral cartilage.

[0075] For repeatability, the standard deviations of the averaged maximum displacements of 3 participants were 492pm / 555pm, 921pm / 1009pm, and 3149pm / 3148pm, which corresponded to 32pm / 37pm, 64pm / 70pm, and 186pm / 187pm at the location of tibiofemoral cartilage (FIG. 13).

[0076] MRI Verification and Validation The averaged Tl, T2, T2*, and T1 p values within the ROIs before / after a 20-minute 25N constant loading were 1026ms / 1028ms, 31.16ms / 34.04ms, 21.15ms / 26.13ms, and

[0077] 37.26ms / 35.29ms (FIG. 14B).

[0078] The DENSE images showed that the participant’s knee was under translational motion and the tibia was rotating clockwise under varus load. The images also showed that the medial side joint space was shrinking followed by inter-cartilage compression, while the lateral side joint space was expanding. Those visual observations were confirmed by the displacement map, where the cartilage moved Omm, 0.85mm, 1.96mm, 2.39mm towards the negative x direction and 0mm, 0.22mm, 0.45mm, 0.58mm towards the positive y direction on average at frame 1, 10, 19, 27 (t=0, 0.37s, 0.7s, Is, respectively) (FIG. 16). This transition happened gradually through all 27 frames.

[0079] The displacement variations within the cartilage resulted in intratissue strain (FIG. 17). During the loading, we found that at frame 1, the values of £«, Eyy, and Exywere close to 0; while at frame 27, the strain values within ROI ranged from -4.8% to and -3.6% to 0.2% for Exy. The transition was gradual over time and was observed through all frames.

[0080] An MRI-compatible varus-valgus loading apparatus for in vivo human knees has been designed, verified, and validated as demonstrated herein. The major findings are as follows: (1) The customed varus-valgus loading apparatus could apply compressive load to the medial condyle; (2) The loading apparatus could provide a high-precision loading scheme; (3) The presence of the device did not hinder qMRI data acquisitions; (4) The cartilage intratissue displacement and strain in response to the cyclic loading could be captured through DENSE MRI.

[0081] The input voltage and the output force show an almost-perfect linear relationship. The imperfection is likely due to pressurized air leakage and friction. The input voltages (0-1 V) and corresponding output forces (0-100N) are selected to cover the range of potential forces that will be practically used (15N-30N). The varus and valgus voltage-force relationship is slightly different due to the geometrical difference in the cross-sectional area of the piston, internal friction, and uneven deformation of the cylinder under pressure.

[0082] The precision results proved that the device could maintain high-level precision throughout the scanning session. Here, the estimated PSD values at the tibiofemoral cartilage location are noticeably lower than the MR images’ pixel size (360 u m), indicating that the inconsistency of the motion induced by the device is almost undetectable by the images throughout the entire imaging session. Even in the worst-case scenario, the maximum PSD is still noticeably lower than the pixel size. This is important because the MRI is inherently sensitive to motion as it may cause various imaging artifacts [Godenschweger, F., et al., 2016. Motion correction in MRI of the brain. Phys Med Biol 61, R32-R56. https: / / doi.Org / 10.1088 / 0031-9155 / 61 / 5 / R32], Meanwhile, the comparison between the non-preconditioned and preconditioned data reveals that the 8-minute preconditioning helps minimize the standard deviation during the cyclic loading. The preconditioning session can assist the cartilage to reach a quasi-steady state given that the cartilage is viscoelastic [Hayes, W.C., Mockros, L.F., 1971. Viscoelastic properties of human articular cartilage. J Appl Physiol 31, 562-568. https: / / doi.Org / 10.1152 / jappl.1971.31.4.562; Mak, A.F., 1986. The Apparent Viscoelastic Behavior of Articular Cartilage — The Contributions From the Intrinsic Matrix Viscoelasticity and Interstitial Fluid Flows. J Biomech Eng 108, 123-130. https: / / doi.Org / 10.1115 / l.3138591; Martin, K.J., Neu, C.P., Hull, M.L., 2009. Quasi-Steady-State Displacement Response of Whole Human Cadaveric Knees in a MRI Scanner. J Biomech Eng 131. https: / / doi.Org / 10.1115 / l.2978986; Neu, C.P., Hull, M.L., 2003. Toward An MRI-Based Method to Measure Non-Uniform Cartilage Deformation: An MRI-Cyclic Loading Apparatus System and Steady-State Cyclic Displacement of Articular Cartilage Under Compressive Loading. J Biomech Eng 125, 180-188. https: / / doi.org / 10.1115 / L 1560141], functions as a wearin process for the participant to familiarize the motion and allows the linear bearing system to selflubricate to reduce friction. All these will lead to smoother and more consistent motion and thus result in better image quality.

[0083] The repeatability results show the difference induced by the apparatus itself throughout different scanning sessions is unlikely to be noticed in the images, hence it is practical for longitudinal studies. The estimated PSD values at the tibiofemoral cartilage location are smaller than the MR images’ pixel size, even though those values are significant at the ankle. However, the standard deviation value is quite high compared to those from the precision test, which indicates that the major source of the error in the longitudinal study may come from the inconsistency of the position of the participant between scans. To minimize the discrepancy, the participant should maintain similar, if not identical, positions among different scans.

[0084] The force and motion estimation method is based on viewing the lower leg as a rigid body rotating in the coronal plane with respect to the intercondylar eminence. However, the force applied to the ankle has to be transmitted through various kinds of tissues like bones and ligaments to finally reach the tibiofemoral cartilage. The viscoelastic properties of different tissues will inevitably dampen the effective force and motion [Fukashiro, S., Noda, M., Shibayama, A., 2001. In vivo determination of muscle viscoelasticity in the human leg. Acta Physiol Scand 172, 241-248. https: / / doi.Org / 10.1046 / j.1365-201x.2001.00866.x; Ristaniemi, A., Stenroth, L., Mikkonen, S., Korhonen, R.K., 2018. Comparison of elastic, viscoelastic and failure tensile material properties of knee ligaments and patellar tendon. J Biomech 79, 31-38. https: / / doi.Org / 10.1016 / j.jbiomech.2018.07.031; Tsuchikane, A., Nakatsuchi, Y., Nomura, A., 1995. The Influence of Joints and Soft Tissue on the Natural Frequency of the Human Tibia Using the Impulse Response Method. Proc Inst Meeh Eng H 209, 149-155. https: / / doi.org / 10.1243 / PIME_PROC_1995_209_337_02], Therefore, the force and motion obtained using the rigid body model are an overestimation of those exerted on the tibiofemoral cartilage.

[0085] In the MRI session, because MR imaging artifacts induced by foreign objects have been frequently observed [Hargreaves, B.A., Worters, P.W., Pauly, K.B., Pauly, J.M., Koch, K.M., Gold, G.E., 2011. Metal -induced artifacts in MRI. AJR Am J Roentgenol 197, 547-55. https: / / doi.org / 10.2214 / AJR. l l.7364; Hecht S, et al. Magnetic resonance imaging susceptibility artifacts due to metallic foreign bodies. Vet Radiol Ultrasound. 2011 Jul-Aug;52(4):409-14. doi: 10.1111 / j.1740-8261.2011.01809.x. Epub 2011 Mar 7. PMID: 21382122; Krupa, K., Bekiesinska- Figatowska, M., 2015. Artifacts in Magnetic Resonance Imaging. Pol J Radiol 80, 93-106. https: / / doi.org / 10.12659 / PJR.892628], we need to guarantee that the presence of the loading apparatus will not affect the relaxometry measurements. Our reported relaxometry values are comparable with other literature [Dunn, T.C., et al., 2004. T2 Relaxation Time of Cartilage at MR Imaging: Comparison with Severity of Knee Osteoarthritis. Radiology 232, 592-598. https: / / doi.org / 10.1148 / radiol.2322030976; Li, X., et al., 2007. In vivo Tip and T2 mapping of articular cartilage in osteoarthritis of the knee using 3T MRI. Osteoarthritis Cartilage 15, 789- 797. https: / / doi.Org / 10.1016 / j.joca.2007.01.011; Mittal, S., et al., 2019. T1 and T2 mapping of articular cartilage and menisci in early osteoarthritis of the knee using 3-Tesla magnetic resonance imaging. Pol J Radiol 84, 549-564. https: / / doi.org / 10.5114 / pjr.2019.91375; Newbould, R.D., et al., 2012. T2* measurement of the knee articular cartilage in osteoarthritis at 3T. Journal of Magnetic Resonance Imaging 35, 1422-1429. https: / / doi.org / 10.1002 / jmri.23598; Yoon, H.J., et al., 2014. T2 Values of Femoral Cartilage of the Knee Joint: Comparison between Pre-Contrast and Post-Contrast Images. Korean J Radiol 15, 123. https: / / doi.org / 10.3348 / kjr.2014.15.L 123], confirming that acquisition and measurement of relaxometry values is not disrupted by the loading device. Additionally, the relaxometry values before and during a constant 25N varus loading are similar, implying that the static loading may not induce critical biochemical changes within the cartilage. The strain pattern indicates that the varus load at the ankle is effectively translated to the compressive force in the tibiofemoral cartilage. Internal strain patterns of Eyyshow that the location of maximum compression gradually transitions from the center to the edge of the medial tibiofemoral cartilage throughout the loading session. The strain patterns of Exx show that the cartilage undergoes expansion in the x direction, while being compressed in they direction. This is consistent with the finding that the cartilage can be viewed as an incompressible material after it reaches a steady state [Bachrach, N.M., Mow, V.C., Guilak, F., 1998. Incompressibility of the solid matrix of articular cartilage under high hydrostatic pressures. J Biomech 31, 445-451. https: / / doi.org / 10.1016 / S0021-9290(98)00035-9], The Exypattern shows that the cartilage is also under shear stress, and those values are comparable to the values. The magnitude of Exx and Exyis lower than that of Eyy, which indicates the cartilage is mainly under compressive force.

[0086] An MRI-compatible varus-valgus loading apparatus that can provide a stable, robust, and precise platform to obtain image-based intratissue displacement, strain, and relaxometry of the human knee in vivo has been designed, verified, and validated as shown herein. The device taught herein is a novel approach to implement varus-valgus stress test in a clinical 3T MRI setup to acquire in vivo full-field (i.e., pixel-based spatial maps of) displacement and strain of the tibiofemoral cartilage and obtain relaxometry values under load. The loading apparatus and related methodology equip us with a unique way to quantify biochemical and biomechanical changes in articular cartilage during mechanical loading and hence provide us with a deeper insight into their pathological changes during the onset of OA pathogenesis.

[0087] EXAMPLE 3 - In vivo elastography reveals cartilage stiffness changes linked to altered gait patterns post- ACL surgery

[0088] Osteoarthritis progression is marked by cartilage softening, a phenomenon that is difficult to quantify non-invasively. Here, we present an MRI-based method that combines displacement encoded with stimulated echoes MRI with an iterative finite elements-based optimization approach to quantify relative cartilage stiffness maps. Our method overcomes the depth limitations of traditional ultrasound elastography and during validation demonstrated accurate stiffness map reconstruction within <6% error, even in the presence of noise. Our technique enables the in vivo quantification of relative cartilage stiffness maps, as well as stiffness heterogeneity. Applying our method within a small cohort of participants at risk for the development of OA due to prior ligament reconstruction, we show that in this population cartilage stiffness heterogeneity is correlated with altered loading patterns during gait, as well as patient-reported outcome scores of physical function and pain. This approach enables the study of in vivo cartilage biomechanics and identifies cartilage stiffness gradients as a possible biomarker for osteoarthritis pathogenesis.

[0089] Osteoarthritis (OA) is a leading cause of joint pain, disability, and health care costs globally. OA is characterized as a complex whole joint disease that involves progressive articular cartilage degradation, marked by cartilage softening and eventual cartilage loss, and often leads to pain, disability, and joint replacement surgery [Kraus VB, and Karsdal MA. Osteoarthritis: Current Molecular Biomarkers and the Way Forward. Calcif Tissue Int. Springer; 2020. p. 3. doi: 10.1007 / s00223-020-00701-7; Katz JN, et al., Diagnosis and Treatment of Hip and Knee Osteoarthritis: A Review. JAMA - Journal of the American Medical Association. American Medical Association; 2021. p. 568-578. doi: 10.1001 / jama.2020.22171.] OA is influenced by a complex interplay of multiscale biomechanical, metabolic, and genetic factors, often triggered by injury and exacerbated by inflammation [Thomas AC, et al., Epidemiology of Posttraumatic Osteoarthritis. J Athl Train. National Athletic Trainers Association; 2017;52(6):491. doi: 10.4085 / 1062-6050-51.5.08], The pathophysiology of OA often involves a point-of-onset joint injury (e.g., ligament rupture), which is followed by articular cartilage matrix changes such as collagen matrix disorganization and proteoglycan loss [Diamond LE, et al., Osteoarthritis year in review 2023: Biomechanics. Osteoarthritis Cartilage. W.B. Saunders; 2024;32(2): 138-147. doi: 10.1016 / J.JOCA.2023.11.015], This structural loss of cartilage major macromolecules manifests mechanically as altered cartilage strains during joint loading as well as altered cartilage stiffness (e.g. cartilage softening and changes to local cartilage stiffness heterogeneity) [ Cai L, et al., Finite deformation elastography of articular cartilage and biomaterials based on imaging and topology optimization. Sci Rep. Nature Research; 2020; 10(1): 1—12. doi: 10.1038 / s41598-020-64723-9], Surgical reconstruction following joint trauma such as an anterior cruciate ligament (ACL) tear is the gold standard treatment for ligament rupture, but does not reliably prevent the progression of cartilage degeneration and onset of post-traumatic OA [Wang LJ, et al., Post-traumatic osteoarthritis following ACL injury. Arthritis Res Ther. BioMed Central Ltd.; 2020;22(l): l-8. doi: 10.1186 / S13075-020-02156-5 / FIGURES / 1; Bodkin SG, et al., Post-traumatic osteoarthritis diagnosed within 5 years following ACL reconstruction. Knee Surgery, Sports Traumatology, Arthroscopy. Springer; 2020;28(3):790-796. doi: 10.1007 / s00167-019-05461-y; Dilley JE, et al., Post-traumatic osteoarthritis: A review of pathogenic mechanisms and novel targets for mitigation. Bone Rep. Elsevier; 2023;18: 101658. doi: 10.1016 / J.BONR.2023.101658], The exact mechanisms linking joint trauma to ongoing cartilage degeneration are unclear, though it is theorized that two possible mechanisms may mediate post-traumatic OA. First, the force required to rupture the ligament in the initial joint injury, may also inflict acute cartilage trauma, triggering a degenerative cascade of inflammatory processes, including chondrocyte apoptosis, mitochondrial dysfunction, and elevated reactive oxygen species, ultimately leading to further cartilage degeneration. Second, the ACL reconstruction surgical procedure does not fully restore normal joint mechanics, resulting in chronic changes to the local mechanical environment of the knee joint. Disrupted cartilage loading patterns may also initiate a degenerative cascade, worsening clinical outcomes over time. The relationship between joint mechanical loading and cartilage degeneration is intricate: Insufficient mechanical loading may contribute to articular cartilage thinning, whereas repetitive joint use and joint overload may lead to the loss of major macromolecules and cartilage degeneration [Herzog W, and Federico S. Considerations on joint and articular cartilage mechanics. Biomech Model Mechanobiol. Biomech Model Mechanobiol; 2006;5(2-3):64-81. doi: 10.1007 / sl0237-006-0029-y], Furthermore, in the regions where tibiofemoral cartilage contact occurs, both underloading in regions normally subject to higher loads and overloading in regions unaccustomed to load during weight-bearing activities, such as gait, may exacerbate the degenerative processes that contribute to the development of post- traumatic OA.

[0090] The majority of ACL tears occur in patients younger than 30, which may significantly alter their quality of life over the coming decades. Within 10 years of ACL injury, the prevalence of symptomatic OA approaches 80%. Given that the lifespan of total knee arthroscopic implants is limited to 20-30 years, individuals who tear an ACL early in life may experience severe cartilage degeneration but could still be ineligible for a knee replacement due to young age. In this population, early detection and intervention are critical due to a clear “start point” for OA development. Unfortunately, the current standard of care for OA diagnosis is radiography, which reliably images only cartilage loss and morphological changes typically seen in the later stages of OA, after irreversible changes to the knee joint have occurred [Eckstein F, et al. Magnitude and regional distribution of cartilage loss associated with grades of joint space narrowing in radiographic osteoarthritis - data from the Osteoarthritis Initiative (OAI). Osteoarthritis Cartilage. Elsevier Ltd; 2010;18(6):760-768. doi: 10.1016 / j.joca.2009.12.009], Preclinical MRI methods, such as quantitative MRI (qMRI), have aimed to identify joint tissue changes in pre-osteoarthritis, an early stage of disease progression that occurs prior to radiographic detection or patient-reported pain, when emerging disease-modifying interventions may be most effective [Chu CR. Can we afford to ignore the biology of joint healing and graft incorporation after ACL reconstruction? Journal of Orthopaedic Research. Wiley; 2021;jor.25145. doi: 10.1002 / jor.25145; Chu CR, and Williams AA. Quantitative MRI UTE-T2* and T2* Show Progressive and Continued Graft Maturation Over 2 Years in Human Patients After Anterior Cruciate Ligament Reconstruction. Orthop J Sports Med. SAGE Publications; 2019;7(8). doi: 10.1177 / 2325967119863056; Williams A, et al., Early articular cartilage MRI T2 changes after anterior cruciate ligament reconstruction correlate with later changes in T2 and cartilage thickness. Journal of Orthopaedic Research. John Wiley & Sons, Ltd; 2017;35(3):699-706. doi: 10.1002 / JOR.23358; Qian Y, et al., MultiComponent T2* Mapping of Knee Cartilage: Technical Feasibility Ex vivo. Magnetic resonance in medicine : official journal of the Society of Magnetic Resonance in Medicine / Society of Magnetic Resonance in Medicine. NIH Public Access; 2010;64(5):1426. doi: 10.1002 / MRM.22450.] However, it remains unclear if these MRI methods are sensitive cartilage changes at the earliest OA disease states and many are not yet clinically accessible to patients.

[0091] Elastography, a non-invasive imaging approach commonly used to quantify the elastic stiffness properties of soft tissues in the body, offers a promising method to assess cartilage softening. Quantifying material stiffness can provide valuable insights into structure-function relationships, particularly in load-bearing tissues such as cartilage, where softening is a hallmark of early OA progression and may be linked to greater changes to the mechanical environment of the knee joint [Xu X, et al. Chondrocyte Adipogenic Differentiation in Softening Osteoarthritic Cartilage. J Dent Res. SAGE Publications Inc.; 2022;101(6):655-663. doi: 10.1177 / 00220345211057539], Therefore, non-invasively quantifying changes in cartilage stiffness in vivo is of substantial interest as a potential diagnostic tool. While elastography appears ideal for this purpose, current conventional elastography methods, primarily using ultrasound, face limitations in penetrating deep into the musculoskeletal tissues of the knee joint space due to the dissipation of high frequency shear waves. Moreover, the use of high frequencies with low loads may not accurately reflect the stiffness properties of load-bearing tissues such as cartilage under physiological low- frequency and high-load activities such as walking [Davis ZR, et al. Intervertebral Disc Elastography to Relate Shear Modulus and Relaxometry in Compression and Bending. bioRxiv. bioRxiv; 2023; doi: 10.1101 / 2023.09.01.555817], Recently, spiral displacement encoded stimulated echo (spiral DENSE) MRI has been used to calculate pixel-level full-field displacement maps of tibiofemoral cartilage under repetitive physiological loads and loading frequencies [Zhu H, et al., In Vivo Human Knee Varus-Valgus Loading Apparatus for Analysis of Mri-Based Intratissue Strain and Relaxometry. doi: 10.2139 / SSRN.4569548; Lee W, et al., High frame rate deformation analysis of knee cartilage by spiral dualMRI and relaxation mapping. Magn Reson Med. John Wiley & Sons, Ltd; 2023;89(2):694-709. doi: 10.1002 / MRM.29487.] When full-field displacement measurements acquired with DENSE MRI are coupled to recent inverse modeling approaches relative stiffness can be quantified without assuming material homogeneity. [Miller CE, et al., Developing a biomechanical model-based elasticity imaging method for assessing hormone receptor positive breast cancer treatment-related myocardial stiffness changes. Journal of Medical Imaging. Society of Photo-Optical Instrumentation Engineers; 2021;8(5):56002- 56003. doi: 10.1117 / 1. JMI.8.5.056002; Mei Y, et al., Introducing regularization into the virtual fields method (VFM) to identify nonhomogeneous elastic property distributions. Comput Meeh. Springer Science and Business Media Deutschland GmbH; 2021;67(6): 1581-1599. doi: 10.1007 / S00466-021-02007-3 / FIGURES / 21; Grediac M, et al., The Virtual Fields Method for Extracting Constitutive Parameters From Full-Field Measurements: a Review. Strain. John Wiley & Sons, Ltd; 2006;42(4):233-253. doi: 10.1111 / J.1475-1305.2006.TB01504.X], These inverse modeling techniques have previously utilized experimental full-field displacement data as an input to calculate shear modulus in intervertebral disk, as well as radial and circumferential elastic modulus in cardiac tissue, and elastic modulus in soft-tissue phantoms [Miller CE, et al., Reproducibility assessment of a biomechanical model-based elasticity imaging method for identifying changes in left ventricular mechanical stiffness. Journal of Medical Imaging. Society of Photo-Optical Instrumentation Engineers; 2022;9(5). doi: 10.1117 / 1. JMI.9.5.056001.]

[0092] In the present study, we tested and validated an elasticity imaging framework where we combined two inverse modeling techniques: the virtual fields method [Luetkemeyer CM, et al., Constitutive modeling of the anterior cruciate ligament bundles and patellar tendon with full-field methods. J Meeh Phys Solids. Pergamon; 2021;156: 104577. doi: 10.1016 / J.JMPS.2021.104577] and biomechanical model-based elasticity imaging method [Weis JA, et al., A consistent pre- clinical / clinical elastography approach for assessing tumor mechanical properties in therapeutic systems. https: / / doi.org / 101117 / 122007425. SPIE; 2013;8672:342-348. doi: 10.1117 / 12.2007425], After validating the framework using simulated datasets representative of articular cartilage material properties and geometries, we applied the elasticity imaging framework in combination with spiral DENSE MRI to calculate relative stiffness maps of articular cartilage in a joint injury cohort study. Specifically, to evaluate pre-osteoarthritic cartilage stiffness changes, we spatially mapped the relative elastic modulus of medial compartment tibiofemoral cartilage in the ACL-reconstructed and contralateral knees of 20 participants who had undergone ACL reconstruction approximately one year prior. Motivated by the zonal structure of cartilage, following the calculation of relative elastic modulus maps in the cohort, we employed a technique to compute the spatial gradient of modulus (A modulus) [Wilson RL, et al., Spatial Gradients of Quantitative MRI as Biomarkers for Early Detection of Osteoarthritis: Data From Human Explants and the Osteoarthritis Initiative. Journal of Magnetic Resonance Imaging. John Wiley & Sons, Ltd; 2023;58(l): 189— 197. doi: 10.1002 / JMRI.28471], Spatial modulus gradients are independent of the scale of the relative elastic modulus, thus describing the true through-thickness variation of absolute cartilage stiffness. Subsequently, we regressed modulus gradient values against patient-reported outcome scores related to general physical function, pain, and quality of life to evaluate the relationship between cartilage stiffness gradient and clinical outcomes. Finally, in a subset of participants, we used markerless motion capture to quantify knee adduction moment and medial contact force during gait, both serving as surrogate measures of medial tibiofemoral joint compartment loading. We then regressed the cartilage modulus gradient against gait metrics to explore the relationship between macroscale changes in joint mechanics and microscale changes in cartilage mechanics.

[0093] RESULTS

[0094] Participant Characteristics: In total, 22 participants underwent scanning at approximately 12 months post-ACL reconstruction on both their ACL-reconstructed and contralateral limbs. Two individuals were excluded from the study due to motion or metal artifact interference with the cartilage region of interest (FIG. 18 A). Therefore, the final sample size consisted of 20 participants, comprising 8 males and 14 females. The average age of participants was 24.9 ± 5.9 years, with a mean body mass index (BMI) of 23.5 ± 3.4. The demographic and surgical characteristics of all participants in the cohort, as well as those in the gait analysis subgroup, are detailed in Table 5. Notably, the subgroup selected for gait analysis by markerless motion analysis included 9 participants, comprising 6 females and 3 males, with an average age of 23.0 ± 3.1 years and an average BMI of 23.3 ± 3.1.

[0095] Validation of elasticity imaging framework: For all five synthetic models, the average percent error was found to be less than 6% (Table 6, FIG. 19). Upon qualitative analysis of the percent error maps (FIG. 19D, 19F), it should be noted that errors increase when the ground truth modulus maps exhibit sharp transitions between stiff and soft moduli. Given the expected gradual transition from soft to stiff moduli in biological cartilage, even during osteoarthritis pathogenesis, we anticipate that errors would remain low within biological cartilage. In Monte Carlo simulations assessing response to noise, the addition of LOWESS smoothing to displacements prior to stiffness calculation rendered the stiffness calculation robust against increasing noise levels from MRI acquisition (FIG. 19G). Applying six rounds of LOWESS smoothing to the displacements minimized the average percent error to less than 5% across all noise cases (FIG. 19G). The error in the functional gradient index remained less than 4% when moduli were scaled to various ranges, indicating that the functional gradient index is representative of the true absolute stiffness gradient, regardless of the absolute stiffness range (Table 7). Relationships between Modulus Metrics and Patient Factors: Statistically significant associations (p < 0.05) were found between the variance of modulus and age, indicating that higher age was associated with lower variance in modulus (p = 0.02, Table 8). Average gradient index values were significantly lower in the femoral region of interest compared to the tibial region of interest (p = 0.03, Table 8). Differences were also observed based on graft type, with higher gradient variances noted in participants with a quadriceps graft (p = 0.03, Table 8). However, the limited sample size of quadriceps tendon grafts (2 out of 20, with the remaining 18 having patellar tendon grafts) within the population may limit the power to reliably discern group differences based on graft type. When all strains were regressed against modulus metrics, only shear strain and the variance of the modulus were found to be significant in the tibial cartilage (p=0.02). All other adjusted regressions of modulus metrics against other strains, or T2, T2*, or Tip, were found to be insignificant.

[0096] Modulus gradient measurements and patient-reported outcomes exhibit variable correlations: Both unadjusted and adjusted regressions were conducted for each cartilage region (femoral or tibial region, ACL-reconstructed or contralateral limb) to explore the relationship between participants' reported outcomes and modulus metrics. The adjusted regressions, depicted in FIG. 20, controlled for age. Several significant correlations emerged from the analysis of modulus gradient data. Specifically, in ACL-reconstructed tibial cartilage, increases in the variance of the gradient were associated with lower KOOS scores, indicating worse physical function (FIG. 20A, p = 0.04). Similar associations were observed with the KOOS quality of life subscore, where increases in both the gradient index and the variance of the gradient were linked to lower (thus worse) KOOS quality of life scores (FIG. 20C, p = 0.002, p = 0.013, respectively). In the contralateral limb, intriguing trends were noted in the relationship between gradient index and KOOS Pain score. In contralateral limb femoral cartilage, a significant positive correlation was observed between the gradient index and KOOS pain score, suggesting that higher gradient index values in the femoral cartilage were associated with less pain. In contrast, in tibial cartilage, increases in the gradient index were associated with increased pain (FIG. 20B, p = 0.01, p = 0.01, respectively).

[0097] Strong correlations observed between modulus metrics and gait metrics in the ACL-reconstructed limb: During the loading phase of gait analysis, we calculated the peak knee adduction moment (KAM) and medial contact force (MCF) for each participant for both limbs. We conducted both unadjusted and adjusted regressions for each cartilage region and limb to investigate the relationship between modulus metrics and gait patterns among participants. The adjusted regressions were controlled for age. Significant correlations were observed exclusively in the ACL-reconstructed limb, with no significant relationships found in the contralateral limb. In femoral cartilage, we observed negative correlations between the gradient index, variance of gradient, and variance of modulus with knee adduction moment, indicating that higher modulus gradient measurements were associated with lower knee adduction moment (FIG. 20B, p < 0.001, p = 0.005, and p = 0.036, respectively). Conversely, in the tibial cartilage of the ACL- reconstructed limb, strong positive correlations were found between the gradient index and variance of gradient with knee adduction moment (FIG. 20B, p = 0.05 and p = 0.03, respectively). A similar trend was identified in tibial cartilage with medial contact force, where increases in gradient index and variance of gradient were associated with higher medial contact forces during gait (p = 0.002 and p = 0.001, respectively).

[0098] After a traumatic joint injury such as a ligament tear, it is well-established that individuals may develop osteoarthritis, even after undergoing ligament reconstructive surgery [Wang LJ, et al., Post-traumatic osteoarthritis following ACL injury. Arthritis Res Ther. BioMed Central Ltd.; 2020;22(l): 1-8], Therefore, cohorts studied shortly after ligament reconstruction present a unique opportunity to evaluate early stages of cartilage degeneration. Developing methods to monitor these pre-osteoarthritic states is crucial for intervening before irreversible joint changes occur. Quantitative assessment of both macroscale joint mechanical patterns and microscale cartilage mechanical behavior may serve as early indicators of these changes, which are closely linked to long-term joint health [Lad NK, Liu B, Ganapathy PK, et al. EFFECT OF NORMAL GAIT ON IN VIVO TIBIOFEMORAL CARTILAGE STRAINS. J Biomech. NIH Public Access; 2016;49(13):2870.; Garcia SA, White MS, Gallegos J, Balza I, Kahan S, Palmieri-Smith RM. Associations between Body Mass Index, Gait Mechanics and Trochlear Cartilage Thickness in those with ACL Reconstruction. Med Sci Sports Exerc. Med Sci Sports Exerc; 2024; Williams JR, Neal K, Alfayyadh A, et al. Patellof emoral contact forces and knee gait mechanics 3 months after ACL reconstruction are associated with cartilage degradation 24 months after surgery. Osteoarthritis Cartilage. W.B. Saunders; 2023;31(1):96— 105; Williams JR, Neal K, Alfayyadh A, et al. Knee cartilage T 2 relaxation times 3 months after ACL reconstruction are associated with knee gait variables linked to knee osteoarthritis. Journal of Orthopaedic Research. Wiley; 2021;jor.25043], This study validated a DENSE MRI-based elasticity imaging framework for assessing articular cartilage stiffness and stiffness variability in participants one-year post-ACL surgery. Additionally, gait mechanics during walking were also evaluated. The findings revealed several important results: increased stiffness gradients and gradient variance in tibial cartilage were associated with worse patient-reported outcome scores, whereas in femoral cartilage, increased stiffness gradient was associated with better outcomes. Similarly, higher stiffness gradient, gradient variance, and modulus variance in femoral cartilage correlated with reduced knee adduction moment, whereas in tibial cartilage, the opposite trend was observed: increased stiffness gradient and gradient variance were associated with higher knee adduction moment. These results highlight that macroscale mechanical patterns may influence cartilage microscale mechanical properties differently, underscoring the importance of understanding site-specific and localized cartilage degeneration in osteoarthritis.

[0099] Our findings align with previous studies that highlight the site-specific biomechanical properties of articular cartilage [Linus A, Tanska P, Nippolainen E, et al. Site-specific elastic and viscoelastic biomechanical properties of healthy and osteoarthritic human knee joint articular cartilage. J Biomech. J Biomech; 2024;169.] While various forms of elastography have been utilized to assess cartilage mechanical properties in ex vivo or in high-field animal MRI studies, our study represents the first successful demonstration of an in vivo cartilage elastography algorithm in humans and at clinical MRI field strengths. The absence of correlations between elastography metrics and quantitative MRI relaxometry techniques, as well as with cartilage strains other than shear strain in the tibial cartilage, underscores the unique capability of cartilage elastography to offer distinct and supplementary insights compared to existing methodologies. This highlights the novelty of elastography as an emerging technique that contributes complementary information to current methods in cartilage assessment. Moreover, in this initial application of cartilage elastography in a human cohort at risk for developing osteoarthritis, the strong correlations we found between modulus measurements and surrogate measures of knee loading during gait, as well as patient-reported outcomes, emphasize the clinical relevance of cartilage elastography as a diagnostic tool for osteoarthritis. Previous research has indicated altered frontal plane knee kinematics during walking in participants with post-traumatic OA. Mechanistically, unloading of the knee after ACL reconstruction may shift the location and magnitude of forces applied to the cartilage during gait, potentially disrupting cartilage structural integrity and leading to changes in regional stiffness and stiffness gradients. [Williams IR, Neal K, Alfayyadh A, et al. Patellof emoral contact forces after ACL reconstruction: A longitudinal study. I Biomech. Elsevier; 2022;134: 110993.] The varying relationships observed between stiffness gradient metrics, outcomes, and gait measures in the femoral and tibial medial compartments likely stem from the localized and site-specific adaptation of articular cartilage to its mechanical environment. Tibial cartilage is subjected to greater compressive forces under both dynamic and static loading conditions, whereas femoral cartilage experiences higher shear forces during dynamic knee movements. Our findings support the association of lower stiffness gradients in femoral cartilage and higher stiffness gradients in tibial cartilage with poorer outcomes and increased knee adduction moments. Importantly, the literature on knee adduction moments in the early post-ACL reconstruction phase presents conflicting evidence. Some studies suggest that reduced loading may lead to long-term osteoarthritis development, while others indicate the same trend with increased loading. [Titchenal MR, et al. Cartilage Subsurface Changes to Magnetic Resonance Imaging UTE-T2* 2 Years After Anterior Cruciate Ligament Reconstruction Correlate With Walking Mechanics Associated With Knee Osteoarthritis. Am J Sports Med. Am J Sports Med; 2018;46(3):565-572.] The mechanistic underpinnings linking adverse outcomes, altered knee loading patterns, and modified stiffness gradients remain complex and challenging to elucidate. Individuals experiencing greater pain, influenced by factors such as rehabilitation adherence or specific surgical techniques like graft tensioning, may adjust their knee loading patterns during daily activities, potentially resulting in localized mechanical adaptations in cartilage stiffness properties. Alternatively, intricate and difficult-to-measure factors such as underlying genetic predispositions or the nature of the initial joint injury may contribute to changes in cartilage material properties, subsequently affecting pain perception and prompting adjustments in gait mechanics aimed at minimizing discomfort. Large-scale longitudinal studies are essential to unravel the relationship between pain, joint loading, and cartilage mechanical properties. Ideally, joint loading and elastography data would be collected from a sizable sample prior to injury, enabling a clearer understanding of whether altered gait precedes changes in mechanical properties leading to pain, or vice versa.

[0100] Several limitations should be acknowledged in our study. Firstly, the elastography method assumed a uniform nearly incompressible Poisson’s ratio across all participants. Previous research has indicated that the Poisson’s ratio of cartilage may vary significantly between individuals and joint locations. Moreover, the elastography method employed in our study assumed linear elastic material properties in cartilage. However, it is well-recognized that cartilage is best modeled by biphasic or triphasic material models, which are considerably more complex than the linear elastic assumption used here. [Jones B, et al., Biphasic Analysis of Cartilage Stresses in the Patellofemoral Joint. J Knee Surg. NTH Public Access; 2016;29(2):92.] Moreover, the study's sample size of 20 participants may limit the generalizability of the results. Variations in participant characteristics, such as age, BMI, and adherence to rehabilitation protocols, could potentially influence the outcomes, although adjusted regressions were employed to mitigate some of the introduced variability from covariate data. Furthermore, subjective patient-reported outcomes, such as pain assessments, may not fully capture the underlying condition of the cartilage. Future research endeavors should involve larger and more diverse cohorts, alongside objective measures of pain and function, to confirm and extend these findings.

[0101] In conclusion, our study validated and applied a novel elasticity imaging framework capable of assessing cartilage stiffness changes that may precede symptomatic osteoarthritis development. To our knowledge, this is the first application of articular cartilage elastography in vivo. The association of cartilage stiffness gradient metrics with both patient outcomes and joint loading during daily activities emphasizes the pivotal role of macroscale mechanical factors in modulating microscale cartilage mechanics. These findings underscore the critical need for early monitoring and intervention to mitigate long-term joint degeneration risks. Advances in imaging and computational techniques hold promise for early detection of cartilage changes and guiding tailored treatment approaches to improve outcomes for ACL reconstruction patients. Future research should prioritize larger cohorts, longitudinal evaluation, and employ objective measures to further elucidate the intricate interplay between macroscale mechanics and microscale cartilage degeneration mechanisms.

[0102] MATERIALS AND METHODS

[0103] Participants: A cohort study was conducted, enrolling participants at the at the University of Colorado Anschutz orthopedic clinic who had recently undergone unilateral ACL reconstructive surgery (FIG. 18A). The study received approval from the University of Colorado IRB, and informed consent was obtained from all participants. Participants were selected based upon specific inclusion and exclusion criteria: included were individuals aged 18 to 40 who had undergone ACL surgery within three months of injury, with no prior symptomatic knee pathologies, and a body mass index (BMI) between 18 and 35. Participants with concomitant knee injuries, such as meniscal tears, were eligible for inclusion. Exclusion criteria comprised any previous knee surgeries prior to the ACL injury or concomitant cartilage damage on either limb. ACL reconstruction procedures utilized either a bone-patella tendon-bone autograft or a quadriceps tendon autograft. Rehabilitation adherence was not actively monitored, though all collaborating surgeons adhered to the MOON ACL protocol for guiding return to physical activity. At approximately twelve months post-ACL reconstruction, an MRI scan was conducted on participants’ ACL-reconstructed and contralateral limbs. Participants completed the Knee Injury and Osteoarthritis Outcome Score (KOOS) questionnaire on the day of the study visit, directly prior to the MRI session.

[0104] Markerless motion capture and musculoskeletal simulation'. A subset of participants from the larger joint injury cohort underwent gait analysis using cloud-based software [Uhlrich SD, et al., OpenCap: Human movement dynamics from smartphone videos. PLoS Comput Biol. Public Library of Science; 2023;19(10):el011462], Nine individuals participated in gait analysis after completing patient-reported outcome forms and before their MRI scan. T-tests were conducted to ensure there were no significant differences in age, time from surgery, BMI, or patient-reported outcome scores between the subgroup of nine included in gait analysis and the larger group of 20. Each participant completed three overground gait collections at a self-selected walking speed. All trials were recorded using smartphone video with the markerless motion software. Advanced deep learning models, in conjunction with inverse kinematics, were utilized to estimate kinematics from the smartphone video recordings. A physics-based musculoskeletal simulation methodology was used to estimate dynamics. For each limb, the peak knee adduction moment (KAM) and medial contact forces (MCF) were calculated during the loading phase of each trial and averaged for statistical analysis.

[0105] Loading Protocol of Human Knee: The above-described MRI-compatible loading apparatus was utilized to apply a varus load to the tibiofemoral cartilage synchronized with DENSE MRI acquisition (FIG. 18A) [Zhu H, et al., In Vivo Human Knee Varus-Valgus Loading Apparatus for Analysis of Mri-Based Intratissue Strain and Relaxometry], The load, equivalent to 0.5 times the participants’ body weight at the medial tibiofemoral cartilage, was applied perpendicular to the ankle joint, and calculated using the principle of moment conservation. Participants underwent a cyclic loading regime designed to mimic a walking cadence, with 1 -second load and unload intervals [Zhu H, et al., In Vivo Human Knee Varus-Valgus Loading Apparatus for Analysis of Mri-Based Intratissue Strain and Relaxometry; Lee W, et al., High frame rate deformation analysis of knee cartilage by spiral dualMRI and relaxation mapping. Magn Reson Med. ohn Wiley & Sons, Ltd; 2023;89(2):694-709], To minimize the viscoelastic effect and allow the tibiofemoral cartilage to reach a steady-state load-deformation response, participants underwent an eight- minute cyclic preconditioning load directly prior to DENSE acquisition. [Lee W, et al., Multiframe biomechanical and relaxometry analysis during in vivo loading of the human knee by spiral dualMRI and compressed sensing. Magn Reson Med. ohn Wiley & Sons, Ltd; 2023;90(3):995- 1009; Chan DD, et al., In vivo articular cartilage deformation: Noninvasive quantification of intratissue strain during joint contact in the human knee. Sci Rep. Nature Publishing Group; 2016;6(l):l-14 ]

[0106] MRI Scanning Protocol: A previously described MRI knee imaging protocol was utilized to scan each participant’s ACL-reconstructed and contralateral knees. MRI sequences included T2, T2*, and Tip relaxometry measurements (FIG. 18 A), and the DENSE sequence (FIG. 18 A), with all images taken in the coronal plane. All imaging was conducted using a 3T MRI system equipped with a 15-channel quadrature knee coil. During the study visit, participants were initially scanned on their ACL reconstructed limb, given a break, and subsequently scanned on their contralateral limb. MRI scanning time for each knee was just under an hour.

[0107] Data Processing: A custom semi-automatic cartilage segmentation algorithm was used to segment the medial articular cartilage into femoral and tibial articular cartilage regions of interest (ROIs) for each sequence (FIG. 18B). ROIs were standardized across both the DENSE and quantitative MRI scans using standardized knee anatomical definitions. Participants were excluded from all analysis if motion or metal artifacts obscured any ROI (FIG. 18 A). To calculate cartilage displacement maps, phase images output by the DENSE sequence were unwrapped using a custom Goldstein branch-cut phase unwrapping algorithm. Raw displacements for each pixel within the cartilage ROIs were determined from the unwrapped phase data [Chan DD, et al., In vivo articular cartilage deformation: Noninvasive quantification of intratissue strain during joint contact in the human knee. Sci Rep. Nature Publishing Group; 2016;6(l): 1- 14] and subsequently smoothed using a locally weighted scatterplot smoothing (LOWESS) filter (FIG 18B) [Chan DD, et al., Displacement smoothing for the precise MRI-based measurement of strain in soft biological tissues. Comput Methods Biomech Biomed Engin. Taylor & Francis; 2013; 16(8):852— 860.] To facilitate comparisons between strain and elastic modulus, in-plane Green-LaGrange strain fields in the transverse, axial, and shear directions were calculated from all displacement fields for each ROI [Geers MGD, et al., Computing strain fields from discrete displacement fields in 2D-solids. Int I Solids Struct. Pergamon; 1996;33(29):4293-4307. doi: 10.1016 / 0020-7683(95)00240-5] Smoothed full-field displacements served as input to the elasticity imaging framework to calculate spatial maps of relative modulus (FIG. 18B). Each ROI (femoral or tibial medial articular cartilage), and each limb (contralateral or ACL-reconstructed) underwent separate processing through the elasticity imaging framework to determine spatial maps of relative modulus.

[0108] Inverse modeling to determine relative elastic moduli: The elasticity imaging framework integrates two inverse modeling techniques to estimate relative moduli of cartilage (FIG. 18C). The inverse modeling can include various degrees of complexity of the choice of material (constitutive; e.g., elastic, viscoelastic, multiphasic, etc.) law, dimensionality (two- or three- dimensions), or anisotropy (linear or nonlinear). A brief overview of these methods is provided, [see also Mei Y, et al., Introducing regularization into the virtual fields method (VFM) to identify nonhomogeneous elastic property distributions. Comput Meeh. Springer Science and Business Media Deutschland GmbH; 2021;67(6): 1581-1599.] For this demonstration, we assumed plane strain, a non-homogenous linear elastic material model, and a nearly incompressible Poisson ratio of 0.49. Initially, an elastic stiffness map was computed based on observed displacements using the first inverse modeling method, the virtual fields method. [Grediac M, et al., The Virtual Fields Method for Extracting Constitutive Parameters From Full-Field Measurements: a Review. Strain. John Wiley & Sons, Ltd; 2006;42(4):233-253. ; Avril S, et al., Sensitivity of the virtual fields method to noisy data. Comput Meeh. Springer Verlag; 2004;34(6):439-452.] The virtual fields method operates on the principle of virtual work, where a cost function minimizes the difference between internal and external virtual work to solve for energy -minimizing material constants. A regularization term was employed within the cost function to ensure convergence and output of moduli. [Mei Y, et al., Introducing regularization into the virtual fields method (VFM) to identify nonhomogeneous elastic property distributions. Comput Meeh. Springer Science and Business Media Deutschland GmbH; 2021;67(6): 1581-1599.] The output relative moduli map from the virtual fields method served as an a priori stiffness input to the second inverse modeling technique, a biomechanical model -based iterative optimization approach (FIG. 18C). In this method, experimental displacements from DENSE imaging, ROI definitions, and moduli maps from the virtual fields method are input. A two-dimensional finite element mesh was generated based on ROI boundaries and DENSE sequence resolution (FIG. 18C). Smoothed displacement fields observed from DENSE imaging were then interpolated onto the finite element mesh for use in the inverse modeling method (FIG. 18C). Model-estimated displacement fields were calculated using the spatial moduli map from the virtual fields method and a linear elastic constitutive model, solved over the finite element mesh using the Galerkin method of weighted residuals. [Miller CE, et al., Journal of Medical Imaging. 2021.] An objective function, defined by the sum of squared error mismatch between the observed (from DENSE) and model-estimated displacement fields, was minimized using an L-BFGS quasi-Newton algorithm (FIG. 19C). This algorithm iteratively updates the elastic modulus estimates for each element in the finite element mesh until a preset convergence tolerance was met. [Weis A-C JA, et al., Development of a diaphragmatic motionbased elastography framework for assessment of liver stiffness; Weis JA, et al., Three- dimensional image-based mechanical modeling for predicting the response of breast cancer to neoadjuvant therapy. Comput Methods Appl Meeh Eng. North-Holland; 2017;314:494-512.] This entire elasticity imaging framework yields spatially resolved estimates of relative modulus within each cartilage ROI (FIG. 18C).

[0109] Validation of elasticity imaging framework: Five distinct finite element models were simulated in FEBio [Maas SA, et al., FEBio: Finite Elements for Biomechanics. J Biomech Eng. American Society of Mechanical Engineers; 2012;134(l): 11005.] to evaluate the feasibility of the elasticity imaging framework for cartilage moduli imaging in conjunction with the DENSE MRI sequence (FIG. 19A). These simulations represented varying levels of cartilage stiffness heterogeneity and geometries, with increasing complexity. All models assumed linear elastic cartilage material properties with depth-dependent moduli and a nearly incompressible Poisson ratio. The models, in order from least to most complex, included: a two-dimensional square with increasing moduli values under unconfined compression (FIG. 19B) [Mostakhdemin M, Nand A, Ramezani M. Articular and Artificial Cartilage, Characteristics, Properties and Testing Approaches — A Review. Polymers (Basel). MDPI AG; 2021 ; 13( 12) :2000]; the first model, with an additional soft inclusion at the top to simulate cartilage bulk softening (FIG. 19C) [Pritzker KPH, Gay S, Jimenez SA, et al. Osteoarthritis cartilage histopathology: Grading and staging. Osteoarthritis Cartilage. W.B. Saunders Ltd; 2006; 14(1): 13-29.]; the first model, but with a narrow soft inclusion to represent a cartilage fissure (FIG. 19D) [Blom RP, Rahim D, Paardekam E, Kerkhoffs GMMJ, lannuzzi D, Smit TH. A Traumatic Impact Immediately Changes the Mechanical Properties of Articular Cartilage. Cartilage. Cartilage; 2024]; a two-dimensional thin cartilage-like half annulus under unconfined compression (FIG. 19E) [Eckstein F, et al. Magnitude and regional distribution of cartilage loss associated with grades of joint space narrowing in radiographic osteoarthritis - data from the Osteoarthritis Initiative (OAI). Osteoarthritis Cartilage. Elsevier Ltd; 2010;18(6):760- 768] and a three-dimensional finite element knee model with three layers of depth dependent articular cartilage moduli, undergoing medial compartmental compression (FIG. 19F) [Paz A, Orozco GA, Tanska P, Garcia JJ, Korhonen RK, Mononen ME. A novel knee joint model in FEBio with inhomogeneous fibril-reinforced biphasic cartilage simulating tissue mechanical responses during gait: data from the osteoarthritis initiative. Comput Methods Biomech Biomed Engin. Taylor & Francis; 2023;26(l 1): 1353-1367], In this model, after the model was run in FEBio, MATLAB code was employed to extract two-dimensional xy displacements in the medial tibial region of interest within a predefined midplane coronal (three-dimensional) z-slice of the model. For validation, two-dimensional displacement fields were extracted from all models upon completion of simulation and used as input for the elasticity imaging framework. The resulting relative modulus maps were compared to the ground truth modulus maps from the finite element models to calculate an average voxel -wise percent error (FIG. 19B-F). To analyze error in the spatial reconstruction of moduli maps in response to noise, we applied Monte Carlo simulations to the least complex model (FIG. 19B, 19G). Within each iteration of the 100 Monte Carlo simulations, random Gaussian noise was added to the ideal displacements of all nodes (FIG. 19G). These noisy displacements were then input into the elasticity imaging framework. Additionally, to assess the impact of smoothing noisy displacements on stiffness reconstruction error, LOWESS smoothing was independently applied to the noisy displacements during Monte Carlo simulations (FIG. 19G). Relative moduli were then derived, and the average percent error compared to the known stiffness values was calculated.

[0110] Calculation of gradient modulus and gradient index: Motivated by the depth-wise, zonal structure of articular cartilage, we evaluated the spatial gradient of relative modulus perpendicular to the articular cartilage surface (i.e., through the thickness) in each ROI using a previously established technique [Wilson RL, et al., Spatial Gradients of Quantitative MRI as Biomarkers for Early Detection of Osteoarthritis: Data From Human Explants and the Osteoarthritis Initiative. Journal of Magnetic Resonance Imaging. John Wiley & Sons, Ltd; 2023 ;58(1): 189— 197], To achieve this, we fit a spline to the cartilage contact surface within each ROI and computed perpendicular trajectories through the cartilage accordingly. Along each trajectory we calculated the spatial gradient of modulus perpendicular to the cartilage surface via a numerical finite difference scheme, resulting in anatomically relevant spatial gradient maps (FIG. 19C) (40). To derive a single numeric metric representative of the full gradient map, we utilized absolute value cumulative histogram thresholding, a dynamic technique where values are cumulatively summed to determine the threshold at the greatest inflection point (FIG. 18C). This method effectively captured relatively high voxel values (-30% of voxels) on the VModulus maps while reducing potential human selection bias. To calculate a gradient index for the ROI, these selected values were then averaged. In addition to the gradient index, we also computed the variance of the gradient and the variance of the relative modulus within each ROI for statistical analysis. Separately, we conducted Monte Carlo simulations to assess the effect of modulus range on the gradient index. Since the elasticity imaging framework is constrained by type 1 displacement boundary conditions, the method can only quantify relative maps of elastic moduli and not absolute maps. All relative moduli maps are automatically normalized to a maximum value of 1. In the Monte Carlo simulations, we randomly generated 100 different modulus maps within a representative cartilage region of interest. Each map was scaled to four different modulus ranges: 0-100, 40-60, 0.5-2, and the original 0-1 range. The gradient index was then calculated for each scaled map. For each scaled map, we calculated the average percent error in the gradient index compared to the original 0-1 map, and this was averaged over the 100 simulations.

[0111] Statistical Analysis'. Statistical analyses were carried out using SAS Version 9.4 (SAS Institute) with a 2-sided significance level of o=0.05. Data that did not adhere to the assumption of normality, as determined by the Shapiro-Wilk test, underwent logarithmic transformation using standard methods. For each ROI, the variance of modulus, variance of modulus gradient, and gradient index were computed. To assess the influence of participant characteristics on each modulus metric and patient-reported outcomes, separate linear mixed-effects models were developed for each response variable. Fixed categorical effects included sex, surgeon, graft type, limb, and tibiofemoral location, while age and BMI served as covariates. Patient ID was treated as a random effect. The relationships between modulus quantitative measures and patient-reported outcomes for each cartilage region and limb were evaluated using general linear models [Williams AA, et al. Patient-Reported Outcomes and Knee Mechanics Correlate With Patellof emoral Deep Cartilage UTE-T2* 2 Years After Anterior Cruciate Ligament Reconstruction; Weis A-C JA, et al., Development of a diaphragmatic motion-based elastography framework for assessment of liver stiffness.]. Age was included as a covariate in these models based on the results of the linear mixed-effects models. Residuals from each model were examined to test for normality and homoscedasticity. Additionally, general linear models following the same protocols were utilized to explore correlations between gait metrics and modulus metrics in the subgroup that underwent gait analysis [Griebel AJ, Khoshgoftar M, Novak T, van Donkelaar CC, Neu CP. Direct noninvasive measurement and numerical modeling of depth-dependent strains in layered agarose constructs. J Biomech. Elsevier; 2014;47(9):2149-2156.] To facilitate comparisons between strain, quantitative MRI, and elastic modulus, general linear models were also used to explore correlations between these values.

[0112] GLOSSARY OF CLAIM TERMS

[0113] As used throughout the entire application, the terms “a” and “an” are used in the sense that they mean “at least one”, “at least a first”, “one or more” or “a plurality” of the referenced components or steps, unless the context clearly dictates otherwise. For example, the term “a cell” includes a plurality of cells, including mixtures thereof.

[0114] The term “and / or” whereever used herein includes the meaning of “and”, “or” and “all or any other combination of the elements connected by said term”.

[0115] The term “about” or “approximately” as used herein means within 20%, preferably within 10%, and more preferably within 5% of a given value or range.

[0116] Notwithstanding that the numerical ranges and parameters setting forth the broad scope of the disclosure are approximations, the numerical values set forth in the specific examples are reported as precisely as possible. Any numerical value, however, inherently contain certain errors necessarily resulting from the standard deviation found in their respective testing measurements. Furthermore, when numerical ranges of varying scope are set forth herein, it is contemplated that any combination of these values inclusive of the recited values may be used.

[0117] As used herein, the term “comprising” is intended to mean that the products, compositions and methods include the referenced components or steps, but not excluding others. “Consisting essentially of’ when used to define products, compositions and methods, shall mean excluding other components or steps of any essential significance. Thus, a composition consisting essentially of the recited components would not exclude trace contaminants and pharmaceutically acceptable carriers. “Consisting of’ shall mean excluding more than trace elements of other components or steps.

[0118] The advantages set forth above, and those made apparent from the foregoing description, are efficiently attained. Since certain changes may be made in the above construction without departing from the scope of the invention, it is intended that all matters contained in the foregoing description or shown in the accompanying drawings shall be interpreted as illustrative and not in a limiting sense.

[0119] All references cited in the present application are incorporated in their entirety herein by reference to the extent not inconsistent herewith.

[0120] It will be seen that the advantages set forth above, and those made apparent from the foregoing description, are efficiently attained and since certain changes may be made in the above construction without departing from the scope of the invention, it is intended that all matters contained in the foregoing description or shown in the accompanying drawings shall be interpreted as illustrative and not in a limiting sense.

[0121] It is also to be understood that the following claims are intended to cover all of the generic and specific features of the invention herein described, and all statements of the scope of the invention which, as a matter of language, might be said to fall therebetween. Now that the invention has been described,

[0122] Table 1. Part list of the thigh constraint. Table 2. Part list of the loading section. Table 3. Part list of the control unit.

[0123] Table 4. Parameters for qMRI sequences. Table 5: Participant demographic, surgical characteristics, and patient-reported outcomes within the 20 subjects included for all analyses and the gait analysis sub-cohort. aData are reported as mean ± SD, or n (%). All KOOS scores are on a 0-100 scale, with 0 representing extreme knee problems. All MARS Scores are scaled from 0 to 16 with 16 representing highly physically active individuals. BMI, body mass index. ACLR, anterior cruciate ligament reconstruction. KOOS, Knee Injury and Osteoarthritis Outcome Score. WOMAC. Western Ontario and McMaster Universities Osteoarthritis Index, MARS, Marx Activity Rating (MARS) Table 6: Error Analysis of Modulus from forward simulations aData are reported as mean ± SD. Percent error was calculated on a voxel wise basis between the relative modulus map output by the elasticity imaging framework and the ground truth modulus map as input into the finite element model. Percent error maps shown in Figure 2. Average percent error was calculated for the entire percent error map.

[0124] Table 7: The gradient index remains the same when moduli are scaled aData are reported as mean ± SD. Monte Carlo simulations were used to generate 100 different modulus maps within a representative cartilage region of interest. Each map was scaled to four different modulus ranges: 0-100, 40-60, 0.5-2, and the original 0-1 range. The gradient index was then calculated for each scaled map. For each scaled map, we calculated the average percent error in the gradient index compared to the original 0-1 map, and this was averaged over the 100 simulations. Table 8: Summary of significance from linear mixed-model estimates for associations between modulus quantitative metrics, patient-reported outcomes and patient factorsa. aData are presented as p values. Separate linear mixed effects models were created for each modulus response variable and patient-reported outcome score. Mixed models consist of the following terms: sex, age, surgeon, BMI, graft type, cartilage location, and limb. Blue boxes indicate significant differences at the level of p < 0.05. Grey boxes indicate significant differences between at the level of p < 0.10.

Claims

What is claimed is:

1. An elastography framework for musculoskeletal and connective tissues comprising: a mechanical loading device adapted to exert a low-frequency, high-force magnitude loading in the physiological range to a human body or a knee of a subject, wherein the loading device is further adapted to operate in the environment of an MRI scanner; a module in communication with the loading device and an MRI scanner having software to synchronize the actions of the mechanical loading device and data acquisition from the MRI scanner; and a module in communication with the MRI scanner having analysis software using inverse and constitutive modeling to provide measurements of relative or absolute stiffness or material properties in musculoskeletal and connective tissues of the human body or a knee of a subject from MRI data from the MRI scanner.

2. The elastography framework for musculoskeletal and connective tissues according to claim 1 wherein the framework is a system for evaluating or quantifying stiffness in joints and connective tissues susceptible to other degenerative arthropathies.

3. The framework according to claim 1 wherein the MRI scanner captures data from the musculoskeletal and connective tissues comprising tissue selected from the group consisting of cartilage, bone, meniscus, ligament, tendon, bone marrow, bursa, synovium, adipose, skin, muscle, and intervertebral discs.

4. The framework according to claim 3 wherein the musculoskeletal and connective tissues are selected form the group consisting of healthy, degenerative, injured, diseased, repaired, regenerated, and engineered tissues.

5. The framework according to claim 3 wherein the musculoskeletal and connective tissues are selected form the group consisting of young, juvenile, or aging or old tissues.

6. The framework according to claim 1 wherein the mechanical loading device comprises: a thigh or hip constraint section, adapted to fix the thigh to eliminate translational motion of the body; anda loading section, adapted to exert a linear or rotational force (varus-valgus, flexionextension, compression-tension, or internal-external) on a subject’s tibiofemoral joint, wherein the loading section results in multicompartmental loading to the medial and / or lateral knee compartment of tibiofemoral cartilage, and / or the patellofemoral cartilage.

7. The mechanical loading apparatus according to claim 6 wherein the thigh constraint section comprises a base and two pillar blocks, wherein the pillar block locations are adjustable to compensate for individual anatomical differences of patients and asymmetries between left or right legs.

8. The mechanical loading apparatus according to claim 7 wherein the pillar blocks are wrapped in thick silicon rubber to provide for the comfort of the patient.

9. The mechanical loading apparatus according to 6 wherein the loading section comprises: a linear or rotational motion system comprising a bearing system, a rear support, foot cast assembly to secure the foot of a subject, and pneumatic cylinder assembly connected to the foot cast assembly, wherein a pneumatic cylinder assembly applies a force to the foot cast assembly; and a main frame coupled to the linear or rotational motion system, adapted to attach the apparatus to an MRI machine and support the linear motion system.

10. The mechanical loading apparatus according to claim 9 further comprising a control unit.

11. The mechanical loading apparatus according to claim 10 wherein the control unit comprises an air pump, an air regulator, a power source, a data acquisition system, and a control circuit, wherein the control unit sends a signal to the air regulator thereby actuating the air pump which interfaces with a cylinder in the pneumatic cylinder assembly.

12. The framework according to claim 10 wherein the control unit synchronizes the mechanical loading apparatus to the pulse sequence actions of the MRI scanner for combined and integrated actions and load-dependent image acquisitions.

13. The framework according to claim 1 wherein the analysis software for inverse and constitutive modeling combines: a inverse algorithm (e.g., virtual fields, topology optimization);linear (e.g., Hookean), nonlinear (e.g., Neo-Hookean or Mooney-Rivlin), or multiphasic or poroelastic constitutive models to extract mechanical quantities like stiffness or modulus; and a finite element analysis to specify or estimate displacement or force boundary conditions to achieve relative or absolute measures of material properties.

14. The framework according to claim 1 wherein the framework provides a system that generates measurements of relative or absolute stiffness or material properties.

15. The system according to claim 14 wherein the measurements of relative or absolute stiffness or material properties is selected from the group consisting of elastic properties, tensile or compressive stiffness or modulus, shear stiffness or modulus, Young’s modulus, equilibrium modulus, viscoelastic constants (e.g., complex, storage, or loss modulus), multiphasic constants (e.g., fluid load fraction) and combinations thereof.

16. The framework according to claim 1 wherein measurements of relative or absolute stiffness or material properties include spatial distributions (e.g., spatial gradient) of relative or absolute stiffness or material properties.

17. The framework according to claim 1 wherein measurements of relative or absolute stiffness or material properties are acquired with measures of MRI morphology and relaxometry, including proton density -weighted double echo steady state (DESS) images, or T1-, T2-, T2*-, and Tip-weighted images.

18. The framework according to claim 1 wherein measurements of relative or absolute stiffness or material properties are acquired with measures of markerless human motion or gait.

19. A mechanical loading device comprising: a thigh or hip constraint section, adapted to fix the thigh to eliminate translational motion of the body; and a loading section, adapted to exert a linear or rotational force (varus-valgus, flexionextension, compression-tension, or internal-external) on a subject’s tibiofemoral joint, wherein the loading section results in multicompartmental loading to the medial and / or lateral knee compartment of tibiofemoral cartilage, and / or the patellofemoral cartilage.

20. The mechanical loading apparatus according to claim 19 wherein the thigh constraint section comprises a base and two pillar blocks, wherein the pillar block locations are adjustable to compensate for individual anatomical differences of patients and asymmetries between left or right legs.

21. The mechanical loading apparatus according to claim 20 wherein the pillar blocks are wrapped in thick silicon rubber to provide for the comfort of the patient.

22. The mechanical loading apparatus according to 19 wherein the loading section comprises: a linear or rotational motion system comprising a bearing system, a rear support, foot cast assembly to secure the foot of a subject, and pneumatic cylinder assembly connected to the foot cast assembly, wherein a pneumatic cylinder assembly applies a force to the foot cast assembly; and a main frame coupled to the linear or rotational motion system, adapted to attach the apparatus to an MRI machine and support the linear motion system.

23. The mechanical loading apparatus according to claim 22 further comprising a control unit.

24. The mechanical loading apparatus according to claim 23 wherein the control unit comprises an air pump, an air regulator, a power source, a data acquisition system, and a control circuit, wherein the control unit sends a signal to the air regulator thereby actuating the air pump which interfaces with a cylinder in the pneumatic cylinder assembly.

25. The mechanical loading apparatus according to claim 23 wherein the control unit synchronizes the mechanical loading apparatus to the pulse sequence actions of the MRI scanner for combined and integrated actions and load-dependent image acquisitions.

26. A method of predicting tissue degeneration or assessing tissue health in a subject comprising the steps of: a. applying a mechanical loading to a joint of the subject at a low-frequency and at high-force magnitude; b. collecting displacement encoding with stimulated echoes (DENSE) magnetic resonance (MR) images of a region of interest in the subject, wherein the MRimages include magnitude and phase and are acquired during the loading time frame; c. calculating displacements from phase data, wherein the displacement are optionally smoothed with a locally weighted scatterplot smoothing (LOWES S) filter; d. creating spatial maps of relative modulus from the displacement calculations for a region of interest (ROI); e. estimating the relative moduli of the tissue by applying inverse modeling, wherein the elasticity imaging framework yields spatially resolved estimates of relative modulus within tissue or region of interest (ROI).

27. The method according to claim 26 wherein the inverse modeling includes various degrees of complexity of the choice of material (constitutive; e.g., elastic, viscoelastic, multiphasic, etc.) law, dimensionality (two- or three-dimensions), or anisotropy (linear or nonlinear).

28. The method according to claim 26 wherein the mechanical loading is applied using a device according to one of claims 19-25.

29. The method according to claim 26 wherein the low-frequency is between 0.001 cycles / second to 100 cycles / second.

30. The method according to claim 26 wherein the high-force magnitude load is between 0.1 to 10 times body weight of the subject.

31. The method according to claim 26 wherein the low-frequency, high force mechanical load mimics a normal physiological activity, such as walking or running.

32. A method to identify cartilage stiffness gradients as a biomarker for osteoarthritis pathogenesis in a subject comprising the steps of: a. applying a mechanical loading to a joint of the subject at a low-frequency and at high-force magnitude; b. collecting displacement encoding with stimulated echoes (DENSE) magnetic resonance (MR) images of a region of interest (ROI) in the subject; c. generating a displacement map from the images of a region of interest (ROI) of the subject;d. estimating the relative moduli of the tissue by applying inverse modeling, wherein the elasticity imaging framework yields spatially resolved estimates of relative modulus within tissue or region of interest (ROI). e. producing cartilage stiffness gradients from the relative moduli estimate, wherein softening patterns in the cartilage are indicative of progression to osteoarthritis.

33. The method according to claim 32 wherein the region of interest is pre-conditioned to the applied load before DENSE acquisition to minimize cartilage viscoelasticity and motion artifacts.

34. The method according to claim 33 wherein the region of interest is pre-conditioned to the applied load for about 4 to about 12 minutes.

35. An MRI-based method that combines displacement encoded with stimulated echoes MRI with an iterative finite elements-based optimization approach to quantify relative cartilage stiffness maps.

36. A mechanical loading device comprising: a thigh or hip constraint section, adapted to fix the thigh to eliminate translational motion of the body of a subject; and a loading section, adapted to exert a linear or rotational force (varus-valgus, flexionextension, compression-tension, or internal-external) on a subject’s tibiofemoral joint, wherein the loading section results in multicompartmental loading to the medial and / or lateral knee compartment of tibiofemoral cartilage, and / or the patellofemoral cartilage.

Images