Multivariate analysis of cartilage degradation using the support vector machine algorithm

Quantitative MRI methods for the assessment of structure, composition, and function of musculoskeletal tissues in basic research and preclinical applications

Osteoarthritis (OA) is a disabling chronic disease involving the gradual degradation of joint structures causing pain and dysfunction. Magnetic resonance imaging (MRI) has been widely used as a non-invasive tool for assessing OA-related changes. While anatomical MRI is limited to the morphological assessment of the joint structures, quantitative MRI (qMRI) allows for the measurement of biophysical properties of the tissues at the molecular level. Quantitative MRI techniques have been employed to characterize tissues' structural integrity, biochemical content, and mechanical properties. Their applications extend to studying degenerative alterations, early OA detection, and evaluating therapeutic intervention. This article is a review of qMRI techniques for musculoskeletal tissue evaluation, with a particular emphasis on articular cartilage. The goal is to describe the underlying mechanism and primary limitations of the qMRI parameters, their association with the tissue physiological properties and their potential in detecting tissue degeneration leading to the development of OA with a primary focus on basic and preclinical research studies. Additionally, the review highlights some clinical applications of qMRI, discussing the role of texture-based radiomics and machine learning in advancing OA research.

Machine Learning Prediction of Collagen Fiber Orientation and Proteoglycan Content From Multiparametric Quantitative MRI in Articular Cartilage

BACKGROUND

Machine learning models trained with multiparametric quantitative MRIs (qMRIs) have the potential to provide valuable information about the structural composition of articular cartilage.

PURPOSE

To study the performance and feasibility of machine learning models combined with qMRIs for noninvasive assessment of collagen fiber orientation and proteoglycan content.

STUDY TYPE

Retrospective, animal model.

ANIMAL MODEL

An open-source single slice MRI dataset obtained from 20 samples of 10 Shetland ponies (seven with surgically induced cartilage lesions followed by treatment and three healthy controls) yielded to 1600 data points, including 10% for test and 90% for train validation.

FIELD STRENGTH/SEQUENCE

A 9.4 T MRI scanner/qMRI sequences: T₁ , T₂ , adiabatic T_1ρ and T_2ρ , continuous-wave T_1ρ and relaxation along a fictitious field (T_RAFF ) maps.

ASSESSMENT

Five machine learning regression models were developed: random forest (RF), support vector regression (SVR), gradient boosting (GB), multilayer perceptron (MLP), and Gaussian process regression (GPR). A nested cross-validation was used for performance evaluation. For reference, proteoglycan content and collagen fiber orientation were determined by quantitative histology from digital densitometry (DD) and polarized light microscopy (PLM), respectively.

STATISTICAL TESTS

Normality was tested using Shapiro-Wilk test, and association between predicted and measured values was evaluated using Spearman's Rho test. A P-value of 0.05 was considered as the limit of statistical significance.

RESULTS

Four out of the five models (RF, GB, MLP, and GPR) yielded high accuracy (R²  = 0.68-0.75 for PLM and 0.62-0.66 for DD), and strong significant correlations between the reference measurements and predicted cartilage matrix properties (Spearman's Rho = 0.72-0.88 for PLM and 0.61-0.83 for DD). GPR algorithm had the highest accuracy (R²  = 0.75 and 0.66) and lowest prediction-error (root mean squared [RMSE] = 1.34 and 2.55) for PLM and DD, respectively.

DATA CONCLUSION

Multiparametric qMRIs in combination with regression models can determine cartilage compositional and structural features, with higher accuracy for collagen fiber orientation than proteoglycan content.

EVIDENCE LEVEL

2 TECHNICAL EFFICACY: Stage 2.

In vivo multimodal imaging of hyaluronan-mediated inflammatory response in articular cartilage

OBJECTIVE

One driving factor in the progression to posttraumatic osteoarthritis (PTOA) is the perpetuation of the inflammatory response to injury into chronic inflammation. Molecular imaging offers many opportunities to complement the sensitivity of current imaging modalities with molecular specificity. The goal of this study was to develop and characterize agents to image hyaluronan (HA)-mediated inflammatory signaling.

DESIGN

We developed optical (Cy5.5-P15-1) and magnetic resonance contrast agents (Gd-DOTA-P15-1) based in a hyaluronan-binding peptide (P15-1) that has shown anti-inflammatory effects on human chondrocytes, and validated them in vitro and in vivo in two animal models of PTOA.

RESULTS

In vitro studies with a near infrared (NIR) Cy5.5-P15-1 imaging agent showed a fast and stable localization of Cy5.5-P15-1 on chondrocytes, but not in synovial cells. In vivo NIR showed significantly higher retention of imaging agent in PTOA knees between 12 and 72 h (n = 8, Cohen's d > 2 after 24 h). NIR fluorescence accumulation correlated with histologic severity in cartilage and meniscus (ρ between 0.37 and 0.57, P < 0.001). By using in vivo magnetic resonance imaging with a Gd-DOTA-P15-1 contrast agent in 12 rats, we detected a significant decrease of T1 on injured knees in all cartilage plates at 48 h (-15%, 95%-confidence interval (CI) = [-18%,-11%]) while no change was observed in the controls (-2%, 95%-CI = [-5%,+1%]).

CONCLUSIONS

This study provides the first in vivo evidence that hyaluronan-related inflammatory response in cartilage after injury is a common finding. Beyond P15-1, we have demonstrated that molecular imaging can provide a versatile technology to investigate and phenotype PTOA pathogenesis, as well as study therapeutic interventions.

A Comparative Systematic Literature Review on Knee Bone Reports from MRI, X-rays and CT Scans Using Deep Learning and Machine Learning Methodologies

The purpose of this research was to provide a “systematic literature review” of knee bone reports that are obtained by MRI, CT scans, and X-rays by using deep learning and machine learning techniques by comparing different approaches—to perform a comprehensive study on the deep learning and machine learning methodologies to diagnose knee bone diseases by detecting symptoms from X-ray, CT scan, and MRI images. This study will help those researchers who want to conduct research in the knee bone field. A comparative systematic literature review was conducted for the accomplishment of our work. A total of 32 papers were reviewed in this research. Six papers consist of X-rays of knee bone with deep learning methodologies, five papers cover the MRI of knee bone using deep learning approaches, and another five papers cover CT scans of knee bone with deep learning techniques. Another 16 papers cover the machine learning techniques for evaluating CT scans, X-rays, and MRIs of knee bone. This research compares the deep learning methodologies for CT scan, MRI, and X-ray reports on knee bone, comparing the accuracy of each technique, which can be used for future development. In the future, this research will be enhanced by comparing X-ray, CT-scan, and MRI reports of knee bone with information retrieval and big data techniques. The results show that deep learning techniques are best for X-ray, MRI, and CT scan images of the knee bone to diagnose diseases.

Non-destructive detection of matrix stabilization correlates with enhanced mechanical properties of self-assembled articular cartilage

Tissue engineers rely on expensive, time-consuming, and destructive techniques to monitor the composition, microstructure, and function of engineered tissue equivalents. A non-destructive solution to monitor tissue quality and maturation would greatly reduce costs and accelerate the development of tissue-engineered products. The objectives of this study were to (a) determine whether matrix stabilization with exogenous lysyl oxidase-like protein-2 (LOXL2) with recombinant hyaluronan and proteoglycan link protein-1 (LINK) would result in increased compressive and tensile properties in self-assembled articular cartilage constructs, (b) evaluate whether label-free, non-destructive fluorescence lifetime imaging (FLIm) could be used to infer changes in both biochemical composition and biomechanical properties, (c) form quantitative relationships between destructive and non-destructive measurements to determine whether the strength of these correlations is sufficient to replace destructive testing methods, and (d) determine whether support vector machine (SVM) learning can predict LOXL2-induced collagen crosslinking. The combination of exogenous LOXL2 and LINK proteins created a synergistic 4.9-fold increase in collagen crosslinking density and an 8.3-fold increase in tensile strength as compared with control (CTL). Compressive relaxation modulus was increased 5.9-fold with addition of LOXL2 and 3.4-fold with combined treatments over CTL. FLIm parameters had strong and significant correlations with tensile properties (R²  = 0.82; p < 0.001) and compressive properties (R²  = 0.59; p < 0.001). SVM learning based on FLIm-derived parameters was capable of automating tissue maturation assessment with a discriminant ability of 98.4%. These results showed marked improvements in mechanical properties with matrix stabilization and suggest that FLIm-based tools have great potential for the non-destructive assessment of tissue-engineered cartilage.

Predicting early symptomatic osteoarthritis in the human knee using machine learning classification of magnetic resonance images from the osteoarthritis initiative

The purpose of this study is to evaluate the ability of a machine learning algorithm to classify in vivo magnetic resonance images (MRI) of human articular cartilage for development of osteoarthritis (OA). Sixty-eight subjects were selected from the osteoarthritis initiative (OAI) control and incidence cohorts. Progression to clinical OA was defined by the development of symptoms as quantified by the Western Ontario and McMaster Universities Arthritis (WOMAC) questionnaire 3 years after baseline evaluation. Multi-slice T2 -weighted knee images, obtained through the OAI, of these subjects were registered using a nonlinear image registration algorithm. T2 maps of cartilage from the central weight bearing slices of the medial femoral condyle were derived from the registered images using the multiple available echo times and were classified for "progression to symptomatic OA" using the machine learning tool, weighted neighbor distance using compound hierarchy of algorithms representing morphology (WND-CHRM). WND-CHRM classified the isolated T2 maps for the progression to symptomatic OA with 75% accuracy.

CLINICAL SIGNIFICANCE

Machine learning algorithms applied to T2 maps have the potential to provide important prognostic information for the development of OA. © 2017 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 35:2243-2250, 2017.

MR Parametric Mapping as a Biomarker of Early Joint Degeneration

Context:

Osteoarthritis (OA) is a common, worldwide disorder. Magnetic resonance (MR) imaging can directly and noninvasively evaluate articular cartilage and has emerged as an essential tool in the study of OA.

Evidence Acquisition:

A PubMed search was performed using the keywords quantitative MRI and cartilage. No limits were set on the range of years searched. Articles were reviewed for relevance with an emphasis on in vivo studies performed at 3 tesla.

Study Design:

Clinical review.

Level of Evidence:

Level 4.

Results:

T2, T2*, T1 (particularly when measured after exogenous contrast administration, such as with the delayed gadolinium-enhanced MR imaging of cartilage [dGEMRIC] technique), and T1ρ are among the most widely utilized quantitative MR imaging techniques to evaluate cartilage and have been implemented in various patient cohorts. Existing challenges include reproducibility of results, insufficient consensus regarding optimal sequences and parameters, and interpretation of values.

Conclusion:

Quantitative assessment of cartilage using MR imaging techniques likely represents the best opportunity to identify early cartilage degeneration and to follow patients after treatment. Despite existing challenges, ongoing work and unique approaches have shown exciting and promising results.

Multiparametric Classification of Skin from Osteogenesis Imperfecta Patients and Controls by Quantitative Magnetic Resonance Microimaging

The purpose of this study is to evaluate the ability of quantitative magnetic resonance imaging (MRI) to discriminate between skin biopsies from individuals with osteogenesis imperfecta (OI) and skin biopsies from individuals without OI. Skin biopsies from nine controls (unaffected) and nine OI patients were imaged to generate maps of five separate MR parameters, T₁, T₂, k_m, MTR and ADC. Parameter values were calculated over the dermal region and used for univariate and multiparametric classification analysis. A substantial degree of overlap of individual MR parameters was observed between control and OI groups, which limited the sensitivity and specificity of univariate classification. Classification accuracies ranging between 39% and 67% were found depending on the variable of investigation, with T₂ yielding the best accuracy of 67%. When several MR parameters were considered simultaneously in a multivariate analysis, the classification accuracies improved up to 89% for specific combinations, including the combination of T₂ and k_m. These results indicate that multiparametric classification by quantitative MRI is able to detect differences between the skin of OI patients and of unaffected individuals, which motivates further study of quantitative MRI for the clinical diagnosis of OI.

Machine learning classification of OARSI-scored human articular cartilage using magnetic resonance imaging

OBJECTIVE

The purpose of this study is to evaluate the ability of machine learning to discriminate between magnetic resonance images (MRI) of normal and pathological human articular cartilage obtained under standard clinical conditions.

METHOD

An approach to MRI classification of cartilage degradation is proposed using pattern recognition and multivariable regression in which image features from MRIs of histologically scored human articular cartilage plugs were computed using weighted neighbor distance using compound hierarchy of algorithms representing morphology (WND-CHRM). The WND-CHRM method was first applied to several clinically available MRI scan types to perform binary classification of normal and osteoarthritic osteochondral plugs based on the Osteoarthritis Research Society International (OARSI) histological system. In addition, the image features computed from WND-CHRM were used to develop a multiple linear least-squares regression model for classification and prediction of an OARSI score for each cartilage plug.

RESULTS

The binary classification of normal and osteoarthritic plugs yielded results of limited quality with accuracies between 36% and 70%. However, multiple linear least-squares regression successfully predicted OARSI scores and classified plugs with accuracies as high as 86%. The present results improve upon the previously-reported accuracy of classification using average MRI signal intensities and parameter values.

CONCLUSION

MRI features detected by WND-CHRM reflect cartilage degradation status as assessed by OARSI histologic grading. WND-CHRM is therefore of potential use in the clinical detection and grading of osteoarthritis.

Analysis of mcDESPOT- and CPMG-derived parameter estimates for two-component nonexchanging systems

PURPOSE

To compare the reliability and stability of the multicomponent-driven equilibrium single pulse observation of T1 and T2 (mcDESPOT) and Carl-Purcell-Meiboom-Gill (CPMG) approaches to parameter estimation.

METHODS

The stability and reliability of mcDESPOT and CPMG-derived parameter estimates were compared through examination of energy surfaces, evaluation of model sloppiness, and Monte Carlo simulations. Comparisons were performed on an equal time basis and assuming a two-component system. Parameter estimation bias, reflecting accuracy, and dispersion, reflecting precision, were derived for a range of signal-to-noise ratios (SNRs) and relaxation parameters.

RESULTS

The energy surfaces for parameters incorporated into the mcDESPOT signal model exhibit flatness, a complex structure of local minima, and instability to noise to a much greater extent than the corresponding surfaces for CPMG. Although both mcDESPOT and CPMG performed well at high SNR, the CPMG approach yielded parameter estimates of considerably greater accuracy and precision at lower SNR.

CONCLUSION

mcDESPOT and CPMG both permit high-quality parameter estimates under SNR that are clinically achievable under many circumstances, depending upon available hardware and resolution and acquisition time constraints. At moderate to high SNR, the mcDESPOT approach incorporating two-step phase increments can yield accurate parameter estimates while providing values for longitudinal relaxation times that are not available through CPMG. However, at low SNR, the CPMG approach is more stable and provides superior parameter estimates. Magn Reson Med 75:2406-2420, 2016. © 2015 Wiley Periodicals, Inc.

Techniques and applications of in vivo diffusion imaging of articular cartilage

Early in the process of osteoarthritis (OA) the composition (water, proteoglycan [PG], and collagen) and structure of articular cartilage is altered leading to changes in its mechanical properties. A technique that can assess the composition and structure of the cartilage in vivo can provide insight in the mechanical integrity of articular cartilage and become a powerful tool for the early diagnosis of OA. Diffusion tensor imaging (DTI) has been proposed as a biomarker for cartilage composition and structure. DTI is sensitive to the PG content through the mean diffusivity and to the collagen architecture through the fractional anisotropy. However, the acquisition of DTI of articular cartilage in vivo is challenging due to the short T2 of articular cartilage (∼40 ms at 3 Tesla) and the high resolution needed (0.5-0.7 mm in plane) to depict the cartilage anatomy. We describe the pulse sequences used for in vivo DTI of articular cartilage and discus general strategies for protocol optimization. We provide a comprehensive review of measurements of DTI of articular cartilage from ex vivo validation experiments to its recent clinical applications.

Classification of sodium MRI data of cartilage using machine learning

PURPOSE

To assess the possible utility of machine learning for classifying subjects with and subjects without osteoarthritis using sodium magnetic resonance imaging data. Theory: Support vector machine, k-nearest neighbors, naïve Bayes, discriminant analysis, linear regression, logistic regression, neural networks, decision tree, and tree bagging were tested.

METHODS

Sodium magnetic resonance imaging with and without fluid suppression by inversion recovery was acquired on the knee cartilage of 19 controls and 28 osteoarthritis patients. Sodium concentrations were measured in regions of interests in the knee for both acquisitions. Mean (MEAN) and standard deviation (STD) of these concentrations were measured in each regions of interest, and the minimum, maximum, and mean of these two measurements were calculated over all regions of interests for each subject. The resulting 12 variables per subject were used as predictors for classification.

RESULTS

Either Min [STD] alone, or in combination with Mean [MEAN] or Min [MEAN], all from fluid suppressed data, were the best predictors with an accuracy >74%, mainly with linear logistic regression and linear support vector machine. Other good classifiers include discriminant analysis, linear regression, and naïve Bayes.

CONCLUSION

Machine learning is a promising technique for classifying osteoarthritis patients and controls from sodium magnetic resonance imaging data.

Sensitivity and specificity of univariate MRI analysis of experimentally degraded cartilage under clinical imaging conditions

BACKGROUND

To evaluate the sensitivity and specificity of classification of pathomimetically degraded bovine nasal cartilage at 3 Tesla and 37°C using univariate MRI measurements of both pure parameter values and intensities of parameter-weighted images.

METHODS

Pre- and posttrypsin degradation values of T1 , T2 , T2 *, magnetization transfer ratio (MTR), and apparent diffusion coefficient (ADC), and corresponding weighted images, were analyzed. Classification based on the Euclidean distance was performed and the quality of classification was assessed through sensitivity, specificity and accuracy (ACC).

RESULTS

The classifiers with the highest accuracy values were ADC (ACC = 0.82 ± 0.06), MTR (ACC = 0.78 ± 0.06), T1 (ACC = 0.99 ± 0.01), T2 derived from a three-dimensional (3D) spin-echo sequence (ACC = 0.74 ± 0.05), and T2 derived from a 2D spin-echo sequence (ACC = 0.77 ± 0.06), along with two of the diffusion-weighted signal intensities (b = 333 s/mm(2) : ACC = 0.80 ± 0.05; b = 666 s/mm(2) : ACC = 0.85 ± 0.04). In particular, T1 values differed substantially between the groups, resulting in atypically high classification accuracy. The second-best classifier, diffusion weighting with b = 666 s/mm(2) , as well as all other parameters evaluated, exhibited substantial overlap between pre- and postdegradation groups, resulting in decreased accuracies.

CONCLUSION

Classification according to T1 values showed excellent test characteristics (ACC = 0.99), with several other parameters also showing reasonable performance (ACC > 0.70). Of these, diffusion weighting is particularly promising as a potentially practical clinical modality. As in previous work, we again find that highly statistically significant group mean differences do not necessarily translate into accurate clinical classification rules.

Prediction of cartilage compressive modulus using multiexponential analysis of T(2) relaxation data and support vector regression

Evaluation of mechanical characteristics of cartilage by magnetic resonance imaging would provide a noninvasive measure of tissue quality both for tissue engineering and when monitoring clinical response to therapeutic interventions for cartilage degradation. We use results from multiexponential transverse relaxation analysis to predict equilibrium and dynamic stiffness of control and degraded bovine nasal cartilage, a biochemical model for articular cartilage. Sulfated glycosaminoglycan concentration/wet weight (ww) and equilibrium and dynamic stiffness decreased with degradation from 103.6 ± 37.0 µg/mg ww, 1.71 ± 1.10 MPa and 15.3 ± 6.7 MPa in controls to 8.25 ± 2.4 µg/mg ww, 0.015 ± 0.006 MPa and 0.89 ± 0.25MPa, respectively, in severely degraded explants. Magnetic resonance measurements were performed on cartilage explants at 4 °C in a 9.4 T wide-bore NMR spectrometer using a Carr-Purcell-Meiboom-Gill sequence. Multiexponential T2 analysis revealed four water compartments with T2 values of approximately 0.14, 3, 40 and 150 ms, with corresponding weight fractions of approximately 3, 2, 4 and 91%. Correlations between weight fractions and stiffness based on conventional univariate and multiple linear regressions exhibited a maximum r(2) of 0.65, while those based on support vector regression (SVR) had a maximum r(2) value of 0.90. These results indicate that (i) compartment weight fractions derived from multiexponential analysis reflect cartilage stiffness and (ii) SVR-based multivariate regression exhibits greatly improved accuracy in predicting mechanical properties as compared with conventional regression.

Osteoarthritis year 2013 in review: imaging

PURPOSE

To review recent original research publications related to imaging of osteoarthritis (OA) and identify emerging trends and significant advances.

METHODS

Relevant articles were identified through a search of the PubMed database using the query terms "OA" in combination with "imaging", "radiography", "MRI", "ultrasound", "computed tomography", and "nuclear medicine"; either published or in press between March 2012 and March 2013. Abstracts were reviewed to exclude review articles, case reports, and studies not focused on imaging using routine clinical imaging measures.

RESULTS

Initial query yielded 932 references, which were reduced to 328 citations following the initial review. MRI (118 references) and radiography (129 refs) remain the primary imaging modalities in OA studies, with fewer reports using computed tomography (CT) (35 refs) and ultrasound (23 refs). MRI parametric mapping techniques remain an active research area (33 refs) with growth in T2*- and T1-rho mapping publications compared to prior years. Although the knee is the major joint studied (210 refs) there is interest in the hip (106 refs) and hand (29 refs). Imaging continues to focus on evaluation of cartilage (173 refs) and bone (119 refs).

CONCLUSION

Imaging plays a major role in OA research with publications continuing along traditional lines of investigation. Translational and clinical research application of compositional MRI techniques is becoming more common driven in part by the availability of T2 mapping data from the Osteoarthritis Initiative (OAI). New imaging techniques continue to be developed with a goal of identifying methods with greater specificity and responsiveness to changes in the joint, and novel functional neuroimaging techniques to study central pain. Publications related to imaging of OA continue to be heavily focused on quantitative and semiquantitative MRI evaluation of the knee with increasing application of compositional MRI techniques in the hip.

Near infrared spectroscopic evaluation of water in hyaline cartilage

In diseased conditions of cartilage such as osteoarthritis, there is typically an increase in water content from the average normal of 60-85% to greater than 90%. As cartilage has very little capability for self-repair, methods of early detection of degeneration are required, and assessment of water could prove to be a useful diagnostic method. Current assessment methods are either destructive, time consuming, or have limited sensitivity. Here, we investigated the hypotheses that non-destructive near infrared spectroscopy (NIRS) of articular cartilage can be used to differentiate between free and bound water, and to quantitatively assess water content. The absorbances centered at 5200 and 6890 cm(-1) were attributed to a combination of free and bound water, and to free water only, respectively. The integrated areas of both absorbance bands were found to correlate linearly with the absolute water content (R = 0.87 and 0.86) and with percent water content (R = 0.97 and 0.96) of the tissue. Partial least square models were also successfully developed and were used to predict water content, and percent free water. These data demonstrate that NIRS can be utilized to quantitatively determine water content in articular cartilage, and may aid in early detection of degenerative tissue changes in a laboratory setting, and with additional validations, possibly in a clinical setting.

Probing articular cartilage damage and disease by quantitative magnetic resonance imaging

Osteoarthritis (OA) is a debilitating disease that reflects a complex interplay of biochemical, biomechanical, metabolic and genetic factors, which are often triggered by injury, and mediated by inflammation, catabolic cytokines and enzymes. An unmet clinical need is the lack of reliable methods that are able to probe the pathogenesis of early OA when disease-rectifying therapies may be most effective. Non-invasive quantitative magnetic resonance imaging (qMRI) techniques have shown potential for characterizing the structural, biochemical and mechanical changes that occur with cartilage degeneration. In this paper, we review the background in articular cartilage and OA as it pertains to conventional MRI and qMRI techniques. We then discuss how conventional MRI and qMRI techniques are used in clinical and research environments to evaluate biochemical and mechanical changes associated with degeneration. Some qMRI techniques allow for the use of relaxometry values as indirect biomarkers for cartilage components. Direct characterization of mechanical behaviour of cartilage is possible via other specialized qMRI techniques. The combination of these qMRI techniques has the potential to fully characterize the biochemical and biomechanical states that represent the initial changes associated with cartilage degeneration. Additionally, knowledge of in vivo cartilage biochemistry and mechanical behaviour in healthy subjects and across a spectrum of osteoarthritic patients could lead to improvements in the detection, management and treatment of OA.

Diffusion-tensor imaging of human articular cartilage specimens with early signs of cartilage damage

PURPOSE

To assess the use of diffusion-tensor (DT) imaging of articular cartilage to detect and grade early cartilage damage in human specimens with early signs of cartilage damage.

MATERIALS AND METHODS

This study was approved by the institutional review board. Forty-three cartilage-on-bone samples drilled from 21 human patellae were examined with 17.6-T magnetic resonance (MR) imaging and a diffusion-weighted spin-echo sequence (spatial resolution, 50 × 100 × 800 μm). Subsequently, samples underwent histologic analysis with safranin O staining. Cartilage damage on safranin O histologic slides was quantified with Osteoarthritis Research Society International (OARSI) grades; grades ranged from 0 (healthy) to 6 (bone remodeling). Maps of longitudinal diffusivity (λ(l)), transverse diffusivity (λ(t)), mean diffusivity (MD), and fractional anisotropy (FA) were calculated. Cartilage was segmented, and region of interest (ROI) analysis was performed and compared with histologic findings. Significant differences in MR parameters between the OARSI groups were assessed with the Tukey test. The value of DT imaging in the diagnosis and grading of cartilage damage was assessed with logistic regression analysis.

RESULTS

Samples had OARSI grades of 0 (n = 14), 1 (n = 11), 2 (n = 12), 3 (n = 4), and 4 (n = 2). Samples with an OARSI grade greater than 0 had significantly increased λ(l), λ(t), and MD (7%-25% increase) in the superficial cartilage growing deeper into cartilage with increasing OARSI grade. Samples with an OARSI grade greater than 0 showed significantly decreased FA in the deep cartilage (-25% to -35% decrease), suggesting that changes in the collagen architecture may occur early in cartilage degradation. DTI showed excellent performance in the detection of cartilage damage (accuracy, 0.95; 41 of 43 samples) and good performance in the grading of cartilage damage (accuracy, 0.74; 32 of 43 samples).

CONCLUSION

DT imaging of articular cartilage can enable physicians to detect and grade early cartilage damage.

Characterization of engineered cartilage constructs using multiexponential T₂ relaxation analysis and support vector regression

Increased sensitivity in the characterization of cartilage matrix status by magnetic resonance (MR) imaging, through the identification of surrogate markers for tissue quality, would be of great use in the noninvasive evaluation of engineered cartilage. Recent advances in MR evaluation of cartilage include multiexponential and multiparametric analysis, which we now extend to engineered cartilage. We studied constructs which developed from chondrocytes seeded in collagen hydrogels. MR measurements of transverse relaxation times were performed on samples after 1, 2, 3, and 4 weeks of development. Corresponding biochemical measurements of sulfated glycosaminoglycan (sGAG) were also performed. sGAG per wet weight increased from 7.74±1.34 μg/mg in week 1 to 21.06±4.14 μg/mg in week 4. Using multiexponential T₂ analysis, we detected at least three distinct water compartments, with T₂ values and weight fractions of (45 ms, 3%), (200 ms, 4%), and (500 ms, 97%), respectively. These values are consistent with known properties of engineered cartilage and previous studies of native cartilage. Correlations between sGAG and MR measurements were examined using conventional univariate analysis with T₂ data from monoexponential fits with individual multiexponential compartment fractions and sums of these fractions, through multiple linear regression based on linear combinations of fractions, and, finally, with multivariate analysis using the support vector regression (SVR) formalism. The phenomenological relationship between T₂ from monoexponential fitting and sGAG exhibited a correlation coefficient of r²=0.56, comparable to the more physically motivated correlations between individual fractions or sums of fractions and sGAG; the correlation based on the sum of the two proteoglycan-associated fractions was r²=0.58. Correlations between measured sGAG and those calculated using standard linear regression were more modest, with r² in the range 0.43-0.54. However, correlations using SVR exhibited r² values in the range 0.68-0.93. These results indicate that the SVR-based multivariate approach was able to determine tissue sGAG with substantially higher accuracy than conventional monoexponential T₂ measurements or conventional regression modeling based on water fractions. This combined technique, in which the results of multiexponential analysis are examined with multivariate statistical techniques, holds the potential to greatly improve the accuracy of cartilage matrix characterization in engineered constructs using noninvasive MR data.