The KNee OsteoArthritis Prediction (KNOAP2020) challenge: An image analysis challenge to predict incident symptomatic radiographic knee osteoarthritis from MRI and X-ray images.

OBJECTIVES: The KNee OsteoArthritis Prediction (KNOAP2020) challenge was organized to objectively compare methods for the prediction of incident symptomatic radiographic knee osteoarthritis within 78 months on a test set with blinded ground truth. DESIGN: The challenge participants were free to use any available data sources to train their models. A test set of 423 knees from the Prevention of Knee Osteoarthritis in Overweight Females (PROOF) study consisting of magnetic resonance imaging (MRI) and X-ray image data along with clinical risk factors at baseline was made available to all challenge participants. The ground truth outcomes, i.e., which knees developed incident symptomatic radiographic knee osteoarthritis (according to the combined ACR criteria) within 78 months, were not provided to the participants. To assess the performance of the submitted models, we used the area under the receiver operating characteristic curve (ROCAUC) and balanced accuracy (BACC). RESULTS: Seven teams submitted 23 entries in total. A majority of the algorithms were trained on data from the Osteoarthritis Initiative. The model with the highest ROCAUC (0.64 (95% confidence interval (CI): 0.57-0.70)) used deep learning to extract information from X-ray images combined with clinical variables. The model with the highest BACC (0.59 (95% CI: 0.52-0.65)) ensembled three different models that used automatically extracted X-ray and MRI features along with clinical variables. CONCLUSION: The KNOAP2020 challenge established a benchmark for predicting incident symptomatic radiographic knee osteoarthritis. Accurate prediction of incident symptomatic radiographic knee osteoarthritis is a complex and still unsolved problem requiring additional investigation.

Diagnosing osteoarthritis from T2 maps using deep learning: an analysis of the entire Osteoarthritis Initiative baseline cohort.

OBJECTIVE: We aim to study to what extent conventional and deep-learning-based T2 relaxometry patterns are able to distinguish between knees with and without radiographic osteoarthritis (OA). METHODS: T2 relaxation time maps were analyzed for 4,384 subjects from the baseline Osteoarthritis Initiative (OAI) Dataset. Voxel Based Relaxometry (VBR) was used for automatic quantification and voxel-based analysis of the differences in T2 between subjects with and without radiographic OA. A Densely Connected Convolutional Neural Network (DenseNet) was trained to diagnose OA from T2 data. For comparison, more classical feature extraction techniques and shallow classifiers were used to benchmark the performance of our algorithm's results. Deep and shallow models were evaluated with and without the inclusion of risk factors. Sensitivity and Specificity values and McNemar test were used to compare the performance of the different classifiers. RESULTS: The best shallow model was obtained when the first ten Principal Components, demographics and pain score were included as features (AUC = 77.77%, Sensitivity = 67.01%, Specificity = 71.79%). In comparison, DenseNet trained on raw T2 data obtained AUC = 83.44%, Sensitivity = 76.99%, Specificity = 77.94%. McNemar test on two misclassified proportions form the shallow and deep model showed that the boost in performance was statistically significant (McNemar's chi-squared = 10.33, degree of freedom (DF) = 1, P-value = 0.0013). CONCLUSION: In this study, we presented a Magnetic Resonance Imaging (MRI)-based data-driven platform using T2 measurements to characterize radiographic OA. Our results showed that feature learning from T2 maps has potential in uncovering information that can potentially better diagnose OA than simple averages or linear patterns decomposition.

3D bone-shape changes and their correlations with cartilage T1ρ and T2 relaxation times and patient-reported outcomes over 3-years after ACL reconstruction.

PURPOSE: (1) To identify bone-shape changes from baseline to 3-years after anterior cruciate ligament reconstruction (ACLR). (2) to assess association between changes in bone-shape from baseline to 6-months and changes in cartilage matrix and patient functions and symptoms from baseline to 3-years after ACLR. METHODS: Bilateral knees of 30 patients with unilateral ACL injuries were scanned at baseline, 6-months, 1-, 2-, and 3-years after ACLR. Bilateral knees of 13 controls were scanned at baseline, 1- and 3-years. Mean T1ρ and T2 values of each cartilage compartment were computed. Bone shape was quantified using statistical shape modeling (SSM) and 3D-MRI. Patient functions and symptoms were evaluated using Knee Injury and Osteoarthritis Outcome Score (KOOS). RESULTS: Statistically significant changes were observed in Femur 2 (medial femoral condyle [MF] shape), Femur 6 (intercondylar notch width), Tibia 1 (tibia plateau area), and Tibia 7 (medial tibia slope) over 3-years after ACLR. Statistically significant differences were observed between injured and control knees in several modes. Statistically significant correlations were found between changes in bone shape (ΔFemur 6, ΔFemur 8 [trochlea inclination and MF height], ΔTibia 1) from baseline to 6-months and that of cartilage T1ρ and T2 and KOOS from baseline to 3-years after ACLR. CONCLUSION: Bone shape remodeling occurs after ACLR, and early bone shape changes (within 6 months) correlated with cartilage matrix and patient outcomes at 3-years after ACLR. Bone shape can be a promising imaging biomarker that stratifies patients at high risk for post-traumatic osteoarthritis (PTOA).

Longitudinal changes in MR T1ρ/T2 signal of meniscus and its association with cartilage T1p/T2 in ACL-injured patients.

OBJECTIVE: To evaluate the longitudinal changes in meniscal T1ρ/T2 signal post-reconstruction in patients with acute anterior cruciate ligament (ACL) injury and to investigate the association with T1ρ/T2 signal in articular knee cartilage. METHOD: In this prospective study, knees of 37 patients with ACL-injury and reconstruction in addition to 13 healthy controls were scanned using magnetic resonance imaging (MRI) T1ρ/T2 mapping. Quantitative analysis of the meniscus was performed in the anterior/posterior horns of lateral/medial meniscus fourteen sub-compartments of cartilage spanning the medial/lateral area of the tibia and femoral condyles. Meniscus T1ρ/T2 signals were compared between injured, contralateral and control knees at baseline, 6-months, 1-year and 2-years using t-tests for cross-sectional comparisons and a mixed model for longitudinal comparisons. Pearson-partial correlations between meniscal and cartilage T1ρ/T2 were evaluated. RESULTS: There was a significant decrease of T1ρ/T2 signal in the posterior horn of lateral meniscus (PHLAT) of injured knees during a 2-year period. In the posterior horn of medial meniscus (PHMED), T1ρ/T2 signal of injured knees was significantly elevated at all time points post-reconstruction compared to contralateral and control knees. Within injured knees, PHMED T1ρ/T2 signal showed significant positive correlations with medial tibia (MT) cartilage T1ρ/T2 signal at all time points. CONCLUSION: A significant decrease in PHLAT T1ρ/T2 signal by 2-years suggests potential tissue recovery after ACL-injury. Elevated T1ρ/T2 signal in the PHMED of injured knees at 2-years correlating with knee cartilage T1ρ/T2 signal elevations suggests involvement of the PHMED in subacute cartilage degeneration after ACL-injury and reconstruction.

Cross-sectional and longitudinal study of the impact of posterior meniscus horn lesions on adjacent cartilage composition, patient-reported outcomes and gait biomechanics in subjects without radiographic osteoarthritis.

OBJECTIVE: The aim of this study was to assess cross-sectional and longitudinal effects of meniscal lesions on adjacent cartilage T1ρ and T2 relaxation times, patient-reported outcomes and gait biomechanics. DESIGN: Thirty patients with no cartilage morphological defects reported by Whole Organ MRI Score (WORMS) magnetic resonance imaging (MRI) grading and no radiographic osteoarthritis (OA) (Kellgren--Lawrence (KL) ≤ 1) were selected, 15 with posterior meniscus horn lesions and 15 matched controls without meniscal lesions. All were imaged on a 3T MR scanner for three consecutive years, except those who dropped from the study. Sagittal and frontal plane kinematic gait data were acquired at baseline. The Knee Injury and Osteoarthritis Outcome Score (KOOS) survey was taken each time. All images were automatically segmented and registered to an atlas for voxel-by-voxel cross-sectional and longitudinal analyses. RESULTS: Relaxation time comparisons between groups showed elevated T1ρ of the lateral tibia (LP) and elevated T2 of the medial tibia (MT) and LT at 1 and 2 years in the lesion group. Longitudinal comparisons within each group revealed greater relaxation time elevations over one and 2 years in the group with lesions. KOOS Quality of Life (QOL) was significantly different between the groups at all time points (P < 0.05), as were other KOOS subcategories. No significant differences in the frontal or sagittal biomechanics were observed between the groups at baseline. CONCLUSIONS: Individuals with healthy cartilage and posterior meniscal horn lesions have increased relaxation times when compared to matched controls, increased relaxation time changes over 2 years, and consistently report a lower KOOS QOL, yet show no difference in gait biomechanics.

T1ρ and T2 relaxation times are associated with progression of hip osteoarthritis.

OBJECTIVE: To evaluate whether baseline T1ρ and T2 relaxation times of hip cartilage are associated with magnetic resonance imaging (MRI) based progression of hip osteoarthritis (OA) at 18 months. METHODS: 3T MRI studies of the hip were obtained at baseline and 18-month follow-up for 54 subjects without evidence of severe OA at baseline [Kellgren-Lawrence (KL) score of 0-3]. 2D fast spin-echo sequences were used for semi-quantitative morphological scoring of cartilage lesions and a combined T1ρ/T2 sequence was used to quantitatively assess cartilage composition. Progression of hip OA was defined based on incident or progression of morphological semi-quantitative grade at 18 months. Baseline T1ρ and T2 relaxation times were compared between progressors and non-progressors using one-way analysis of variance and Mann-Whitney U tests and used to predict progression with binary logistic regression after adjusting for age, gender, body mass index, and KL score. Additionally, a novel voxel-based relaxometry technique was used to compare the spatial distribution of baseline T1ρ and T2 between progressors and non-progressors. RESULTS: Significantly higher baseline T1ρ and T2 values were observed in hip OA progressors compared to non-progressors, particularly in the posterosuperior and anterior aspects of the femoral cartilage. Logistic regression showed that higher baseline T1ρ or T2 values in the femoral cartilage were significantly associated with progression of femoral cartilage lesions at 18 months. CONCLUSION: T1ρ and T2 relaxation parameters are associated with morphological cartilage degeneration at 18 months and may serve as potential imaging biomarkers for progression of cartilage lesions in hip OA.

The association between MR T1ρ and T2 of cartilage and patient-reported outcomes after ACL injury and reconstruction.

OBJECTIVE: To determine if cartilage T1ρ and T2 relaxation time measures after ACL injury and prior to reconstruction (baseline) are associated with patient-reported outcomes at baseline, 6-months, and 1-year after surgery. DESIGN: Fifty-four ACL-injured participants were scanned in both knees at baseline using 3T MR T1ρ and T2 mapping. Participants also completed Knee-injury and Osteoarthritis Outcome Score (KOOS) and Marx activity level questionnaires at baseline, 6-months, and 1-year after reconstruction. The difference between cartilage T1ρ or T2 of the injured and contralateral knee (side-to-side difference, SSD) was calculated to account for physiological variations among patients. Linear regression models were built to evaluate the association between the baseline SSD T1ρ or T2 and KOOS or Marx at all time points. RESULTS: Higher baseline SSD T1ρ posterolateral tibia (pLT) was associated with worse KOOS in all subscales except symptoms at baseline, worse KOOS pain at 6-months, and worse KOOS in all subscales except sports function at 1-year. Higher baseline SSD T2 femoral trochlea (TrF) was associated with worse KOOS activities of daily living (ADL) at 1-year. Higher baseline SSD T1ρ pLT was associated with lower Marx activity level at 1-year. More severe cartilage lesions, as assessed by Whole-Organ MRI Scoring (WORMS), was significantly associated with worse KOOS pain at 6-months and 1-year. CONCLUSION: T1ρ and T2 of cartilage after ACL injury were associated with KOOS after injury and both KOOS and Marx after reconstruction. Such associations may help clinicians stratify outcomes post-injury, and thus, improve patient management.

Three-dimensional MRI-based statistical shape model and application to a cohort of knees with acute ACL injury.

OBJECTIVE: The aim of this study is to develop a novel 3D magnetic resonance imaging (MRI)-based Statistical Shape Modeling (SSM) and apply it in knee MRIs in order to extract and compare relevant shapes of the tibia and femur in patients with and without acute Anterior cruciate ligament (ACL) injuries. METHODS: Bilateral MR images were acquired and analyzed for 50 patients with acute ACL injuries and for 19 control subjects. A shape model was extracted for the tibia and femur using an SSM algorithm based on a set of matched landmarks that are computed in a fully automatic manner. RESULTS: Shape differences were detected between the knees in the ACL-injury group and control group, suggesting a common shape feature that may predispose these knees to injury. Some of the detected shape features that discriminate between injured and control knees are related to intercondylar width and posterior tibia slope, features that have been suggested in previous studies as ACL morphological risk factors. However, shape modeling has the great potential to quantify these characteristics with a comprehensive description of the surfaces describing complex 3D deformation that cannot be represented with simple geometric indexes. CONCLUSIONS: 3D MRI-based bone shape quantification has the ability to identify specific anatomic risk factors for ACL injury. A better understanding of the role in bony shape on ligamentous injuries could help in the identification of subjects with an increased risk for an ACL tear and to develop targeted prevention strategies, including education and training.

Cartilage T1ρ and T2 relaxation times: longitudinal reproducibility and variations using different coils, MR systems and sites.

OBJECTIVE: To evaluate the longitudinal reproducibility and variations of cartilage T1ρ and T2 measurements using different coils, MR systems and sites. METHODS: Single-Site study: Phantom data were collected monthly for up to 29 months on four GE 3T MR systems. Data from phantoms and human subjects were collected on two MR systems using the same model of coil; and were collected on one MR system using two models of coils. Multi-site study: Three participating sites used the same model of MR systems and coils, and identical imaging protocols. Phantom data were collected monthly. Human subjects were scanned and rescanned on the same day at each site. Two traveling human subjects were scanned at all three sites. RESULTS: Single-Site Study: The phantom longitudinal RMS-CVs ranged from 1.8% to 2.7% for T1ρ and 1.8-2.8% for T2. Significant differences were found in T1ρ and T2 values using different MR systems and coils. Multi-Site Study: The phantom longitudinal RMS-CVs ranged from 1.3% to 2.6% for T1ρ and 1.2-2.7% for T2. Across three sites (n = 16), the in vivo scan-rescan RMS-CV was 3.1% and 4.0% for T1ρ and T2, respectively. Phantom T1ρ and T2 values were significantly different between three sites but highly correlated (R > 0.99). No significant difference was found in T1ρ and T2 values of traveling controls, with cross-site RMS-CV as 4.9% and 4.4% for T1ρ and T2, respectively. CONCLUSION: With careful quality control and cross-calibration, quantitative MRI can be readily applied in multi-site studies and clinical trials for evaluating cartilage degeneration.

Evaluation of 2 Novel Ratio-Based Metrics for Lumbar Spinal Stenosis.

BACKGROUND AND PURPOSE: Quantitative metrics of the dural sac such as the cross-sectional area are commonly used to evaluate central canal stenosis. The aim of this study was to analyze 2 new metrics to measure spinal stenosis on the basis of the ratio between the dural sac and disc cross-sectional areas (DDRCA) and the dural sac and disc anterior-posterior diameters (DDRDIA) and compare them with established quantitative metrics of the dural sac. MATERIALS AND METHODS: T2-weighted axial MR images (n = 260 patients) were retrospectively evaluated, graded for central canal stenosis as normal (no stenosis), mild, moderate, or severe from L1/L2 through L5/S1 with 1 grade per spinal level and annotated to measure the DDRCA and DDRDIA. Thresholds were obtained using a decision tree classifier on a subset of patients (n = 130) and evaluated on the remaining patients (n = 130) for accuracy and consistency across demographics, anatomic variation, and clinical outcomes. RESULTS: DDRCA and DDRDIA had areas under the receiver operating characteristic curve of 98.6 (97.4-99.3) and 98.0 (96.7-98.9) compared with dural sac cross-sectional area at 96.5 (95.0-97.7) for binary classification. DDRDIA and DDRCA had κ scores of 0.75 (0.71-0.79) and 0.80 (0.75-0.83) compared with dural sac cross-sectional area at 0.62 (0.57-0.66) for multigrade classification. No significant differences (P > .1) in the area under the receiver operating characteristic curve were observed for the DDRDIA across variations in the body mass index. The DDRDIA also had the highest area under the receiver operating characteristic curve among symptomatic patients (visual analog scale ≥ 7) or patients who underwent surgery. CONCLUSIONS: Ratio-based metrics (DDRDIA and DDRCA) are accurate and robust to anatomic and demographic variability compared with quantitative metrics of the dural sac and better correlated with symptomatology and surgical outcomes.