Pub Date : 2025-01-01Epub Date: 2025-07-02DOI: 10.1016/j.ostima.2025.100348
R.E. Harari , J. Collins , S.E. Smith , S. Wells , J. Duryea
<div><h3>INTRODUCTION</h3><div>Accurate prediction of knee osteoarthritis (KOA) progression remains a clinical challenge due to its heterogeneous nature and discordance between structural and symptomatic outcomes. Integrated imaging and machine learning (ML) approaches may enhance prognostic modeling but often suffer from limited interpretability or reliance on static features.</div></div><div><h3>OBJECTIVE</h3><div>We aim to develop explainable ML models for predicting KOA progression using baseline and longitudinal imaging and clinical features. This study also aims to identify key imaging biomarkers associated with structural and symptomatic progression.</div></div><div><h3>METHODS</h3><div>Data and 3T MRI measurements from 600 participants in the FNIH OA Biomarkers Consortium were analyzed. Participants were grouped into four progression categories based on 48-month joint space narrowing and WOMAC pain: (1) radiographic + pain progressors, (2) radiographic-only, (3) pain-only, and (4) non-progressors. Two binary classification frameworks were defined: (1) radiographic + pain vs. all others (primary), and (2) all radiographic progressors vs. pain-only + non-progressors (secondary). ML models included Random Forest, XGBoost, logistic regression, decision tree, and multilayer perceptron (MLP). The model used demographic information and imaging features from semi-automated segmentation software. We measured the volume of medial compartment femur cartilage (Cart), bone marrow lesion (BML) in the MF, LF, MT, LT, patella, and trochlea, osteophytes (Ost) in the MF, LF, MT, and LT, Hoffa’s synovitis (HS), and effusion/synovitis (ES). Longitudinal delta values were computed over 24 months. Performance was assessed via 10-fold stratified cross-validation (AUC, F1-score). Explainability tools included SHAP, Gini importance, coefficients, and permutation importance.</div></div><div><h3>RESULTS</h3><div>In the cross-sectional setting, the Random Forest classifier achieved the highest discrimination performance, with AUC values of 0.672 for the primary task (radiographic + pain progressors vs. others) and 0.791 for the secondary task (all radiographic progressors vs. others). The MLP model showed similar results in the secondary task (AUC = 0.743). AUC performance metrics for all models are shown in Table 1. Model performance improved notably when incorporating 24-month changes in imaging features. In the longitudinal analysis, Random Forest again performed best in the secondary task (AUC = 0.873), followed by XGBoost and MLP. The strongest predictors in these models were changes in medial femoral cartilage thickness, medial tibial bone marrow lesions, and osteophyte scores. To better understand the basis of model predictions, we applied four feature ranking methods. Among them, the SHAP method produced the most consistent and clinically interpretable results. As an example, shown in Figure 1 which show top 15 important features, SHAP highlighted 24-month r
{"title":"PREDICTING KNEE OSTEOARTHRITIS PROGRESSION USING EXPLAINABLE MACHINE LEARNING AND CLINICAL IMAGING DATA","authors":"R.E. Harari , J. Collins , S.E. Smith , S. Wells , J. Duryea","doi":"10.1016/j.ostima.2025.100348","DOIUrl":"10.1016/j.ostima.2025.100348","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Accurate prediction of knee osteoarthritis (KOA) progression remains a clinical challenge due to its heterogeneous nature and discordance between structural and symptomatic outcomes. Integrated imaging and machine learning (ML) approaches may enhance prognostic modeling but often suffer from limited interpretability or reliance on static features.</div></div><div><h3>OBJECTIVE</h3><div>We aim to develop explainable ML models for predicting KOA progression using baseline and longitudinal imaging and clinical features. This study also aims to identify key imaging biomarkers associated with structural and symptomatic progression.</div></div><div><h3>METHODS</h3><div>Data and 3T MRI measurements from 600 participants in the FNIH OA Biomarkers Consortium were analyzed. Participants were grouped into four progression categories based on 48-month joint space narrowing and WOMAC pain: (1) radiographic + pain progressors, (2) radiographic-only, (3) pain-only, and (4) non-progressors. Two binary classification frameworks were defined: (1) radiographic + pain vs. all others (primary), and (2) all radiographic progressors vs. pain-only + non-progressors (secondary). ML models included Random Forest, XGBoost, logistic regression, decision tree, and multilayer perceptron (MLP). The model used demographic information and imaging features from semi-automated segmentation software. We measured the volume of medial compartment femur cartilage (Cart), bone marrow lesion (BML) in the MF, LF, MT, LT, patella, and trochlea, osteophytes (Ost) in the MF, LF, MT, and LT, Hoffa’s synovitis (HS), and effusion/synovitis (ES). Longitudinal delta values were computed over 24 months. Performance was assessed via 10-fold stratified cross-validation (AUC, F1-score). Explainability tools included SHAP, Gini importance, coefficients, and permutation importance.</div></div><div><h3>RESULTS</h3><div>In the cross-sectional setting, the Random Forest classifier achieved the highest discrimination performance, with AUC values of 0.672 for the primary task (radiographic + pain progressors vs. others) and 0.791 for the secondary task (all radiographic progressors vs. others). The MLP model showed similar results in the secondary task (AUC = 0.743). AUC performance metrics for all models are shown in Table 1. Model performance improved notably when incorporating 24-month changes in imaging features. In the longitudinal analysis, Random Forest again performed best in the secondary task (AUC = 0.873), followed by XGBoost and MLP. The strongest predictors in these models were changes in medial femoral cartilage thickness, medial tibial bone marrow lesions, and osteophyte scores. To better understand the basis of model predictions, we applied four feature ranking methods. Among them, the SHAP method produced the most consistent and clinically interpretable results. As an example, shown in Figure 1 which show top 15 important features, SHAP highlighted 24-month r","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"5 ","pages":"Article 100348"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144523627","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2025-07-02DOI: 10.1016/j.ostima.2025.100277
K. Balaji , M. Mendoza , P.M. Vicente , C. Galazis , S. Kukran , A.A. Bharath , P.J. Lally , N.K. Bangerter
<div><h3>INTRODUCTION</h3><div>Cartilage T<sub>2</sub> is a non-invasive, microstructural MRI biomarker for KOA, with elevated T<sub>2</sub> indicating early KOA onset. Cartilage T<sub>2</sub> maps could be used in clinical trials to test a drug candidate’s effect on microstructure. Quantitative DESS (qDESS) is widely used for cartilage imaging as it simultaneously acquires 3D, morphological whole knee images and quantitative T<sub>2</sub> maps in ∼5 minutes. Researchers are also developing T<sub>2</sub> mapping techniques using phase-cycled balanced Steady State Free Precession (pc-bSSFP). It is rapid and has higher SNR efficiency than qDESS, which could lead to better 3D morphological image quality and more reliable T<sub>2</sub> maps. PLANET is a technique that uses a minimum of six different pc-bSSFP acquisitions to analytically calculate T<sub>2</sub>. This is too time-consuming to be clinically feasible. In this study, we trained Random Forest (RaFo) machine learning models to estimate T<sub>2</sub> from fewer pc-bSSFP acquisitions to reduce scan time while still estimating reliable voxel-level T<sub>2</sub> values.</div></div><div><h3>OBJECTIVE</h3><div>1) Train and test RaFo models on simulated 4 and 6 pc-bSSFP data and benchmark performance with PLANET. 2) Test RaFo models on in vivo knee data and benchmark performance with the reference T<sub>2</sub> mapping technique (spin echo), PLANET, and qDESS.</div></div><div><h3>METHODS</h3><div>70,000-sample training and 30,000-sample testing datasets were simulated. Each sample corresponded to 12 different pc-bSSFP measurements of the same voxel location in the tissue. The physics-informed simulated datasets were pre-processed, which included sub-sampling from 12 pc-bSSFP measurements to 4 or 6. RaFo models were then trained to estimate T<sub>2</sub> and tested on these pre-processed datasets. Finally, to evaluate performance on noisier <em>in vivo</em> data, fully sampled knee images of two healthy volunteers (HVs, 2F:24-25) were acquired on a 3T Siemens Verio (Erlangen, Germany) with an 8-channel knee coil using 12 measurements of bSSFP (water excitation, 8.6/4.3 ms TR/TE; 22° flip angle; 1 × 1 × 5 mm<sup>3</sup> voxel volume; 128 × 128 × 130 mm<sup>3</sup>), qDESS (water excitation; 20° flip angle; 21.77 ms TR; 6 ms TE; 364 Hz/Px receiver bandwidth; 0 dummy scans per volume), and a gold-standard spin-echo T<sub>2</sub> mapping approach (2500 ms TR; 15, 45, 75 ms TE, 90° and 180° flip angle) with appropriate ethics approval. All images had 1 × 1 × 5 mm<sup>3</sup> voxel volume and 128 × 128 mm<sup>2</sup> field of view. PLANET was tested on 6 pc-bSSFP measurements (labelled PLANET-6). RaFo models were tested on 4 and 6 bSSFP measurements (labelled RaFo-4 and RaFo-6, respectively).</div></div><div><h3>RESULTS</h3><div>Fig1 shows results from simulated data tests, with similar performance across the RaFo models and PLANET. Fig2 shows the in vivo T<sub>2</sub> maps, with the RaFo models visually
{"title":"SIMULTANEOUS 3D CARTILAGE T2 MAPPING AND MORPHOLOGICAL IMAGING WITH RAFO-4 MRI, A MACHINE LEARNING ALGORITHM","authors":"K. Balaji , M. Mendoza , P.M. Vicente , C. Galazis , S. Kukran , A.A. Bharath , P.J. Lally , N.K. Bangerter","doi":"10.1016/j.ostima.2025.100277","DOIUrl":"10.1016/j.ostima.2025.100277","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Cartilage T<sub>2</sub> is a non-invasive, microstructural MRI biomarker for KOA, with elevated T<sub>2</sub> indicating early KOA onset. Cartilage T<sub>2</sub> maps could be used in clinical trials to test a drug candidate’s effect on microstructure. Quantitative DESS (qDESS) is widely used for cartilage imaging as it simultaneously acquires 3D, morphological whole knee images and quantitative T<sub>2</sub> maps in ∼5 minutes. Researchers are also developing T<sub>2</sub> mapping techniques using phase-cycled balanced Steady State Free Precession (pc-bSSFP). It is rapid and has higher SNR efficiency than qDESS, which could lead to better 3D morphological image quality and more reliable T<sub>2</sub> maps. PLANET is a technique that uses a minimum of six different pc-bSSFP acquisitions to analytically calculate T<sub>2</sub>. This is too time-consuming to be clinically feasible. In this study, we trained Random Forest (RaFo) machine learning models to estimate T<sub>2</sub> from fewer pc-bSSFP acquisitions to reduce scan time while still estimating reliable voxel-level T<sub>2</sub> values.</div></div><div><h3>OBJECTIVE</h3><div>1) Train and test RaFo models on simulated 4 and 6 pc-bSSFP data and benchmark performance with PLANET. 2) Test RaFo models on in vivo knee data and benchmark performance with the reference T<sub>2</sub> mapping technique (spin echo), PLANET, and qDESS.</div></div><div><h3>METHODS</h3><div>70,000-sample training and 30,000-sample testing datasets were simulated. Each sample corresponded to 12 different pc-bSSFP measurements of the same voxel location in the tissue. The physics-informed simulated datasets were pre-processed, which included sub-sampling from 12 pc-bSSFP measurements to 4 or 6. RaFo models were then trained to estimate T<sub>2</sub> and tested on these pre-processed datasets. Finally, to evaluate performance on noisier <em>in vivo</em> data, fully sampled knee images of two healthy volunteers (HVs, 2F:24-25) were acquired on a 3T Siemens Verio (Erlangen, Germany) with an 8-channel knee coil using 12 measurements of bSSFP (water excitation, 8.6/4.3 ms TR/TE; 22° flip angle; 1 × 1 × 5 mm<sup>3</sup> voxel volume; 128 × 128 × 130 mm<sup>3</sup>), qDESS (water excitation; 20° flip angle; 21.77 ms TR; 6 ms TE; 364 Hz/Px receiver bandwidth; 0 dummy scans per volume), and a gold-standard spin-echo T<sub>2</sub> mapping approach (2500 ms TR; 15, 45, 75 ms TE, 90° and 180° flip angle) with appropriate ethics approval. All images had 1 × 1 × 5 mm<sup>3</sup> voxel volume and 128 × 128 mm<sup>2</sup> field of view. PLANET was tested on 6 pc-bSSFP measurements (labelled PLANET-6). RaFo models were tested on 4 and 6 bSSFP measurements (labelled RaFo-4 and RaFo-6, respectively).</div></div><div><h3>RESULTS</h3><div>Fig1 shows results from simulated data tests, with similar performance across the RaFo models and PLANET. Fig2 shows the in vivo T<sub>2</sub> maps, with the RaFo models visually","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"5 ","pages":"Article 100277"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144523632","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2025-07-02DOI: 10.1016/j.ostima.2025.100280
M.W. Brejnebøl , T. Haugegaard , R. Christensen , H. Gudbergsen , H. Bliddal , P. Hansen , L.E. Kristensen , C.T. Nielsen , C.L. Daugaard , J.U. Nybing , M. Henriksen , M. Boesen
OBJECTIVE
To compare the effect of weight loss and glucagon-like peptide-1 receptor agonist (GLP-1RA) (liraglutide), relative to weight loss and placebo, on structural knee osteoarthritis.
METHODS
This secondary analysis followed a superiority framework of data from the LOSEIT trial, a randomised, parallel-group, placebo-controlled trial. Participants aged 18 to 74 years with overweight (BMI ≥27 kg/m²), symptomatic and early-to-moderate radiographic knee OA were recruited. They underwent 8-week intensive diet intervention followed by randomisation to receive a GLP-1RA (liraglutide 3 mg/d) or placebo for 52 weeks. The primary outcome was the change in radiographic medial minimal joint space width (mmJSW). Analyses were conducted on the intention-to-treat population.
RESULTS
From November 14, 2016, through September 12, 2017, 156 participants were randomly assigned to GLP-1RA (n = 80) or to placebo (n = 76). As reported in the primary analysis of the data, the GLP-1RA group lost more weight than the placebo group (mean difference, - 3.21 kg, 95%CI: - 6.39 to - 0.03; P=0.050). The GLP-1RA group demonstrated an increase in mean mmJSW of 0.22 mm (95%CI: 0.06 to 0.38) while the placebo group did not change (0.07 mm, 95%CI: - 0.11 to 0.25). No evidence of a difference in mean mmJSW was observed between groups (0.15 mm, 95%CI: -0.06 to 0.36; P=0.17).
CONCLUSION
While the results indicate a potentially favourable effect on mmJSW within the GLP-1RA group, the observed difference in structural knee OA changes on radiographs compared to placebo did not reach statistical significance.
目的比较减肥和胰高血糖素样肽-1受体激动剂（GLP-1RA）（利拉鲁肽）相对于减肥和安慰剂对结构性膝骨关节炎的影响。方法本二次分析采用LOSEIT试验的优势数据框架，该试验是一项随机、平行组、安慰剂对照试验。参与者年龄在18至74岁之间，体重超重（BMI≥27 kg/m²），有症状和早期至中度膝关节炎。他们接受了8周的强化饮食干预，随后随机分配接受GLP-1RA（利拉鲁肽3mg /d）或安慰剂52周。主要观察指标是影像学上内侧最小关节间隙宽度（mmJSW）的变化。对意向治疗人群进行了分析。从2016年11月14日至2017年9月12日，156名参与者被随机分配到GLP-1RA组（n = 80）或安慰剂组（n = 76）。在数据的初步分析中，GLP-1RA组比安慰剂组减轻了更多的体重(平均差，- 3.21 kg, 95%CI: - 6.39至- 0.03；P = 0.050)。GLP-1RA组显示平均mmJSW增加0.22 mm (95%CI: 0.06至0.38)，而安慰剂组没有变化（0.07 mm, 95%CI: - 0.11至0.25）。没有证据表明两组之间的平均mmJSW有差异(0.15 mm, 95%CI: -0.06 ~ 0.36；P = 0.17)。结论：虽然结果表明GLP-1RA组对mmJSW有潜在的有利影响，但与安慰剂相比，在x线片上观察到的膝关节OA结构性变化的差异没有达到统计学意义。
{"title":"THE EFFECT OF WEIGHT LOSS AND GLUCAGON-LIKE PEPTIDE-1 RECEPTOR AGONIST ON STRUCTURAL CHANGES IN KNEE OSTEOARTHRITIS: SECONDARY ANALYSIS OF THE RANDOMISED, PLACEBO-CONTROLLED LOSEIT TRIAL","authors":"M.W. Brejnebøl , T. Haugegaard , R. Christensen , H. Gudbergsen , H. Bliddal , P. Hansen , L.E. Kristensen , C.T. Nielsen , C.L. Daugaard , J.U. Nybing , M. Henriksen , M. Boesen","doi":"10.1016/j.ostima.2025.100280","DOIUrl":"10.1016/j.ostima.2025.100280","url":null,"abstract":"<div><h3>OBJECTIVE</h3><div>To compare the effect of weight loss and glucagon-like peptide-1 receptor agonist (GLP-1RA) (liraglutide), relative to weight loss and placebo, on structural knee osteoarthritis.</div></div><div><h3>METHODS</h3><div>This secondary analysis followed a superiority framework of data from the LOSEIT trial, a randomised, parallel-group, placebo-controlled trial. Participants aged 18 to 74 years with overweight (BMI ≥27 kg/m²), symptomatic and early-to-moderate radiographic knee OA were recruited. They underwent 8-week intensive diet intervention followed by randomisation to receive a GLP-1RA (liraglutide 3 mg/d) or placebo for 52 weeks. The primary outcome was the change in radiographic medial minimal joint space width (mmJSW). Analyses were conducted on the intention-to-treat population.</div></div><div><h3>RESULTS</h3><div>From November 14, 2016, through September 12, 2017, 156 participants were randomly assigned to GLP-1RA (n = 80) or to placebo (n = 76). As reported in the primary analysis of the data, the GLP-1RA group lost more weight than the placebo group (mean difference, - 3.21 kg, 95%CI: - 6.39 to - 0.03; P=0.050). The GLP-1RA group demonstrated an increase in mean mmJSW of 0.22 mm (95%CI: 0.06 to 0.38) while the placebo group did not change (0.07 mm, 95%CI: - 0.11 to 0.25). No evidence of a difference in mean mmJSW was observed between groups (0.15 mm, 95%CI: -0.06 to 0.36; P=0.17).</div></div><div><h3>CONCLUSION</h3><div>While the results indicate a potentially favourable effect on mmJSW within the GLP-1RA group, the observed difference in structural knee OA changes on radiographs compared to placebo did not reach statistical significance.</div></div>","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"5 ","pages":"Article 100280"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144523633","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2025-07-02DOI: 10.1016/j.ostima.2025.100340
S.L. Westbrook , A. Guermazi , P.G. Conaghan
<div><h3>INTRODUCTION</h3><div>Bone marrow lesions (BMLs), detectable on MRI as areas of ill-defined high signal intensity on fluid-sensitive sequences, are a common feature of osteoarthritis (OA), representing areas of increased bone turnover, oedema, and fibrosis. BMLs are prevalent in ∼80% of symptomatic knee OA patients, correlate with radiographic severity (Kellgren-Lawrence [KL] grade) and knee pain. Changes in BMLs are associated with fluctuations in knee pain. Excess neurotrophins (NTs) are implicated in OA pain. LEVI-04, a first-in-class p75NTR-Fc fusion protein that supplements endogenous p75NTR, provides analgesia primarily via inhibition of neurotrophin-3 (NT-3) activity. In this Phase II RCT, LEVI-04 demonstrated statistically significant and clinically meaningful improvements versus placebo for the primary endpoint (WOMAC pain) and secondary endpoints including WOMAC physical function and stiffness, patient global assessment (PGA) and pain on movement (StEPP) across all doses. LEVI-04 was generally well tolerated, with no increased incidence of SAEs, TEAEs, or AESIs concerning joint pathologies compared to placebo.<sup>1</sup></div></div><div><h3>OBJECTIVE</h3><div>This analysis investigated LEVI-04′s effects on BMLs in people with painful knee OA.</div></div><div><h3>METHODS</h3><div>518 participants with symptomatic knee OA (WOMAC pain ≥ 4/10, KL grade ≥ 2) were enrolled in a Phase II multicentre randomized double-blinded placebo-controlled trial. Participants received placebo or LEVI-04 (0.3, 1, or 2 mg/kg) every 4 weeks through week 16. BML area (mm²) was measured in a blinded fashion from coronal proton density-weighted fat-suppressed (PD-FS) sequences (slice thickness 3 mm, TE/TR 35/3000 ms) of the target knee at baseline and week 20. For each participant, the BML area was determined as the largest area within the MRI sequence of ill-defined high signal intensity of the subchondral bone marrow, and without presence of a fracture line. The perimeter of each BML was highlighted and the area measured electronically using IAG Dynamika Software™. For BML presence, participants were categorized as BML positive if one or more lesions were identified in the target knee. The presence of BML and change in BML area were assessed in response to LEVI-04.</div></div><div><h3>RESULTS</h3><div>BML area was greater in knees with higher KL grade (figure 1). The presence of BMLs at baseline was similar across treatment and placebo groups (74-79%). At week 20, there was a significant and dose-dependent reduction in the proportion of patients with BMLs in the LEVI-04 groups (figure 2). Furthermore, a statistically-significant, dose-dependent reduction in mean BML area from baseline to week 20 was observed in LEVI-04 groups compared to placebo (figure 3).</div></div><div><h3>CONCLUSION</h3><div>In this Phase II trial, a statistically significant and dose-dependent reduction in both the presence of BMLs and BML area was seen for all LEV-04 treatment
{"title":"LEVI-04 REDUCES BONE MARROW LESION AREA AND PRESENCE IN KNEE OSTEOARTHRITIS: RESULTS FROM A PHASE II RCT","authors":"S.L. Westbrook , A. Guermazi , P.G. Conaghan","doi":"10.1016/j.ostima.2025.100340","DOIUrl":"10.1016/j.ostima.2025.100340","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Bone marrow lesions (BMLs), detectable on MRI as areas of ill-defined high signal intensity on fluid-sensitive sequences, are a common feature of osteoarthritis (OA), representing areas of increased bone turnover, oedema, and fibrosis. BMLs are prevalent in ∼80% of symptomatic knee OA patients, correlate with radiographic severity (Kellgren-Lawrence [KL] grade) and knee pain. Changes in BMLs are associated with fluctuations in knee pain. Excess neurotrophins (NTs) are implicated in OA pain. LEVI-04, a first-in-class p75NTR-Fc fusion protein that supplements endogenous p75NTR, provides analgesia primarily via inhibition of neurotrophin-3 (NT-3) activity. In this Phase II RCT, LEVI-04 demonstrated statistically significant and clinically meaningful improvements versus placebo for the primary endpoint (WOMAC pain) and secondary endpoints including WOMAC physical function and stiffness, patient global assessment (PGA) and pain on movement (StEPP) across all doses. LEVI-04 was generally well tolerated, with no increased incidence of SAEs, TEAEs, or AESIs concerning joint pathologies compared to placebo.<sup>1</sup></div></div><div><h3>OBJECTIVE</h3><div>This analysis investigated LEVI-04′s effects on BMLs in people with painful knee OA.</div></div><div><h3>METHODS</h3><div>518 participants with symptomatic knee OA (WOMAC pain ≥ 4/10, KL grade ≥ 2) were enrolled in a Phase II multicentre randomized double-blinded placebo-controlled trial. Participants received placebo or LEVI-04 (0.3, 1, or 2 mg/kg) every 4 weeks through week 16. BML area (mm²) was measured in a blinded fashion from coronal proton density-weighted fat-suppressed (PD-FS) sequences (slice thickness 3 mm, TE/TR 35/3000 ms) of the target knee at baseline and week 20. For each participant, the BML area was determined as the largest area within the MRI sequence of ill-defined high signal intensity of the subchondral bone marrow, and without presence of a fracture line. The perimeter of each BML was highlighted and the area measured electronically using IAG Dynamika Software™. For BML presence, participants were categorized as BML positive if one or more lesions were identified in the target knee. The presence of BML and change in BML area were assessed in response to LEVI-04.</div></div><div><h3>RESULTS</h3><div>BML area was greater in knees with higher KL grade (figure 1). The presence of BMLs at baseline was similar across treatment and placebo groups (74-79%). At week 20, there was a significant and dose-dependent reduction in the proportion of patients with BMLs in the LEVI-04 groups (figure 2). Furthermore, a statistically-significant, dose-dependent reduction in mean BML area from baseline to week 20 was observed in LEVI-04 groups compared to placebo (figure 3).</div></div><div><h3>CONCLUSION</h3><div>In this Phase II trial, a statistically significant and dose-dependent reduction in both the presence of BMLs and BML area was seen for all LEV-04 treatment","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"5 ","pages":"Article 100340"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144523446","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2025-07-02DOI: 10.1016/j.ostima.2025.100343
M.A. van den Berg , F. Boel , M.M.A. van Buuren , N.S. Riedstra , J. Tang , H. Ahedi , N. Arden , S.M.A. Bierma-Zeinstra , C.G. Boer , F.M. Cicuttini , T.F. Cootes , K.M. Crossley , D.T. Felson , W.P. Gielis , J.J. Heerey , G. Jones , S. Kluzek , N.E. Lane , C. Lindner , J.A. Lynch , R. Agricola
<div><h3>INTRODUCTION</h3><div>Radiographic hip osteoarthritis (RHOA) is a multifactorial disease, making early detection of individuals at risk challenging yet essential for timely intervention and evaluation of preventive strategies. Integrating information on multiple different data modalities using individual participant data from diverse cohorts may enhance predictive modeling in the early stages of RHOA. A focus on model interpretability may further enable the identification of clinically relevant patient subgroups and potential intervention targets.</div></div><div><h3>OBJECTIVE</h3><div>Creating a multi-modal prediction model for improving the performance of RHOA incidence prediction models compared to clinical features alone, and investigating the estimated predictor effects and the generalizability of the models to similar populations.</div></div><div><h3>METHODS</h3><div>We pooled individual participant data from nine prospective cohort studies within the Worldwide Collaboration on OsteoArthritis prediCtion for the Hip (World COACH consortium). All studies included standardized anteroposterior pelvic, long-limb, and/or hip radiographs, assessed for RHOA at baseline and after 4–8 years of follow-up. Incident RHOA was defined as the development of RHOA (grade ≥2) in hips without definite RHOA at baseline (grade <2). The original cohort values of clinical predictors including age, birth-assigned sex, body mass index (BMI), smoking status, diabetes, and hip pain were harmonized into one consistent dataset. X-ray-derived predictors describing the hip morphology, the alpha angle and the lateral center edge angle, were automatically and uniformly determined using automated landmark points placed with Bonefinder®. Additionally, the values of 13 shape modes explaining 85% of the variation from a landmark-based statistical shape model were included. This SSM was built on all baseline RHOA grade <2 hips within World COACH. Risk prediction models were built with generalized linear mixed effects models (GLMM) and Random Forest (RF) models while adjusting for correlations within cohorts and individuals. The discriminative performance (AUC) of different model configurations and the linear versus non-linear approaches were compared through stratified 5-fold cross-validation. For each model configuration, predictions were made with and without cohort labels to assess heterogeneity between cohorts.</div></div><div><h3>RESULTS</h3><div>In total, 29,110 hips without definite RHOA at baseline were included of which 5.0% developed RHOA within 4-8 years (mean age 63.7 (8.6) years, 75.5% female, mean BMI 27.5 (4.7) kg/m<sup>2</sup>). When comparing our uni-modal prediction model using only the clinical predictors (Model 1) to those with X-ray information added (Table 1), we observed a higher discriminative performance for the multi-modal models. Overall, including cohort information significantly improved model performance (p < 0.05), and the RF mode
{"title":"ADVANCING HIP OSTEOARTHRITIS PREDICTION: INSIGHTS FROM MULTI-MODAL PREDICTIVE MODELING WITH INDIVIDUAL PARTICIPANT DATA OF THE WORLD COACH CONSORTIUM","authors":"M.A. van den Berg , F. Boel , M.M.A. van Buuren , N.S. Riedstra , J. Tang , H. Ahedi , N. Arden , S.M.A. Bierma-Zeinstra , C.G. Boer , F.M. Cicuttini , T.F. Cootes , K.M. Crossley , D.T. Felson , W.P. Gielis , J.J. Heerey , G. Jones , S. Kluzek , N.E. Lane , C. Lindner , J.A. Lynch , R. Agricola","doi":"10.1016/j.ostima.2025.100343","DOIUrl":"10.1016/j.ostima.2025.100343","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Radiographic hip osteoarthritis (RHOA) is a multifactorial disease, making early detection of individuals at risk challenging yet essential for timely intervention and evaluation of preventive strategies. Integrating information on multiple different data modalities using individual participant data from diverse cohorts may enhance predictive modeling in the early stages of RHOA. A focus on model interpretability may further enable the identification of clinically relevant patient subgroups and potential intervention targets.</div></div><div><h3>OBJECTIVE</h3><div>Creating a multi-modal prediction model for improving the performance of RHOA incidence prediction models compared to clinical features alone, and investigating the estimated predictor effects and the generalizability of the models to similar populations.</div></div><div><h3>METHODS</h3><div>We pooled individual participant data from nine prospective cohort studies within the Worldwide Collaboration on OsteoArthritis prediCtion for the Hip (World COACH consortium). All studies included standardized anteroposterior pelvic, long-limb, and/or hip radiographs, assessed for RHOA at baseline and after 4–8 years of follow-up. Incident RHOA was defined as the development of RHOA (grade ≥2) in hips without definite RHOA at baseline (grade <2). The original cohort values of clinical predictors including age, birth-assigned sex, body mass index (BMI), smoking status, diabetes, and hip pain were harmonized into one consistent dataset. X-ray-derived predictors describing the hip morphology, the alpha angle and the lateral center edge angle, were automatically and uniformly determined using automated landmark points placed with Bonefinder®. Additionally, the values of 13 shape modes explaining 85% of the variation from a landmark-based statistical shape model were included. This SSM was built on all baseline RHOA grade <2 hips within World COACH. Risk prediction models were built with generalized linear mixed effects models (GLMM) and Random Forest (RF) models while adjusting for correlations within cohorts and individuals. The discriminative performance (AUC) of different model configurations and the linear versus non-linear approaches were compared through stratified 5-fold cross-validation. For each model configuration, predictions were made with and without cohort labels to assess heterogeneity between cohorts.</div></div><div><h3>RESULTS</h3><div>In total, 29,110 hips without definite RHOA at baseline were included of which 5.0% developed RHOA within 4-8 years (mean age 63.7 (8.6) years, 75.5% female, mean BMI 27.5 (4.7) kg/m<sup>2</sup>). When comparing our uni-modal prediction model using only the clinical predictors (Model 1) to those with X-ray information added (Table 1), we observed a higher discriminative performance for the multi-modal models. Overall, including cohort information significantly improved model performance (p < 0.05), and the RF mode","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"5 ","pages":"Article 100343"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144523923","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2025-07-02DOI: 10.1016/j.ostima.2025.100292
N. Hendriks , F. Boel , C. Lindner , F. Rivadeneira , C.J. Tiderius , S.M.A. Bierma-Zeinstra , R. Agricola , J. Runhaar
INTRODUCTION
Acetabular dysplasia (AD) is an important risk factor for early hip OA in adults. In Europe, infants are screened for developmental hip dysplasia. However, AD can also develop during skeletal maturation and these cases often remain unrecognized. Potentially, AD could be influenced prior to the closure of the hip growth plates. Understanding AD development during growth is crucial to prevent future joint degeneration. Different definitions are used to measure AD, depending on the stage of skeletal maturation. More knowledge of the prevalence of AD in the general population is required to understand its development during growth.
OBJECTIVE
1) To estimate the prevalence of AD in 6-year-olds from the general population, and 2) to compare different AD definitions in this age group.
METHODS
Data from The Generation R Study, a population-based study examining growth and health from fetal life to adulthood, was used. All participants aged 6 years, with high-resolution dual-energy x-ray absorptiometry (DXA) anteroposterior image of the right hip available were included. The hip shape was outlined with 70 landmarks using BoneFinder®. Using these landmarks, the acetabular index (AI), a measurement of acetabular roof inclination, was calculated to assess AD (AI>20°). While AI is commonly used in children, the lateral center-edge angle (LCEA), as indicator for acetabular roof coverage of the femoral head, was also calculated. Mean LCEA and prevalence of AD (LCEA<15°) were compared to measures using AI.
RESULTS
In total, 3,270 participants were included with a mean age of 6.2 (SD 0.6) years, and 51% was female. The mean AI was 11.3° (SD 5.0°) and the mean LCEA was 19.5° (SD 5.9°). The distribution for both AD definitions is shown in Figure 1. An AI>20° was found in 124 participants, indicating a AD prevalence of 3.8% (95%CI, 3.1% - 4.5%). Based on the LCEA, the AD prevalence was 21.3% (95%CI, 19.9% - 22.7%).
CONCLUSION
The prevalence of AD in 6-year-olds is 3.8%, based on the AI. The LCEA classifies more hips as dysplastic in 6-year-olds. The validity of the LCEA in this age group and clinical relevance of these newly classified dysplastic hips need to be determined. A better understanding of the development of AD is important, as recovery during growth may be feasible and could contribute to the prevention of OA.
{"title":"PREVALENCE OF ACETABULAR DYSPLASIA IN 6-YEAR-OLDS IN A GENERAL POPULATION","authors":"N. Hendriks , F. Boel , C. Lindner , F. Rivadeneira , C.J. Tiderius , S.M.A. Bierma-Zeinstra , R. Agricola , J. Runhaar","doi":"10.1016/j.ostima.2025.100292","DOIUrl":"10.1016/j.ostima.2025.100292","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Acetabular dysplasia (AD) is an important risk factor for early hip OA in adults. In Europe, infants are screened for developmental hip dysplasia. However, AD can also develop during skeletal maturation and these cases often remain unrecognized. Potentially, AD could be influenced prior to the closure of the hip growth plates. Understanding AD development during growth is crucial to prevent future joint degeneration. Different definitions are used to measure AD, depending on the stage of skeletal maturation. More knowledge of the prevalence of AD in the general population is required to understand its development during growth.</div></div><div><h3>OBJECTIVE</h3><div>1) To estimate the prevalence of AD in 6-year-olds from the general population, and 2) to compare different AD definitions in this age group.</div></div><div><h3>METHODS</h3><div>Data from The Generation R Study, a population-based study examining growth and health from fetal life to adulthood, was used. All participants aged 6 years, with high-resolution dual-energy x-ray absorptiometry (DXA) anteroposterior image of the right hip available were included. The hip shape was outlined with 70 landmarks using BoneFinder®. Using these landmarks, the acetabular index (AI), a measurement of acetabular roof inclination, was calculated to assess AD (AI>20°). While AI is commonly used in children, the lateral center-edge angle (LCEA), as indicator for acetabular roof coverage of the femoral head, was also calculated. Mean LCEA and prevalence of AD (LCEA<15°) were compared to measures using AI.</div></div><div><h3>RESULTS</h3><div>In total, 3,270 participants were included with a mean age of 6.2 (SD 0.6) years, and 51% was female. The mean AI was 11.3° (SD 5.0°) and the mean LCEA was 19.5° (SD 5.9°). The distribution for both AD definitions is shown in Figure 1. An AI>20° was found in 124 participants, indicating a AD prevalence of 3.8% (95%CI, 3.1% - 4.5%). Based on the LCEA, the AD prevalence was 21.3% (95%CI, 19.9% - 22.7%).</div></div><div><h3>CONCLUSION</h3><div>The prevalence of AD in 6-year-olds is 3.8%, based on the AI. The LCEA classifies more hips as dysplastic in 6-year-olds. The validity of the LCEA in this age group and clinical relevance of these newly classified dysplastic hips need to be determined. A better understanding of the development of AD is important, as recovery during growth may be feasible and could contribute to the prevention of OA.</div></div>","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"5 ","pages":"Article 100292"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144523990","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2025-07-02DOI: 10.1016/j.ostima.2025.100300
H. Liu, J.L. Gregory, M.O. Silva, C.E. Davey, K.S. Stok
<div><h3>INTRODUCTION</h3><div>Longitudinal assessment of knee joint structure holds promise for providing invaluable spatial-temporal information on the progression of degenerative musculoskeletal (MSK) diseases involving the knee joint.</div></div><div><h3>OBJECTIVE</h3><div>This proof-of-concept study aims to establish a time-lapse <em>in vivo</em> imaging protocol with high temporal resolution to longitudinally track multi-scale structural changes, including mechanical alteration to whole joint structure, sensitive microstructural changes to subchondral bone, and abnormal bone remodeling activity, in a mouse collagenase-induced osteoarthritis (OA) model.</div></div><div><h3>METHODS</h3><div>Eight male C57BL/10 mice aged nine weeks were recruited and assigned to two longitudinal groups, control (CT) and OA. Of these, four ten-week-old mice assigned to the OA group received intra-articular injection of collagenase on the right knee to destabilize the right tibiofemoral joint. Longitudinal <em>in vivo</em> micro-computed tomography (microCT) scans were performed one day before collagenase injection and then weekly for eight weeks in total, resulting in nine scans for each animal. <em>In vivo</em> microCT (Scanco Medical) was performed with a source voltage of 70 kVp, an integration time of 350 <em>ms</em>, a current of 114 μ<em>A</em>, and an isotropic nominal resolution of 10.4 μ<em>m</em> with 1000 projections, with each scanning taking around 30 minutes. Quantitative morphometric analysis (QMA) was performed to measure longitudinal changes to structure of whole joint and subchondral bone, including joint space width (mm), and trabecular thickness (mm). Visualization of dynamic bone remodeling was performed by registering serial microCT scans. Bone resorption rate, BRR (%/day), and bone formation rate, BFR (%/day) were measured to quantify bone remodeling activity. To test the differences between CT and OA group at each time point from week 1 to week 8, a one-way analysis of covariance was used.</div></div><div><h3>RESULTS</h3><div>Three weeks post OA-induction, a significantly smaller joint space width was observed in medial osteoarthritic joint (202 μm), when compared to CT joint (228 μm) (p < 0.01). Regarding trabecular thickness, significant differences were observed at multiple time points between CT and OA groups, specifically in the first three weeks at the early stage of OA progression at lateral side (p < 0.01). Representative 3D visualization of bone formation and bone resorption is shown in <strong>Figure 1 A-B</strong>. Abnormal bone remodeling activities were observed in osteoarthritic femur. When compared to control femur, significantly larger bone resorption rate was observed in the first week post collagenase injection in both the lateral (p < 0.01) and medial femur (p < 0.01), as shown in <strong>Figure 1 C-D</strong>.</div></div><div><h3>CONCLUSION</h3><div>This proof-of-concept study, for the first time, demonstr
{"title":"IN VIVO MICRO COMPUTED TOMOGRAPHY IMAGING ALLOWS LONGITUDINAL ASSESSMENT OF MULTI-SCALE CHANGES TO WHOLE JOINT WITH PROGRESSION OF OA","authors":"H. Liu, J.L. Gregory, M.O. Silva, C.E. Davey, K.S. Stok","doi":"10.1016/j.ostima.2025.100300","DOIUrl":"10.1016/j.ostima.2025.100300","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Longitudinal assessment of knee joint structure holds promise for providing invaluable spatial-temporal information on the progression of degenerative musculoskeletal (MSK) diseases involving the knee joint.</div></div><div><h3>OBJECTIVE</h3><div>This proof-of-concept study aims to establish a time-lapse <em>in vivo</em> imaging protocol with high temporal resolution to longitudinally track multi-scale structural changes, including mechanical alteration to whole joint structure, sensitive microstructural changes to subchondral bone, and abnormal bone remodeling activity, in a mouse collagenase-induced osteoarthritis (OA) model.</div></div><div><h3>METHODS</h3><div>Eight male C57BL/10 mice aged nine weeks were recruited and assigned to two longitudinal groups, control (CT) and OA. Of these, four ten-week-old mice assigned to the OA group received intra-articular injection of collagenase on the right knee to destabilize the right tibiofemoral joint. Longitudinal <em>in vivo</em> micro-computed tomography (microCT) scans were performed one day before collagenase injection and then weekly for eight weeks in total, resulting in nine scans for each animal. <em>In vivo</em> microCT (Scanco Medical) was performed with a source voltage of 70 kVp, an integration time of 350 <em>ms</em>, a current of 114 μ<em>A</em>, and an isotropic nominal resolution of 10.4 μ<em>m</em> with 1000 projections, with each scanning taking around 30 minutes. Quantitative morphometric analysis (QMA) was performed to measure longitudinal changes to structure of whole joint and subchondral bone, including joint space width (mm), and trabecular thickness (mm). Visualization of dynamic bone remodeling was performed by registering serial microCT scans. Bone resorption rate, BRR (%/day), and bone formation rate, BFR (%/day) were measured to quantify bone remodeling activity. To test the differences between CT and OA group at each time point from week 1 to week 8, a one-way analysis of covariance was used.</div></div><div><h3>RESULTS</h3><div>Three weeks post OA-induction, a significantly smaller joint space width was observed in medial osteoarthritic joint (202 μm), when compared to CT joint (228 μm) (p < 0.01). Regarding trabecular thickness, significant differences were observed at multiple time points between CT and OA groups, specifically in the first three weeks at the early stage of OA progression at lateral side (p < 0.01). Representative 3D visualization of bone formation and bone resorption is shown in <strong>Figure 1 A-B</strong>. Abnormal bone remodeling activities were observed in osteoarthritic femur. When compared to control femur, significantly larger bone resorption rate was observed in the first week post collagenase injection in both the lateral (p < 0.01) and medial femur (p < 0.01), as shown in <strong>Figure 1 C-D</strong>.</div></div><div><h3>CONCLUSION</h3><div>This proof-of-concept study, for the first time, demonstr","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"5 ","pages":"Article 100300"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144524185","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2025-07-02DOI: 10.1016/j.ostima.2025.100349
W. Wirth , F. Eckstein
<div><h3>INTRODUCTION</h3><div>Automated cartilage segmentation using convolutional neural networks (CNN) has been shown to provide moderate to high accuracy in comparison with gold-standard manual approaches. It also displays similar sensitivity to longitudinal change and to between-group differences in change as has been reported for manual analysis [1-3]. Denuded areas of subchondral bone (dAB) provide challenges and impair the accuracy of automated cartilage segmentation in knees with severe radiographic OA (KLG 4). The reason is that CNNs are trained to detect cartilage, but encounter “difficulties” to properly segment areas where cartilage is lost entirely. CNNs therefore often segment cartilage cover in some areas of actual full thickness loss or ignore dABs entirely. This was observed to result in an overestimation of cartilage thickness and an underestimation of dABs in knees with severe OA [4].</div></div><div><h3>OBJECTIVE</h3><div>To improve CNN-based automated segmentation in severely osteoarthritic knee cartilage by using an automated post-processing algorithm that relies on a multi-atlas registration for reconstructing the total area of subchondral bone (tAB). We evaluate the agreement, accuracy and longitudinal sensitivity to cartilage change of this new methodology.</div></div><div><h3>METHODS</h3><div>Sagittal DESS and coronal FLASH MRIs were acquired by the Osteoarthritis Initiative (OAI). 2D U-Net models were trained for both MRI protocols using manual cartilage segmentations of knees with radiographic OA (KLG2-4, n training / validation set: 86/18 knees, baseline scans only) or severe radiographic OA (KLG4, n training/ validation set: 29/6 knees. These were trained either from baseline scans only [KLG4<sub>BL</sub>] or from baseline and follow-up scans [KLG4<sub>BL+FU</sub>]. The trained models were then applied to the test set comprising 10 KLG4 knees with manual cartilage segmentations from both DESS and FLASH MRI available and to n=125/14 knees with manual cartilage segmentations from either DESS or FLASH MRI available. Automated, registration-based post-processing was applied to reconstruct missing parts of the tAB and to refine the segmentations (Fig. 1), particularly in areas of denuded bone. The agreement and accuracy of automated cartilage analysis were evaluated in the test set for individual cartilages using Dice Similarity coefficients (DSC), correlation analysis, and by determining systematic offsets between manual and automated analysis. The sensitivity to one-year change was assessed using the standardized response mean (SRM) across the entire femorotibial joint in 104/24 (DESS/FLASH) knees with manual baseline and follow-up segmentations.</div></div><div><h3>RESULTS</h3><div>The strongest agreement (DSC 0.80±0.07 to 0.89±0.05) and lowest systematic offsets for cartilage thickness (1.2% to 8.5%) were observed for CNNs trained on KLG2-4 rather than KLG4 knees. Similar observations were made for dABs (-40.6% to 3.
{"title":"A FULLY-AUTOMATED TECHNIQUE FOR KNEE CARTILAGE AND DENUDED BONE AREA MORPHOMETRY IN SEVERE RADIOGRAPHIC KNEE OA – METHOD DEVELOPMENT AND VALIDATION","authors":"W. Wirth , F. Eckstein","doi":"10.1016/j.ostima.2025.100349","DOIUrl":"10.1016/j.ostima.2025.100349","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Automated cartilage segmentation using convolutional neural networks (CNN) has been shown to provide moderate to high accuracy in comparison with gold-standard manual approaches. It also displays similar sensitivity to longitudinal change and to between-group differences in change as has been reported for manual analysis [1-3]. Denuded areas of subchondral bone (dAB) provide challenges and impair the accuracy of automated cartilage segmentation in knees with severe radiographic OA (KLG 4). The reason is that CNNs are trained to detect cartilage, but encounter “difficulties” to properly segment areas where cartilage is lost entirely. CNNs therefore often segment cartilage cover in some areas of actual full thickness loss or ignore dABs entirely. This was observed to result in an overestimation of cartilage thickness and an underestimation of dABs in knees with severe OA [4].</div></div><div><h3>OBJECTIVE</h3><div>To improve CNN-based automated segmentation in severely osteoarthritic knee cartilage by using an automated post-processing algorithm that relies on a multi-atlas registration for reconstructing the total area of subchondral bone (tAB). We evaluate the agreement, accuracy and longitudinal sensitivity to cartilage change of this new methodology.</div></div><div><h3>METHODS</h3><div>Sagittal DESS and coronal FLASH MRIs were acquired by the Osteoarthritis Initiative (OAI). 2D U-Net models were trained for both MRI protocols using manual cartilage segmentations of knees with radiographic OA (KLG2-4, n training / validation set: 86/18 knees, baseline scans only) or severe radiographic OA (KLG4, n training/ validation set: 29/6 knees. These were trained either from baseline scans only [KLG4<sub>BL</sub>] or from baseline and follow-up scans [KLG4<sub>BL+FU</sub>]. The trained models were then applied to the test set comprising 10 KLG4 knees with manual cartilage segmentations from both DESS and FLASH MRI available and to n=125/14 knees with manual cartilage segmentations from either DESS or FLASH MRI available. Automated, registration-based post-processing was applied to reconstruct missing parts of the tAB and to refine the segmentations (Fig. 1), particularly in areas of denuded bone. The agreement and accuracy of automated cartilage analysis were evaluated in the test set for individual cartilages using Dice Similarity coefficients (DSC), correlation analysis, and by determining systematic offsets between manual and automated analysis. The sensitivity to one-year change was assessed using the standardized response mean (SRM) across the entire femorotibial joint in 104/24 (DESS/FLASH) knees with manual baseline and follow-up segmentations.</div></div><div><h3>RESULTS</h3><div>The strongest agreement (DSC 0.80±0.07 to 0.89±0.05) and lowest systematic offsets for cartilage thickness (1.2% to 8.5%) were observed for CNNs trained on KLG2-4 rather than KLG4 knees. Similar observations were made for dABs (-40.6% to 3.","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"5 ","pages":"Article 100349"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144522381","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2025-07-02DOI: 10.1016/j.ostima.2025.100341
F. Boel , M.A. van den Berg , N.S. Riedstra , M.M.A. van Buuren , J. Tang , H. Ahedi , N. Arden , S.M.A. Bierma-Zeinstra , C.G. Boer , F.M. Cicuttini , T.F. Cootes , K.M. Crossley , D.T. Felson , W.P. Gielis , J.J. Heerey , G. Jones , S. Kluzek , N.E. Lane , C. Lindner , J.A. Lynch , R. Agricola
<div><h3>INTRODUCTION</h3><div>Hip morphology has been recognized as an important risk factor for the development of hip OA. In previous studies within the Worldwide Collaboration on OsteoArthritis prediCtion for the Hip consortium (World COACH), both acetabular dysplasia (AD) and pincer morphology–characterized by acetabular under- and overcoverage of the femoral head–were associated with the development of radiographic hip OA (RHOA) within 4-8 years, with an odds ratio (OR) of 1.80 (95% confidence interval (CI) 1.40-2.34) and 1.50 (95% CI 1.05-2.15), respectively. However, we know that not everyone with AD or pincer morphology will develop RHOA. Specific baseline characteristics or variations in hip shape among individuals with AD and pincer morphology may influence their risk of developing RHOA. Statistical shape models (SSM), describing the mean hip shape of a population and a range of independent shape variations, can be utilized to study these variations in hip shape.</div></div><div><h3>OBJECTIVE</h3><div>To evaluate whether specific hip shape variations or baseline characteristics within individuals with either AD or pincer morphology are associated with the development of RHOA within 4-8 years.</div></div><div><h3>METHODS</h3><div>We pooled individual participant data from seven prospective cohort studies within the World COACH consortium. Standardized anteroposterior (AP) pelvic radiographs were obtained at baseline and within 4-8 years follow-up. RHOA was scored by KLG or (modified) Croft grade. We harmonized the RHOA scores into “No OA” (KLG/Croft = 0), “doubtful OA” (KLG/Croft = 1), or “definite OA” (KLG/Croft ≥ 2 or total hip replacement). The Wiberg center edge angle (WCEA), measuring the weight-bearing femoral head coverage, and the lateral center edge angle (LCEA), measuring the bony femoral head coverage, were automatically determined using a validated method. Hips were included if they had baseline and follow-up RHOA scores, no RHOA at baseline, and either AD defined by a WCEA ≤ 25° or pincer morphology defined by a LCEA ≥45°. For both populations, an SSM was created of the acetabular roof, posterior wall, femoral head and neck, and teardrop (Fig 1). We analyzed the first 13 shape modes that explained around 90% of total shape variation in the population. The association between each shape mode, sex, baseline age, BMI, diabetes and smoking habits, and the development of RHOA was estimated using univariate generalized linear mixed-effects models. The mixed effects were added to account for the potential clustering within cohorts and participants. The results were expressed as ORs with 95% CIs.</div></div><div><h3>RESULTS</h3><div>The AD population consisted of 4,737 hips, of which 2.6% developed incident RHOA (Table 1). Four of the 13 shape modes (Fig 1) were associated with the development of RHOA. Additionally, in hips with AD, females had higher odds of incident RHOA than males (OR 2.85, 95% CI 1.46 – 5.58), and each year inc
髋关节形态已被认为是髋关节骨关节炎发生的重要危险因素。在全球髋关节骨关节炎预测合作联盟（World COACH）之前的研究中，髋臼发育不良（AD）和钳形（以髋臼股骨头覆盖不足和过度为特征）与4-8年内髋关节骨性关节炎（RHOA）的发生相关，比值比（OR）分别为1.80（95%可信区间（CI） 1.40-2.34）和1.50 （95% CI 1.05-2.15）。然而，我们知道不是每个患有AD或钳形形态的人都会发展RHOA。特定的基线特征或AD和钳形形态个体的臀部形状变化可能影响他们发展RHOA的风险。统计形状模型（SSM）描述了一个群体的平均臀部形状和一系列独立的形状变化，可以用来研究臀部形状的这些变化。目的评估AD或钳形形态患者的特定髋关节形状变化或基线特征是否与4-8年内RHOA的发生有关。方法：我们汇集了来自世界COACH联盟的7项前瞻性队列研究的个体参与者数据。在基线和4-8年随访期间获得标准化骨盆正位（AP） x线片。RHOA采用KLG或（改良的）Croft评分。我们将RHOA评分统一为“无OA”（KLG/Croft = 0）、“可疑OA”（KLG/Croft = 1）或“明确OA”（KLG/Croft≥2或全髋关节置换术）。采用经过验证的方法自动确定Wiberg中心边缘角（WCEA）和外侧中心边缘角（LCEA），分别用于测量负重股骨头覆盖率和骨股骨头覆盖率。如果髋关节有基线和随访的RHOA评分，基线时无RHOA，且WCEA≤25°定义的AD或LCEA≥45°定义的钳形形态，则纳入髋部。对于这两组患者，对髋臼顶、后壁、股骨头、颈和泪滴进行SSM（图1）。我们分析了前13种形状模式，它们解释了种群中约90%的总形状变化。使用单变量广义线性混合效应模型估计每种体型模式、性别、基线年龄、BMI、糖尿病和吸烟习惯与RHOA发展之间的关系。加入混合效应是为了解释在队列和参与者中潜在的聚类。结果以or表示，ci为95%。结果AD人群包括4737例髋关节，其中2.6%发生了RHOA（表1）。13种形状模式中的4种（图1）与RHOA的发展有关。此外，在患有AD的髋关节中，女性发生RHOA的几率高于男性（OR 2.85, 95% CI 1.46 - 5.58），并且基线年龄的逐年增加与RHOA发生的几率升高相关（OR 1.05, 95% CI 1.02 - 1.09）。基线BMI、糖尿病和吸烟习惯都与AD患者的RHOA无关。钳形人群包括1118髋，其中2.8%发生偶发RHOA。只有一种形状模式与入射RHOA相关（图1）。性别、基线年龄、BMI、糖尿病和吸烟习惯与钳形形态患者的RHOA无关。结论AD患者的形状和钳形形态的差异与RHOA的发生几率有关。在AD患者中，性别和基线年龄也与RHOA的发生有关。然而，在钳形形态的患者中没有观察到这种情况。这些发现可能为髋关节骨关节炎的个性化风险评估工具和预防策略的发展提供信息。
{"title":"BEYOND ACETABULAR DYSPLASIA AND PINCER MORPHOLOGY: REFINING HIP OSTEOARTHRITIS RISK ASSESSMENT THROUGH STATISTICAL SHAPE MODELING","authors":"F. Boel , M.A. van den Berg , N.S. Riedstra , M.M.A. van Buuren , J. Tang , H. Ahedi , N. Arden , S.M.A. Bierma-Zeinstra , C.G. Boer , F.M. Cicuttini , T.F. Cootes , K.M. Crossley , D.T. Felson , W.P. Gielis , J.J. Heerey , G. Jones , S. Kluzek , N.E. Lane , C. Lindner , J.A. Lynch , R. Agricola","doi":"10.1016/j.ostima.2025.100341","DOIUrl":"10.1016/j.ostima.2025.100341","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Hip morphology has been recognized as an important risk factor for the development of hip OA. In previous studies within the Worldwide Collaboration on OsteoArthritis prediCtion for the Hip consortium (World COACH), both acetabular dysplasia (AD) and pincer morphology–characterized by acetabular under- and overcoverage of the femoral head–were associated with the development of radiographic hip OA (RHOA) within 4-8 years, with an odds ratio (OR) of 1.80 (95% confidence interval (CI) 1.40-2.34) and 1.50 (95% CI 1.05-2.15), respectively. However, we know that not everyone with AD or pincer morphology will develop RHOA. Specific baseline characteristics or variations in hip shape among individuals with AD and pincer morphology may influence their risk of developing RHOA. Statistical shape models (SSM), describing the mean hip shape of a population and a range of independent shape variations, can be utilized to study these variations in hip shape.</div></div><div><h3>OBJECTIVE</h3><div>To evaluate whether specific hip shape variations or baseline characteristics within individuals with either AD or pincer morphology are associated with the development of RHOA within 4-8 years.</div></div><div><h3>METHODS</h3><div>We pooled individual participant data from seven prospective cohort studies within the World COACH consortium. Standardized anteroposterior (AP) pelvic radiographs were obtained at baseline and within 4-8 years follow-up. RHOA was scored by KLG or (modified) Croft grade. We harmonized the RHOA scores into “No OA” (KLG/Croft = 0), “doubtful OA” (KLG/Croft = 1), or “definite OA” (KLG/Croft ≥ 2 or total hip replacement). The Wiberg center edge angle (WCEA), measuring the weight-bearing femoral head coverage, and the lateral center edge angle (LCEA), measuring the bony femoral head coverage, were automatically determined using a validated method. Hips were included if they had baseline and follow-up RHOA scores, no RHOA at baseline, and either AD defined by a WCEA ≤ 25° or pincer morphology defined by a LCEA ≥45°. For both populations, an SSM was created of the acetabular roof, posterior wall, femoral head and neck, and teardrop (Fig 1). We analyzed the first 13 shape modes that explained around 90% of total shape variation in the population. The association between each shape mode, sex, baseline age, BMI, diabetes and smoking habits, and the development of RHOA was estimated using univariate generalized linear mixed-effects models. The mixed effects were added to account for the potential clustering within cohorts and participants. The results were expressed as ORs with 95% CIs.</div></div><div><h3>RESULTS</h3><div>The AD population consisted of 4,737 hips, of which 2.6% developed incident RHOA (Table 1). Four of the 13 shape modes (Fig 1) were associated with the development of RHOA. Additionally, in hips with AD, females had higher odds of incident RHOA than males (OR 2.85, 95% CI 1.46 – 5.58), and each year inc","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"5 ","pages":"Article 100341"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144522508","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2025-07-02DOI: 10.1016/j.ostima.2025.100336
K. Balaji , P.M. Vicente , S. Kukran , M. Mendoza , A.A. Bharath , P.J. Lally , N.K. Bangerter
<div><h3>INTRODUCTION</h3><div>Cartilage T<sub>2</sub> is a non-invasive MRI biomarker for KOA as it is sensitive to the underlying collagen hydration/organization. Cartilage microstructural changes seen in early KOA result in elevated T<sub>2</sub>. Cartilage T<sub>2</sub> maps could be used in DMOAD clinical trials.</div><div>Quantitative DESS (qDESS) simultaneously acquires 3D, morphological whole knee images and quantitative T<sub>2</sub> maps in ∼5 minutes. Recently, we developed RaFo-4 balanced Steady State Free Precession (RaFo-4 bSSFP) that also has the potential to simultaneously acquire 3D, morphological whole knee images with high SNR efficiency and quantitative cartilage T<sub>2</sub> maps in ∼5 minutes. RaFo-4 uses machine learning (Random Forest) to estimate voxel-level cartilage T<sub>2</sub> from bSSFP images. In this preliminary study, we compared qDESS and RaFo-4 bSSFP in morphological imaging and cartilage T<sub>2</sub> mapping.</div></div><div><h3>OBJECTIVE</h3><div>1) Which technique (qDESS or RaFo-4 bSSFP) has better test-retest repeatability of cartilage T<sub>2</sub> maps? 2) Which technique gives higher quality morphological images, as quantified using SNR of femoral, patellar, and tibial cartilage and CNR of cartilage-muscle, cartilage-synovial fluid, and synovial fluid-muscle?</div></div><div><h3>METHODS</h3><div>10 healthy volunteers (HVs: 7F, 3M, 20-40 age range) were scanned on a 3T Siemens Verio (Erlangen, Germany) using an 8-channel knee coil with ethics approval. Test-retest 3D (80 slices) sagittal knee images were acquired using qDESS (water excitation, 20° flip angle, 21.77 ms TR, 6 ms TE, 364 Hz/Px receiver bandwidth, 0 dummy scans per volume) and bSSFP (water excitation, 22° flip angle, 8.6 ms TR, 4.3 ms TE, 364 Hz/Px receiver bandwidth, 0 dummy scans per volume) for both knees of each HV with knee repositioning. qDESS and bSSFP were resolution- (0.4 × 0.4 × 1.5 mm<sup>3</sup> voxel volume, 150 × 150 × 120 mm<sup>3</sup> field of view) and scan time-matched (5:05 min. for qDESS and 5:04 min for bSSFP). 4 separate phase-cycled bSSFP images were acquired with phase cycling increments [0°, 90°, 180°, 270°]. Parallel imaging was used (GRAPPA R=2 for bSSFP and qDESS with 24 reference lines; 6/8<sup>th</sup> phase/slice partial Fourier for bSSFP). Cartilage in qDESS images was segmented using DOSMA and those segmentation masks were used on the bSSFP images. Test-retest repeatability was calculated using the ICC and coefficient of variation (CoV) after removing outlier T<sub>2</sub> estimates (T<sub>2</sub> < 20 ms, T<sub>2</sub> > 90 ms). The percentage of outlier estimates was also calculated. For quantitatively evaluating morphological image quality, SNR and CNR were calculated from the Root Sum of Squares (RSOS) of the two qDESS echos and four phase-cycled bSSFP images.</div></div><div><h3>RESULTS</h3><div>1) In Fig1, RaFo-4 preserves cartilage T<sub>2</sub> spatial variations seen in qDESS T<sub>2</sub> ma
{"title":"COMPARATIVE STUDY: QDESS VERSUS RAFO-4 PERFORMANCE IN 5-MINUTE, SIMULTANEOUS, RELIABLE 3D T2 MAPPING AND MORPHOLOGICAL MR IMAGING","authors":"K. Balaji , P.M. Vicente , S. Kukran , M. Mendoza , A.A. Bharath , P.J. Lally , N.K. Bangerter","doi":"10.1016/j.ostima.2025.100336","DOIUrl":"10.1016/j.ostima.2025.100336","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Cartilage T<sub>2</sub> is a non-invasive MRI biomarker for KOA as it is sensitive to the underlying collagen hydration/organization. Cartilage microstructural changes seen in early KOA result in elevated T<sub>2</sub>. Cartilage T<sub>2</sub> maps could be used in DMOAD clinical trials.</div><div>Quantitative DESS (qDESS) simultaneously acquires 3D, morphological whole knee images and quantitative T<sub>2</sub> maps in ∼5 minutes. Recently, we developed RaFo-4 balanced Steady State Free Precession (RaFo-4 bSSFP) that also has the potential to simultaneously acquire 3D, morphological whole knee images with high SNR efficiency and quantitative cartilage T<sub>2</sub> maps in ∼5 minutes. RaFo-4 uses machine learning (Random Forest) to estimate voxel-level cartilage T<sub>2</sub> from bSSFP images. In this preliminary study, we compared qDESS and RaFo-4 bSSFP in morphological imaging and cartilage T<sub>2</sub> mapping.</div></div><div><h3>OBJECTIVE</h3><div>1) Which technique (qDESS or RaFo-4 bSSFP) has better test-retest repeatability of cartilage T<sub>2</sub> maps? 2) Which technique gives higher quality morphological images, as quantified using SNR of femoral, patellar, and tibial cartilage and CNR of cartilage-muscle, cartilage-synovial fluid, and synovial fluid-muscle?</div></div><div><h3>METHODS</h3><div>10 healthy volunteers (HVs: 7F, 3M, 20-40 age range) were scanned on a 3T Siemens Verio (Erlangen, Germany) using an 8-channel knee coil with ethics approval. Test-retest 3D (80 slices) sagittal knee images were acquired using qDESS (water excitation, 20° flip angle, 21.77 ms TR, 6 ms TE, 364 Hz/Px receiver bandwidth, 0 dummy scans per volume) and bSSFP (water excitation, 22° flip angle, 8.6 ms TR, 4.3 ms TE, 364 Hz/Px receiver bandwidth, 0 dummy scans per volume) for both knees of each HV with knee repositioning. qDESS and bSSFP were resolution- (0.4 × 0.4 × 1.5 mm<sup>3</sup> voxel volume, 150 × 150 × 120 mm<sup>3</sup> field of view) and scan time-matched (5:05 min. for qDESS and 5:04 min for bSSFP). 4 separate phase-cycled bSSFP images were acquired with phase cycling increments [0°, 90°, 180°, 270°]. Parallel imaging was used (GRAPPA R=2 for bSSFP and qDESS with 24 reference lines; 6/8<sup>th</sup> phase/slice partial Fourier for bSSFP). Cartilage in qDESS images was segmented using DOSMA and those segmentation masks were used on the bSSFP images. Test-retest repeatability was calculated using the ICC and coefficient of variation (CoV) after removing outlier T<sub>2</sub> estimates (T<sub>2</sub> < 20 ms, T<sub>2</sub> > 90 ms). The percentage of outlier estimates was also calculated. For quantitatively evaluating morphological image quality, SNR and CNR were calculated from the Root Sum of Squares (RSOS) of the two qDESS echos and four phase-cycled bSSFP images.</div></div><div><h3>RESULTS</h3><div>1) In Fig1, RaFo-4 preserves cartilage T<sub>2</sub> spatial variations seen in qDESS T<sub>2</sub> ma","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"5 ","pages":"Article 100336"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144523629","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：