Osteoarthritis imaging最新文献

{"title":"ADAPT-SEG: A MODALITY-AGNOSTIC PIPELINE FOR ROBUST PATIENT-SPECIFIC MSK SEGMENTATION","authors":"L.G. Johnson,&nbsp;S.J. Shefelbine","doi":"10.1016/j.ostima.2026.100430","DOIUrl":"10.1016/j.ostima.2026.100430","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Accurate segmentation of musculoskeletal MRI is a critical step in patient-specific musculoskeletal modeling and quantitative analysis of cartilage. However, segmentation remains a time-consuming bottleneck limiting the feasibility of large-scale studies. While machine learning tools can automate segmentation in large, homogeneous datasets, they often struggle to generalize to unfamiliar image types, poor-quality images, or extreme joint pathology. Augmenting machine learning segmentation with modality-independent statistical shape models (SSMs) could improve generalization by constraining outputs to fit reasonable shape priors, thereby reducing the manual segmentation burden in small and diverse datasets.</div></div><div><h3>OBJECTIVE</h3><div>We hypothesize that augmenting deep learning segmentation (nnUNet) with modality-independent shape priors will improve segmentation accuracy in unfamiliar MRI sequences compared with nnUNet alone.</div></div><div><h3>METHODS</h3><div>We developed ADAPT-seg (Agnostic Deep-learning and Prior-guided Tissue Segmentation), a three-step pipeline consisting of: 1) mutual-information based registration to a reference image, 2) initial segmentation boundary estimation with Active Shape Modeling (ASM), and 3) refinement with nnUNet (Figure 1). Using labeled DESS-sequence knee images from the OAI-ZIB dataset (n=40, KL grade 0), we created a 3D SSM of the femur and tibia and an nnUNet “fullres-cascade” two-stage segmentation model (Figure 1). Five additional KLG 0 participants with DESS, FLASH and TSE-sequence images comprised the test set. All test images were segmented using nnUNet alone and the full ADAPT-seg pipeline, with ASM-derived labels serving as the low-resolution stage. Performance against manual segmentations was evaluated in a region of interest (ROI) covering the key articulating anatomy using Dice scores, average surface distances, and boundary F1 scores.</div></div><div><h3>RESULTS</h3><div>ADAPT-seg demonstrated greatly improved segmentation accuracy on unfamiliar FLASH and TSE sequences in all metrics and performed comparably to nnUNet on DESS images (Table 1, Figure 2). nnUNet failed to correctly label the tibia in all TSE and FLASH images, a catastrophic failure prevented by SSM-based initialization. Both nnUNet and ADAPT-seg added tibia labels to the patella in all FLASH scans.</div></div><div><h3>CONCLUSION</h3><div>SSM-based augmentation is a feasible strategy to improve generalization of nnUNet segmentation across unfamiliar MRI sequences. ADAPT-seg remains under active development, with the aim of extending the pipeline to cover other joints, more extreme pathomorphology, and multiple tissue types including articular cartilage. Our long-term goal is to develop ADAPT-seg into an open-source tool which, used alongside expert review and correction, can greatly reduce the manual segmentation burden for MSK researchers.</div></div>","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"6 ","pages":"Article 100430"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148543770","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"WALKING MECHANICS DIFFERS BETWEEN ASYMPTOMATIC INDIVIDUALS WITH OR WITHOUT MEDIAL KNEE FULL-THICKNESS CARTILAGE LOSS: DATA FROM THE LAUSANNE KNEE STUDY (LKS)","authors":"LC Pereira,&nbsp;SM Shokouhyan,&nbsp;P Margain,&nbsp;J Runhaar,&nbsp;S Bierma-Zeinstra,&nbsp;BM Jolles,&nbsp;P Omoumi,&nbsp;J Favre","doi":"10.1016/j.ostima.2026.100440","DOIUrl":"10.1016/j.ostima.2026.100440","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Walking mechanics is an important factor of knee osteoarthritis, but its relationship with structural lesions and symptoms remains unclear. Studying asymptomatic individuals with MRI-defined full-thickness cartilage loss (FTCL) may help isolate gait alterations associated with structural disease rather than pain.</div></div><div><h3>OBJECTIVE</h3><div>To compare walking mechanics between asymptomatic individuals without FTCL at the knee and those with FTCL in the medial femorotibial compartment.</div></div><div><h3>METHODS</h3><div>A randomly selected knee for 204 asymptomatic participants of the Lausanne Knee Study was classified as medial FTCL (n = 16; 7 females; median age 59 years) or no FTCL (n = 188; 113 females; median age 56 years) based on a magnetic resonance exam. All participants underwent a three-dimensional gait analysis using a marker-based motion capture system. Walking kinetics and kinematics were calculated using a previously validated method. Parameters were compared between groups using Mann–Whitney U tests. Effect size was measured using the rank-biserial correlation (r).</div></div><div><h3>RESULTS</h3><div>A consistent and coherent pattern of statistically significant differences of small effect size (0.004 ≤ p ≤0.043; 0.14 ≤ r ≤ 0.20) was observed between groups. During early stance, the medial FTCL group exhibited a larger knee flexion moment, a larger ankle dorsiflexion angle, and a lower fore ground reaction force. During midstance, the knee adduction moment was larger in the medial FTCL group. During late stance, the medial FTCL group showed larger knee and hip flexed angle, a lower hip extension moment, and a lower vertical ground reaction force.</div></div><div><h3>CONCLUSION</h3><div>This study highlighted a pattern of differences between asymptomatic individuals with medial FTCL and no FTCL. This original result on asymptomatic extends prior literature on symptomatic OA, enhancing our overall understanding of the tripartite interplay among gait alterations, symptoms, and structural lesions.</div></div>","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"6 ","pages":"Article 100440"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148557876","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"OPPORTUNISTIC POPLITEAL ARTERY ASSESSMENT ON KNEE MRI AND MEDIAL TIBIAL STRUCTURAL DAMAGE IN OSTEOARTHRITIS: FNIH DATA","authors":"P. Mirghaderi,&nbsp;N. Eshraghi,&nbsp;F. Pishgar,&nbsp;N. Balu,&nbsp;C. Yuan,&nbsp;H.A. Chansky,&nbsp;D.J. Hunter,&nbsp;F.W. Roemer,&nbsp;A. Guermazi,&nbsp;M. Chalian","doi":"10.1016/j.ostima.2026.100437","DOIUrl":"10.1016/j.ostima.2026.100437","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Vascular pathology has been proposed as a contributor to knee OA structural damage, but evidence remains inconsistent. Popliteal artery wall biomarkers extracted opportunistically from knee MRI may help characterize vascular-related structural phenotypes.</div></div><div><h3>OBJECTIVE</h3><div>To evaluate whether baseline MRI-derived popliteal artery wall biomarkers are associated with baseline severity and 24-month worsening of medial tibial cartilage loss and bone marrow lesions (BMLs) in the FNIH Osteoarthritis Biomarkers Consortium.</div></div><div><h3>METHODS</h3><div>This analysis included 600 FNIH participants, with one knee per participant. Baseline sagittal 3D DESS knee MRI was processed using FRAPPE, a fully automated popliteal artery wall segmentation tool, to extract eccentricity ratio (maximum-to-minimum wall thickness ratio), normalized wall index (wall area divided by total vessel area), wall area, and lumen area. Baseline and 24-month MRI were assessed by two musculoskeletal radiologists using MOAKS. Linear regression models evaluated associations with medial tibial cartilage damage and BML scores and 24-month worsening, adjusting for age, sex, BMI, and FNIH subgroup.</div></div><div><h3>RESULTS</h3><div>Mean age was 61.6 ± 8.9 years, mean BMI was 30.7 ± 4.8 kg/m², and 353/600 participants (58.8%) were female. Higher eccentricity ratio was associated with greater baseline medial tibial cartilage loss, including more involved subregions (β=0.44, 95% CI 0.09-0.79; p&lt;0.05), higher summed full-thickness damage MOAKS scores (β=0.49, 95% CI 0.18-0.80; p&lt;0.01), and higher summed surface-area damage MOAKS scores (β=0.88, 95% CI 0.16-1.60; p&lt;0.05). Higher eccentricity ratio was also associated with greater medial tibial BML burden, including more involved subregions (β=0.40, 95% CI 0.10-0.70; p&lt;0.01) and higher summed BML size MOAKS scores (β=0.67, 95% CI 0.16-1.17; p&lt;0.01). Popliteal artery wall biomarkers were not associated with 24-month worsening of cartilage loss or BMLs.</div></div><div><h3>CONCLUSION</h3><div>Popliteal artery wall eccentricity on knee MRI was associated with greater baseline medial tibial cartilage loss and BML severity, but not with 24-month worsening. Opportunistic vascular assessment on knee MRI may help characterize vascular-related baseline structural disease burden in knee OA.</div></div>","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"6 ","pages":"Article 100437"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148557874","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"MINING THIGH MUSCLE INFORMATION FROM MRI USING ARTIFICIAL INTELLIGENCE TO PREDICT INCIDENT KNEE OSTEOARTHRITIS","authors":"Q Dang,&nbsp;H Wang,&nbsp;Z Wang,&nbsp;J Ma,&nbsp;C Ding","doi":"10.1016/j.ostima.2026.100441","DOIUrl":"10.1016/j.ostima.2026.100441","url":null,"abstract":"<div><h3>BACKGROUND</h3><div>Knee osteoarthritis (OA) is the most common chronic joint disease with a substantial public health burden, and early diagnosis of knee OA is crucial for timely intervention and effective disease management. Thigh muscle changes play an important role in knee OA pathogenesis and are potential targets for intervention, which makes thigh muscles a promising target for developing predictive models. Here we developed MKOTIR (muscle-oriented knee osteoarthritis tailored incidence risk), a fully automated artificial intelligence (AI) model based on thigh MRI to predict incident knee OA over 8 years.</div></div><div><h3>METHODS</h3><div>A total of 4070 thigh MRI scans were collected from five clinical centers, constituting a training set, a validation set, an internal testing set, and two external testing sets. MKOTIR processes original MRI scans and directly outputs the predicted probability of developing knee OA over 8 years. Model performance was evaluated using the area under the curve (AUC), accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and F1 score. A clinical model and a radiomics model were also constructed as benchmarks for comparison. To enhance interpretability, we correlated MKOTIR with key clinical characteristics and classical radiomics features. Furthermore, the independent prognostic value of MKOTIR was evaluated in the clinical context.</div></div><div><h3>RESULTS</h3><div>MKOTIR outperformed the benchmark models and demonstrated excellent predictive performance in two external testing sets, with AUC values of 0.977 and 0.965, accuracy of 0.982 and 0.982, sensitivity of 0.911 and 0.902, specificity of 0.991 and 0.997, PPV of 0.935 and 0.982, and NPV of 0.988 and 0.982, respectively. The F1 scores were 0.923 and 0.940, further supporting its robust and well-balanced performance. MKOTIR prediction was significantly associated with some clinical characteristics, confirming its clinical relevance, and was independent of potential confounders. Although MKOTIR prediction correlated with certain radiomics features, the radiomics model alone failed to achieve comparable performance, indicating that MKOTIR learned additional imaging patterns beyond those captured by the conventional radiomics approach. Moreover, the performance of MKOTIR remained robust in subgroup analyses.</div></div><div><h3>CONCLUSION</h3><div>The application of AI to thigh MRI enables mining muscle information for predicting incident knee OA over 8 years. Our developed MKOTIR model demonstrates excellent performance and therefore holds promise for the early initiation of muscle-targeted therapies.</div></div>","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"6 ","pages":"Article 100441"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148557875","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DIFFERENTIAL DIAGNOSIS OF HAND ARTHRITIS ON RADIOGRAPHS USING DEEP LEARNING","authors":"J. Eder,&nbsp;M. Kellner,&nbsp;C. Watzenboeck,&nbsp;V. Ovsianik,&nbsp;R. Janizcek,&nbsp;M. Deman,&nbsp;D. Sieghart,&nbsp;G. Langs,&nbsp;P. Seeboeck,&nbsp;W. Wirth,&nbsp;D. Aletaha,&nbsp;P. Mandl","doi":"10.1016/j.ostima.2026.100401","DOIUrl":"10.1016/j.ostima.2026.100401","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Radiographic differential diagnosis of hand arthritides remains challenging, particularly when findings overlap across rheumatoid arthritis (RA), psoriatic arthritis (PsA), and hand osteoarthritis (HOA), a common situation in older patients. Deep learning may support image-based classification, but clinical datasets are often limited, imbalanced, and affected by cohort-related heterogeneity.</div></div><div><h3>OBJECTIVE</h3><div>To present initial work-in-progress results of a deep learning-based four-class classification approach distinguishing RA, PsA, HOA, and no documented radiographic changes (NDRC) on hand radiographs from the Medical University of Vienna (MUW) and the Osteoarthritis Initiative (OAI).</div></div><div><h3>METHODS</h3><div>This retrospective feasibility study included 5,404 unilateral hand radiographs from MUW and OAI, split at the patient level into 4,814 training, 201 validation, and 389 independent test radiographs (Table 1). Target classes were rheumatoid arthritis (RA), psoriatic arthritis (PsA), hand osteoarthritis (HOA), and no documented radiographic changes (NDRC), with NDRC indicating absence of documented arthritic changes rather than clinically verified healthy controls. MUW and OAI cases were pooled for HOA and NDRC, while RA and PsA radiographs originated from MUW. Images were classified using an ImageNet-pretrained ResNet-18. Performance on the independent test set was assessed using accuracy, balanced accuracy, macro and weighted F1-scores, Cohen’s kappa, and macro-averaged multiclass ROC-AUC.</div></div><div><h3>RESULTS</h3><div>In the mixed four-class setting, the model achieved an accuracy of 0.59, balanced accuracy of 0.59, macro F1-score of 0.58, weighted F1-score of 0.58, Cohen’s kappa of 0.45, and macro-averaged multiclass ROC-AUC of 0.84. Per-class performance is shown in Table 2, with recall ranging from 0.38 to 0.79 and F1-scores from 0.41 to 0.80. The relatively high ROC-AUC compared with moderate accuracy and F1-scores suggests that the model captured discriminative ranking information, whereas final single-label four-class assignment remained challenging in this heterogeneous multi-source setting.</div></div><div><h3>CONCLUSION</h3><div>These initial findings suggest that deep learning can capture disease-related signal in heterogeneous hand radiograph datasets, but robust differential diagnosis of RA, PsA, and HOA, as well as separation from NDRC, remains challenging. A key limitation is that NDRC reflects absence of documented radiographic changes in the source data rather than clinically verified healthy controls. Future work will focus on expanding the dataset with more balanced multi-centre data, including radiographs from healthy individuals, and improving robustness to cohort-related domain shift.</div></div>","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"6 ","pages":"Article 100401"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148557881","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ASSOCIATION OF SYNOVITIS ACTIVITY MEASURED BY SYNOVIAL TISSUE VOLUME (STV) AND DYNAMIC CONTRAST-ENHANCED (DCE)-MRI IN PATIENTS WITH INFLAMMATORY KNEE OSTEOARTHRITIS","authors":"A. Brett,&nbsp;N. Maguire,&nbsp;J. Burlison,&nbsp;D. Laurent,&nbsp;F. Saxer,&nbsp;K. Stevenson,&nbsp;K. Meiser,&nbsp;J. Riek,&nbsp;F. Roemer,&nbsp;A. Guermazi,&nbsp;M. Schieker","doi":"10.1016/j.ostima.2026.100407","DOIUrl":"10.1016/j.ostima.2026.100407","url":null,"abstract":"&lt;div&gt;&lt;h3&gt;INTRODUCTION&lt;/h3&gt;&lt;div&gt;Osteoarthritis (OA) is a complex condition influenced by genetic predisposition, mechanical stress and/or inflammatory processes. Synovitis, which is present in more than half of all OA patients, predicts more severe cartilage damage, faster disease progression and has been associated with pain. The quantitative measurement of synovitis using MR imaging requires the use of contrast enhancement (CE) to distinguish it from effusion in the joint. Quantitative measurements of synovitis based on T1w CE sequences include synovial tissue volume (STV), which represents the overall degree of whole joint synovitis or by dynamic contrast-enhanced (DCE)-MRI, which reflects tissue perfusion. However, DCE-MRI is more complex and requires power-injectors, limiting sites that may participate in a multicentre trial.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;OBJECTIVE&lt;/h3&gt;&lt;div&gt;Our aim was to compare measurements of STV and Ktrans, a marker of synovitis via a measure of capillary permeability derived from DCE-MRI, in a cohort of patients with inflammatory knee OA from a Phase II study (NCT04886258).&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;METHODS&lt;/h3&gt;&lt;div&gt;Patients were aged 50 to 80 with at least moderate synovitis at screening by 11-site semi-quantitative scoring of T1w CE 3T MRI of the entire knee. Synovitis activity was assessed by acquiring T1-weighted images covering the entire knee at intervals of a few seconds, before and after administration of a contrast agent. To segment the synovial tissue, active appearance models (AAMs) were fitted to the bones in each image together with a 3D synovial region of interest (ROI) mask that encompassed the joint while excluding adjacent tissues and blood vessels. Ktrans was computed within this ROI and calculated for each voxel where the data fit the Tofts model with sufficient accuracy (R² ≥ 0.6). The median Ktrans value was used for comparison to avoid outliers. The ROI was then used together with the pre-contrast and last of the post-contrast low-resolution DCE-MRI images to compute STV as the volume of voxels in the post contrast ROI-masked image enhanced at a greyscale intensity more than the mean plus 3SD of the pre-contrast foreground intensity determined by Otsu thresholding (Figure 1). Fractional volume of STV was calculated as the enhancing volume divided by the entire ROI volume.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;RESULTS&lt;/h3&gt;&lt;div&gt;106 image sets at baseline and 90 at 12 weeks were available for DCE-MRI and STV comparison. 87 paired baseline-12week image sets were available for difference comparison. Comparison of all baseline and 12-week median Ktrans and fractional STV results showed strong correlation (r = 0.70 and 0.69), with minimal zero Ktrans offset (0.0016 and 0.0028min-1), see Figure 1. Comparison of baseline-12-week changes showed very strong correlation (r = 0.79) again with minimal zero Ktrans offset (-0.0011 min-1), see Figure 2.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;CONCLUSION&lt;/h3&gt;&lt;div&gt;Although it may be expected that physiologic measures of ","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"6 ","pages":"Article 100407"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148543640","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"AUTOMATED BML SEGMENTATION USING UNET FOR TOTAL KNEE REPLACEMENT PREDICTION","authors":"T. Hetali,&nbsp;Z. Ming,&nbsp;S. Khaing,&nbsp;R. Zender,&nbsp;V. Khushi,&nbsp;W. Kevin,&nbsp;S. Juan","doi":"10.1016/j.ostima.2026.100429","DOIUrl":"10.1016/j.ostima.2026.100429","url":null,"abstract":"&lt;div&gt;&lt;h3&gt;INTRODUCTION&lt;/h3&gt;&lt;div&gt;Knee osteoarthritis (OA) is a leading cause of disability worldwide and often progresses to total knee replacement (TKR). Bone marrow lesions (BMLs), visible as hyperintense regions on MRI, are strong imaging predictors of TKR progression. Although manual BML segmentation is considered the gold standard, it is highly time-consuming and impractical for large-scale clinical use. Deep learning-based segmentation offers a scalable alternative, but it remains unclear whether automatically extracted BML features provide the same predictive value as expert annotations.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;OBJECTIVE&lt;/h3&gt;&lt;div&gt;This study evaluates the clinical feasibility of using a UNet-based segmentation model for automated BML quantification and investigates whether UNet-derived volumetric features can effectively replace manual annotations for TKR prediction.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;METHODS&lt;/h3&gt;&lt;div&gt;Two datasets from the Osteoarthritis Initiative (OAI) were used: Dataset 1 included 300 patients with manual annotations for training, validation, and testing, while Dataset 2 included 1,393 patients for external testing. CLAHE preprocessing was applied to enhance MRI contrast. Separate UNet models were trained for femur, tibia, and patella segmentation using a combined binary cross-entropy and Dice loss. Predicted masks were used to extract 2D and 3D BML volumetric features from medial and lateral compartments. These automated features were compared with manual features using six classifiers: Logistic Regression, SVM, Random Forest, CNN, XGBoost, and Decision Tree. Segmentation performance was evaluated using Dice scores and Pearson volume correlation, while classification performance was assessed using AUC, F1-score, and accuracy.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;RESULTS&lt;/h3&gt;&lt;div&gt;On Dataset 1, UNet achieved Dice scores of 0.8203 (femur), 0.7387 (tibia), and 0.8264 (patella), with strong Pearson volume correlations above 0.83. On the external Dataset 2, 2D Dice remained robust at 0.7713 (femur), 0.7235 (tibia), and 0.7817 (patella), with Pearson volume correlations all exceeding 0.70, demonstrating cross-dataset generalizability. In TKR prediction, automated UNet-derived features matched or outperformed manual annotations across multiple classifiers. Random Forest achieved the best performance with an AUC of 0.950 using automated features, compared to 0.866 using manual features. SVM also improved from 0.906 to 0.948 AUC when using automated features.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;CONCLUSION&lt;/h3&gt;&lt;div&gt;UNet-derived BML volumetric features provide predictive performance comparable to, and in some cases better than, expert manual annotations for TKR prediction. The strong volumetric agreement across datasets suggests that automated segmentation captures clinically meaningful lesion information, even when boundary-level overlap is moderate. These findings support the use of automated UNet-based pipelines as a scalable alternative to manual BML quantification for MRI-based T","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"6 ","pages":"Article 100429"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148543683","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"IDENTIFICATION OF MACHINE-INFORMED REGIONAL HOTSPOTS WITHIN MAGNETIC RESONANCE IMAGES ASSOCIATED WITH KNEE OSTEOARTHRITIS OUTCOMES","authors":"C.Q.Y. Lin,&nbsp;O. Saarela,&nbsp;C. McIntosh,&nbsp;A.K.O. Wong","doi":"10.1016/j.ostima.2026.100433","DOIUrl":"10.1016/j.ostima.2026.100433","url":null,"abstract":"&lt;div&gt;&lt;h3&gt;INTRODUCTION&lt;/h3&gt;&lt;div&gt;It is difficult to identify the origin of knee OA as it is a heterogeneous disease, with the subchondral bone being a relatively neglected tissue. The current bottom-up hypothesis-driven process is costly, slow, and sometimes results in failure of a biomarker that scientists have spent many years refining. Therefore, employing a top-down approach to randomly select as many regions of interests and associated imaging features as possible to converge on potential hotspots exhibiting multiple clinical correlations could enhance our ability to pinpoint high priority symptom-correlated regions for future hypothesis-testing poised for success.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;OBJECTIVE&lt;/h3&gt;&lt;div&gt;Discover subchondral volumes of interest (VOI) and radiomic knee MRI features that represent underlying physiological causes of knee OA symptoms by applying high-dimensional methods.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;METHODS&lt;/h3&gt;&lt;div&gt;This study used MRIs and clinical data of 2804 women and 1992 men (n=4796), 45-79 years old from the OAI. Images were acquired in double-bullseye orientation using coronal IM FSE MRI at 3T, 0mm gap, with voxel size of 0.365 × 0.456 × 3.0 mm&lt;sup&gt;3&lt;/sup&gt;. A 3D anatomical MRI atlas of the knee bone structure was generated by combining MRI scans for each KLG (0-4), and sex (male and female) until the mean absolute change in voxel difference reached a plateau. Five watershed configurations (maximum settings of 10, 100, 250, 500, and 750 VOIs per bone) were applied to the atlas to define a large array of contrast-guided VOIs within the femur and tibia. The watershed-labeled atlas was co-registered to each participant’s knee MRI and radiomic features for each VOI was extracted using PyRadiomics. Multilinear principal component analysis was used to reduce dimensionality of the 107 2D matrix-structured radiomics data, yielding latent factor scores, an approach previously applied with brain functional MRI. Radiomic latent factor scores were related to Knee Osteoarthritis Outcome Score (KOOS) subscales measured at 10 timepoints across 10 years. Associations were modelled in a random-effects linear regression for knee-level subscales, accounting for age, sex, body mass index, KLG, physical activity, income, education level, alcohol use, smoking, comorbidities, depression, hormone replacement, and glucocorticoid use. Standardized estimates per watershed VOI, and per set of radiomic latent factor scores were mapped back to the knee atlas and overlaid to compute mean absolute effect estimates representing parametric maps per timepoint. Hotspots were defined as regions with the highest average absolute effect, indicating significant associations with multiple outcomes.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;RESULTS&lt;/h3&gt;&lt;div&gt;For the KOOS symptoms and pain subscales, the timepoint with the largest overall mean effect size was at baseline. Hotspots for both subscales were localized to the mid-anterior aspect of the medial tibial condyle just inferior to the subchond","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"6 ","pages":"Article 100433"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148543689","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ACCELERATED ANKLE MRI PROTOCOL USING COMBINED VARIABLE DENSITY UNDERSAMPLING OF K-SPACE WITH ECHO SPACING OPTIMIZATION le Density","authors":"D Ferreira Branco,&nbsp;M Alzaher,&nbsp;FJ Simeone,&nbsp;M Jarraya,&nbsp;M Fung,&nbsp;A Gaddipati,&nbsp;S Mandava,&nbsp;D Rettmann,&nbsp;E Milshteyn,&nbsp;JS Husseini","doi":"10.1016/j.ostima.2026.100422","DOIUrl":"10.1016/j.ostima.2026.100422","url":null,"abstract":"&lt;div&gt;&lt;h3&gt;PURPOSE&lt;/h3&gt;&lt;div&gt;To compare image quality and acquisition times for a commercial deep-learning reconstruction technique with a new technique that combines a variable density undersampling of k-space with echo spacing optimization for 3-fold acceleration in ankle MR.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;METHODS&lt;/h3&gt;&lt;div&gt;In this prospective study, 28 patients underwent ankle MR exams as part of standard clinical care using our institution’s routine protocol on a GE Premier 3T magnet with a protocol employing a commercialized DL Recon algorithm (AIR™ Recon DL, GE HealthCare), which is based on a Convolutional Neural Network (CNN) trained to remove ringing, reduce noise, and increase sharpness. Each patient underwent an additional 3-fold accelerated MRI exam which combined 2 features: a vendor-supplied prototype known as Sonic DLTM 2D, which is an acquisition and reconstruction technique that combines highly accelerated variable density undersampled data acquisition with an unrolled DL-based reconstruction and echo spacing optimization, which allows for longer echo train lengths (ETL) and SAR reduction. Commercial AIR Recon DL images were compared to the accelerated images. Reviewers evaluated lateral ankle ligaments, as well as tibiotalar articular cartilage and subchondral bone for pathology. Reviewers also performed a subjective evaluation of image quality, assessment of main pathology, and diagnostic confidence in commercial and accelerated images. Image-quality Likert scores were compared between protocols using Wilcoxon signed-rank tests; inter-reader agreement was quantified with Gwet's AC2 for image quality and diagnostic confidence, and with weighted Cohen's κ for lesion assessment.&lt;/div&gt;&lt;/div&gt;&lt;div&gt;&lt;h3&gt;RESULTS&lt;/h3&gt;&lt;div&gt;Mean acquisition time was 705.5 ± 91.5 s for the standard protocol versus 471.1 ± 52.1 s for the accelerated protocol (33% reduction). Image-quality Likert scores did not differ between protocols for any reader (Wilcoxon signed-rank, p = 0.21, 0.10, 0.65). Inter-reader agreement (Gwet's AC2) for image-quality was excellent in both protocols (standard 0.90 [95% CI 0.81–0.99]; accelerated 0.83 [0.73–0.93]); diagnostic confidence agreement was good to excellent (0.79 [0.67–0.92] vs 0.77 [0.62–0.92] respectively). For diagnostic lesion assessment, inter-reader agreement evaluated using Cohen’s kappa was consistently good to excellent across all structures. For standard sequences, kappa values ranged from 0.72 (95% CI: 0.54–0.90) for calcaneofibular ligament (CFL) injury to 0.93 (CI: 0.84–1.00) for subchondral edema, with anterior talofibular ligament (ATFL) injury (κ = 0.84 CI: 0.71–0.97), cartilage (κ = 0.86 CI: 0.70–1.00), and subchondral cyst (κ = 0.89, CI: 0.76–1.00) also demonstrating high agreement. For accelerated sequences, kappa values ranged from 0.80 (CI: 0.65–0.95) for CFL to 0.97 (CI: 0.89–1.00) for subchondral cysts, with ATFL (κ = 0.87, CI: 0.75–0.98), cartilage (κ = 0.87, CI: 0.71–1.00), and subchondral edema (κ = 0.93, CI","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"6 ","pages":"Article 100422"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148543765","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"LONGITUDINAL SENSITIVITY TO CHANGE AND PROGNOSTIC VALIDITY OF AUTOMATICALLY MEASURED EFFUSION SYNOVITIS VOLUME (ESV)","authors":"F. Eckstein,&nbsp;K. Sheikhi Valashani,&nbsp;T. Winkler,&nbsp;J. Collins,&nbsp;D.J. Hunter,&nbsp;F.W. Roemer,&nbsp;A. Guermazi,&nbsp;W. Wirth","doi":"10.1016/j.ostima.2026.100409","DOIUrl":"10.1016/j.ostima.2026.100409","url":null,"abstract":"<div><h3>INTRODUCTION</h3><div>Albeit not specific to inflammation only, effusion-synovitis (ES) - assessed using non-contrast enhanced MRI - is used as a proxy of synovial activation in knee OA. ES is defined as hyperintensity within the synovial cavity, and is commonly graded using established scoring systems, based on the amount of capsular distension. Quantitative approaches to measure the amount of ES volume (ESV) have been proposed and permit quantitative, precise, and longitudinal evaluation of ES using a continuous measure. We have previously developed a fully automated AI-based technology for measuring ESV and demonstrated criterion validity vs. manual segmentation. The FNIH1 Biomarker Consortium Study is ideally suited to evaluate ESV as a baseline or longitudinal predictor of progression.</div></div><div><h3>OBJECTIVE</h3><div>To test the longitudinal sensitivity to change of ESV in progressor vs. non-progressor knees, and to test its baseline, follow-up (FU), and longitudinal prognostic validity.</div></div><div><h3>METHODS</h3><div>The analyses were performed in the FNIH-1 Biomarker Consortium Cohort: 194 knees exhibited combined radiographic and pain progression at 2, 3, or 4 years [Y] FU [ComP], 103 radiographic only [RoP], 103 pain progression only [PoP], and 200 neither type [NoP]. Fully automated segmentation of peripatellar ESV was carried out on baseline, Y1, and Y2 axial reformats of DESS MRI. An AI-based convolutional neural network (CNN) model was trained with 77 manually segmented knees. We evaluated the rate (µl and %) and sensitivity to change (SRM) of ESV in the four groups. The strength of the association was examined using odds ratios (ORs), and the effect size of the between (progressor) group differences by the Cohen’s d.</div></div><div><h3>RESULTS</h3><div>The ESV increased by 38% over 2Y in ComP (SRM: 0.46), by 18% in RoP (0.19), and by 9% in PoP (0.15). In contrast, it diminished by 11% in NoP (-0.20) (Table 1). Between ComP and NoPs, the ESV differed by 1013 µl at baseline, by 3010 µl at Y1, and by 4417 µl at Y2. The ORs for the ESV in ComP vs. NPs were 1.15 (95% CI 0.94, 1.40) at baseline, 1.64 (95% CI 1.33, 2.03) at Y1, and 1.96 (95% CI 1.60, 2.39) at Y2 (Table 1). Fig. 1 shows the Cohen’s d values throughout groups, time points, or observation periods.</div></div><div><h3>CONCLUSION</h3><div>The ESV - determined from non-contrast enhanced DESS MRI - was sensitive to longitudinal change (an increase) in knees with OA disease progression. The current results should be confirmed using standard clinical MRI sequences (IW TSE FS) and in other cohorts, ideally in direct comparison with contrast enhanced MRI. Yet, the current analysis is promising in that the ESV yielded strong prognostic validity and differentiated well between knees with different types of progression, as well as between (combined) progressors vs. non-progressors.</div></div>","PeriodicalId":74378,"journal":{"name":"Osteoarthritis imaging","volume":"6 ","pages":"Article 100409"},"PeriodicalIF":0.0,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148543595","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}