ERElka B. Rubin

Elka B. Rubin

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Elka B. Rubin has a total of 60 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Advanced MRI Approaches for Evaluating Common Lower Extremity Injuries in Basketball Players: Current and Emerging Techniques

    by Elka B. Rubin, Andrew M. Schmidt, Matthew F. Koff, Feliks Kogan, Kenneth Gao, Sharmila Majumdar, Hollis Potter, Garry E. Gold

    Magnetic resonance imaging (MRI) can provide accurate and non‐invasive diagnoses of lower extremity injuries in athletes. Sport‐related injuries commonly occur in and around the knee and can affect the articular cartilage, patellar tendon, hamstring muscles, and bone. Sports medicine physicians utilize MRI to evaluate and diagnose injury, track recovery, estimate return to sport timelines, and assess the risk of recurrent injury. This article reviews the current literature and describes novel developments of quantitative MRI tools that can further advance our understanding of sports injury diagnosis, prevention, and treatment while minimizing injury risk and rehabilitation time. Innovative approaches for enhancing the early diagnosis and treatment of musculoskeletal injuries in basketball players span a spectrum of techniques. These encompass the utilization of T2, T1ρ, and T2* quantitative MRI, along with dGEMRIC and Na‐MRI to assess articular cartilage injuries, 3D‐Ultrashort echo time MRI for patellar tendon injuries, diffusion tensor imaging for acute myotendinous injuries, and sagittal short tau inversion recovery and axial long‐axis T1‐weighted, and 3D Cube sequences for bone stress imaging. Future studies should further refine and validate these MR‐based quantitative techniques while exploring the lifelong cumulative impact of basketball on players' knees.Level of Evidence: 5Technical Efficacy: Stage 2


  • Open Access

    Improving Data-Efficiency and Robustness of Medical Imaging Segmentation Using Inpainting-Based Self-Supervised Learning

    by Jeffrey Dominic, Nandita Bhaskhar, Arjun D. Desai, Andrew Schmidt, Elka Rubin, Beliz Gunel, Garry E. Gold, Brian A. Hargreaves, Leon Lenchik, Robert Boutin, et al.

    We systematically evaluate the training methodology and efficacy of two inpainting-based pretext tasks of context prediction and context restoration for medical image segmentation using self-supervised learning (SSL). Multiple versions of self-supervised U-Net models were trained to segment MRI and CT datasets, each using a different combination of design choices and pretext tasks to determine the effect of these design choices on segmentation performance. The optimal design choices were used to train SSL models that were then compared with baseline supervised models for computing clinically-relevant metrics in label-limited scenarios. We observed that SSL pretraining with context restoration using 32 × 32 patches and Poission-disc sampling, transferring only the pretrained encoder weights, and fine-tuning immediately with an initial learning rate of 1 × 10−3 provided the most benefit over supervised learning for MRI and CT tissue segmentation accuracy (p < 0.001). For both datasets and most label-limited scenarios, scaling the size of unlabeled pretraining data resulted in improved segmentation performance. SSL models pretrained with this amount of data outperformed baseline supervised models in the computation of clinically-relevant metrics, especially when the performance of supervised learning was low. Our results demonstrate that SSL pretraining using inpainting-based pretext tasks can help increase the robustness of models in label-limited scenarios and reduce worst-case errors that occur with supervised learning.


  • Generalizability of Deep Learning Segmentation Algorithms for Automated Assessment of Cartilage Morphology and MRI Relaxometry

    by Andrew M. Schmidt, Arjun D. Desai, Lauren E. Watkins, Hollis A. Crowder, Marianne S. Black, Valentina Mazzoli, Elka B. Rubin, Quin Lu, James W. MacKay, Robert D. Boutin, et al.

    Background: Deep learning (DL)‐based automatic segmentation models can expedite manual segmentation yet require resource‐intensive fine‐tuning before deployment on new datasets. The generalizability of DL methods to new datasets without fine‐tuning is not well characterized.Purpose: Evaluate the generalizability of DL‐based models by deploying pretrained models on independent datasets varying by MR scanner, acquisition parameters, and subject population.Study Type: Retrospective based on prospectively acquired data.Population: Overall test dataset: 59 subjects (26 females); Study 1: 5 healthy subjects (zero females), Study 2: 8 healthy subjects (eight females), Study 3: 10 subjects with osteoarthritis (eight females), Study 4: 36 subjects with various knee pathology (10 females).Field Strength/Sequence: A 3‐T, quantitative double‐echo steady state (qDESS).Assessment: Four annotators manually segmented knee cartilage. Each reader segmented one of four qDESS datasets in the test dataset. Two DL models, one trained on qDESS data and another on Osteoarthritis Initiative (OAI)‐DESS data, were assessed. Manual and automatic segmentations were compared by quantifying variations in segmentation accuracy, volume, and T2 relaxation times for superficial and deep cartilage.Statistical Tests: Dice similarity coefficient (DSC) for segmentation accuracy. Lin's concordance correlation coefficient (CCC), Wilcoxon rank‐sum tests, root‐mean‐squared error‐coefficient‐of‐variation to quantify manual vs. automatic T2 and volume variations. Bland–Altman plots for manual vs. automatic T2 agreement. A P value < 0.05 was considered statistically significant.Results: DSCs for the qDESS‐trained model, 0.79–0.93, were higher than those for the OAI‐DESS‐trained model, 0.59–0.79. T2 and volume CCCs for the qDESS‐trained model, 0.75–0.98 and 0.47–0.95, were higher than respective CCCs for the OAI‐DESS‐trained model, 0.35–0.90 and 0.13–0.84. Bland–Altman 95% limits of agreement for superficial and deep cartilage T2 were lower for the qDESS‐trained model, ±2.4 msec and ±4.0 msec, than the OAI‐DESS‐trained model, ±4.4 msec and ±5.2 msec.Data Conclusion: The qDESS‐trained model may generalize well to independent qDESS datasets regardless of MR scanner, acquisition parameters, and subject population.Evidence Level: 1Technical Efficacy: Stage 1


  • Open Access

    SKM-TEA: A Dataset for Accelerated MRI Reconstruction with Dense Image Labels for Quantitative Clinical Evaluation

    by Arjun D Desai, Andrew M Schmidt, Elka B Rubin, Christopher M Sandino, Marianne S Black, Valentina Mazzoli, Kathryn J Stevens, Robert Boutin, Christopher Ré, Garry E Gold, et al.

    Magnetic resonance imaging (MRI) is a cornerstone of modern medical imaging. However, long image acquisition times, the need for qualitative expert analysis, and the lack of (and difficulty extracting) quantitative indicators that are sensitive to tissue health have curtailed widespread clinical and research studies. While recent machine learning methods for MRI reconstruction and analysis have shown promise for reducing this burden, these techniques are primarily validated with imperfect image quality metrics, which are discordant with clinically-relevant measures that ultimately hamper clinical deployment and clinician trust. To mitigate this challenge, we present the Stanford Knee MRI with Multi-Task Evaluation (SKM-TEA) dataset, a collection of quantitative knee MRI (qMRI) scans that enables end-to-end, clinically-relevant evaluation of MRI reconstruction and analysis tools. This 1.6TB dataset consists of raw-data measurements of ~25,000 slices (155 patients) of anonymized patient MRI scans, the corresponding scanner-generated DICOM images, manual segmentations of four tissues, and bounding box annotations for sixteen clinically relevant pathologies. We provide a framework for using qMRI parameter maps, along with image reconstructions and dense image labels, for measuring the quality of qMRI biomarker estimates extracted from MRI reconstruction, segmentation, and detection techniques. Finally, we use this framework to benchmark state-of-the-art baselines on this dataset. We hope our SKM-TEA dataset and code can enable a broad spectrum of research for modular image reconstruction and image analysis in a clinically informed manner. Dataset access, code, and benchmarks are available at https://github.com/StanfordMIMI/skm-tea.


  • Effects of the Competitive Season and Off‐Season on Knee Articular Cartilage in Collegiate Basketball Players Using Quantitative MRI: A Multicenter Study

    by Elka B. Rubin, Valentina Mazzoli, Marianne S. Black, Katherine Young, Arjun D. Desai, Matthew F. Koff, Ashwin Sreedhar, Feliks Kogan, Marc R. Safran, Dominic J. Vincentini, et al.

    Background: Injuries to the articular cartilage in the knee are common in jumping athletes, particularly high‐level basketball players. Unfortunately, these are often diagnosed at a late stage of the disease process, after tissue loss has already occurred.Purpose/Hypothesis: To evaluate longitudinal changes in knee articular cartilage and knee function in National Collegiate Athletic Association (NCAA) basketball players and their evolution over the competitive season and off‐season.Study Type: Longitudinal, multisite cohort study.Population: Thirty‐two NCAA Division 1 athletes: 22 basketball players and 10 swimmers.Field Strength/Sequence: Bilateral magnetic resonance imaging (MRI) using a combined T1ρ and T2 magnetization‐prepared angle‐modulated portioned k‐space spoiled gradient‐echo snapshots (MAPSS) sequence at 3T.Assessment: We calculated T2 and T1ρ relaxation times to compare compositional cartilage changes between three timepoints: preseason 1, postseason 1, and preseason 2. Knee Osteoarthritis Outcome Scores (KOOS) were used to assess knee health.Statistical Tests: One‐way variance model hypothesis test, general linear model, and chi‐squared test.Results: In the femoral articular cartilage of all athletes, we saw a global decrease in T2 and T1ρ relaxation times during the competitive season (all P < 0.05) and an increase in T2 and T1ρ relaxation times during the off‐season (all P < 0.05). In the basketball players' femoral cartilage, the anterior and central compartments respectively had the highest T2 and T1ρ relaxation times following the competitive season and off‐season. The basketball players had significantly lower KOOS measures in every domain compared with the swimmers: Pain (P < 0.05), Symptoms (P < 0.05), Function in Daily Living (P < 0.05), Function in Sport/Recreation (P < 0.05), and Quality of Life (P < 0.05).Conclusion: Our results indicate that T2 and T1ρ MRI can detect significant seasonal changes in the articular cartilage of basketball players and that there are regional differences in the articular cartilage that are indicative of basketball‐specific stress on the femoral cartilage. This study demonstrates the potential of quantitative MRI to monitor global and regional cartilage health in athletes at risk of developing cartilage problems.Level of Evidence: 2Technical Efficacy Stage: 2


  • Rapid volumetric gagCEST imaging of knee articular cartilage at 3 T: evaluation of improved dynamic range and an osteoarthritic population

    by Lauren E. Watkins, Elka B. Rubin, Valentina Mazzoli, Scott D. Uhlrich, Arjun D. Desai, Marianne Black, Gabe K. Ho, Scott L. Delp, Marc E. Levenston, Gary S. Beaupré, et al.

    Chemical exchange saturation transfer of glycosaminoglycans, gagCEST, is a quantitative MR technique that has potential for assessing cartilage proteoglycan content at field strengths of 7 T and higher. However, its utility at 3 T remains unclear. The objective of this work was to implement a rapid volumetric gagCEST sequence with higher gagCEST asymmetry at 3 T to evaluate its sensitivity to osteoarthritic changes in knee articular cartilage and in comparison with T2 and T1ρ measures. We hypothesize that gagCEST asymmetry at 3 T decreases with increasing severity of osteoarthritis (OA). Forty‐two human volunteers, including 10 healthy subjects and 32 subjects with medial OA, were included in the study. Knee Injury and Osteoarthritis Outcome Scores (KOOS) were assessed for all subjects, and Kellgren‐Lawrence grading was performed for OA volunteers. Healthy subjects were scanned consecutively at 3 T to assess the repeatability of the volumetric gagCEST sequence at 3 T. For healthy and OA subjects, gagCEST asymmetry and T2 and T1ρ relaxation times were calculated for the femoral articular cartilage to assess sensitivity to OA severity. Volumetric gagCEST imaging had higher gagCEST asymmetry than single‐slice acquisitions (p = 0.015). The average scan‐rescan coefficient of variation was 6.8%. There were no significant differences in average gagCEST asymmetry between younger and older healthy controls (p = 0.655) or between healthy controls and OA subjects (p = 0.310). T2 and T1ρ relaxation times were elevated in OA subjects (p < 0.001 for both) compared with healthy controls and both were moderately correlated with total KOOS scores (rho = −0.181 and rho = −0.332 respectively). The gagCEST technique developed here, with volumetric scan times under 10 min and high gagCEST asymmetry at 3 T, did not vary significantly between healthy subjects and those with mild‐moderate OA. This further supports a limited utility for gagCEST imaging at 3 T for assessment of early changes in cartilage composition in OA.


  • Does Injection of Hyaluronic Acid Protect Against Early Cartilage Injury Seen After Marathon Running? A Randomized Controlled Trial Utilizing High-Field Magnetic Resonance Imaging

    by Amit Nathani, Garry E. Gold, Uchechukwuka Monu, Brian Hargreaves, Andrea K. Finlay, Elka B. Rubin, Marc R. Safran

    Background: Previous studies have shown that runners demonstrate elevated T2 and T1ρ values on magnetic resonance imaging (MRI) after running a marathon, with the greatest changes in the patellofemoral and medial compartment, which can persist after 3 months of reduced activity. Additionally, marathon running has been shown to increase serum inflammatory markers. Hyaluronic acid (HA) purportedly improves viscoelasticity of synovial fluid, serving as a lubricant while also having chondroprotective and anti-inflammatory effects. Purpose/Hypothesis: The purpose was to investigate whether intra-articular HA injection can protect articular cartilage from injury attributed to marathon running. The hypothesis was that the addition of intra-articular HA 1 week before running a marathon would reduce the magnitude of early cartilage breakdown measured by MRI. Study Design: Randomized controlled trial; Level of evidence, 2. Methods: After institutional review board approval, 20 runners were randomized into receiving an intra-articular injection of HA or normal saline (NS) 1 week before running a marathon. Exclusionary criteria included any prior knee injury or surgery and having run >3 prior marathons. Baseline 3-T knee MRI was obtained within 48 hours before the marathon (approximately 5 days after injection). Follow-up 3-T MRI scans of the same knee were obtained 48 to 72 hours and 3 months after the marathon. The T2 and T1ρ relaxation times of articular cartilage were measured in 8 locations—the medial and lateral compartments (including 2 areas of each femoral condyle) and the patellofemoral joint. The statistical analysis compared changes in T2 and T1ρ relaxation times (ms) from baseline to immediate and 3-month postmarathon scans between the HA and NS groups with repeated measures analysis of variance. Results: Fifteen runners completed the study: 6 women and 2 men in the HA group (mean age, 31 years; range, 23-50 years) and 6 women and 1 man in the NS group (mean age, 27 years; range, 20-49 years). There were no gross morphologic MRI changes after running the marathon. Postmarathon studies revealed no statistically significant changes between the HA and NS groups in all articular cartilage areas of the knee on both T2 and T1ρ relaxation times. Conclusion: Increased T2 and T1ρ relaxation times have been observed in marathon runners, suggesting early cartilage injury. The addition of intra-articular HA did not significantly affect relaxation times in all areas of the knee when compared with an NS control.


  • Gait retraining as a conservative treatment for medial knee osteoarthritis

    by V. Mazzoli, S. Uhlrich, E. Rubin, F. Kogan, B. Heargraves, S. Delp, G.S. Beaupre, G.E. Gold


  • Assessing the Effectiveness of Evidence-Based Medicine in Practice

    by Carolyn A. Hutyra, Benjamin Streufert, Cary S. Politzer, Perez Agaba, Elka Rubin, Lori A. Orlando, Dean C. Taylor, Richard C. Mather

    Background: The dissemination of evidence-based information into medical practice is essential to provide patients with optimal care and realize society’s substantial investments in medical research. Effective information delivery and treatment utilization may lead to improvements in patient outcome, reductions in cost, and an overall lower burden on the health-care system. This study examines the dissemination of medical evidence following a first-time anterior shoulder dislocation (FTASD) and assesses the impact of potential dissemination strategies.Methods: The state of evidence dissemination into clinical practice for FTASD was evaluated with use of the Reach, Effectiveness, Adoption, Implementation, and Maintenance (RE-AIM) framework. The treatment pathway for patients with FTASDs was mapped and evaluated using data that were collected through an orthopaedic shoulder-specialist survey and with review of a claims database.Results: A total of 1,755 patients with an FTASD were identified through a national claims database; 50% of patients followed up with a care provider within 30 days after an emergency department (ED) or urgent care visit. Based on shoulder-specialist survey data, physician estimates of the risk of redislocation within a 2-year window aligned with medical evidence 59% of the time. Only 29% of patients obtained information for FTASD that aligns with high-level medical evidence.Conclusions: There are gaps and deficiencies in the dissemination and application of evidence in the treatment of FTASDs. Specifically, patients have limited exposure to health-care encounters where appropriate information related to low rates of follow-up following ED or urgent care visits may be communicated. Evaluating the current state of practice and identifying areas of improvement for the dissemination of evidence regarding FTASDs can be achieved through application of the RE-AIM framework. Greater consideration and resourcing of dissemination and implementation strategies may improve the dissemination and the impact of existing medical evidence.