pith. sign in

arxiv: 2501.13376 · v3 · submitted 2025-01-23 · 📡 eess.IV · cs.CV

Clinical utility of foundation models in musculoskeletal MRI for biomarker fidelity and predictive outcomes

Gabrielle Hoyer , Michelle W Tong , Rupsa Bhattacharjee , Valentina Pedoia , Sharmila Majumdar This is my paper

Pith reviewed 2026-05-23 05:41 UTC · model grok-4.3

classification 📡 eess.IV cs.CV
keywords musculoskeletal MRIfoundation modelsimage segmentationquantitative biomarkersknee osteoarthritisclinical triagepredictive modeling
0
0 comments X

The pith

Fine-tuned foundation models produce clinically reliable quantitative biomarkers from routine musculoskeletal MRI that support both reduced-workload triage and long-term outcome prediction.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows that promptable foundation models, after fine-tuning on varied musculoskeletal MRI datasets and paired with automated prompting, generate segmentations whose derived measurements of cartilage, bone, and soft tissue match expert annotations at levels suitable for clinical use. These same measurements then feed a three-stage knee triage process that lowers the volume of cases needing human review while preserving detection sensitivity, and they also train 48-month landmark models that predict knee replacement and incident osteoarthritis with reported calibration and decision-curve net benefit. A reader would care because the approach converts everyday scans into standardized, scalable data without requiring exhaustive new expert labeling for every site. The architecture is presented as model-agnostic and open-source so that others can replicate or extend the measurement-to-decision pipeline.

Core claim

Fine-tuned versions of SAM, SAM2, and MedSAM, driven by automated detection for prompting, yield segmentations whose extracted biomarkers achieve high concordance with expert annotations across cartilage, bone, and soft-tissue structures; the same biomarker set powers a three-stage knee triage cascade that reduces verification workload while retaining sensitivity and enables 48-month landmark models that forecast knee replacement and incident osteoarthritis with favorable calibration and net benefit across clinically relevant thresholds.

What carries the argument

Fine-tuned promptable foundation segmenters (SAM, SAM2, MedSAM) coupled to automated detection for fully automatic prompting, which convert raw MRI into standardized quantitative biomarkers.

If this is right

  • Automated biomarker extraction can be inserted into existing radiology workflows to cut the fraction of knee MRI cases requiring full expert review.
  • The same biomarker values can be used directly as inputs to risk-stratification models that identify patients likely to need knee replacement within four years.
  • The modular, model-agnostic design allows substitution of newer foundation models without rebuilding the downstream triage or prediction stages.
  • Open-source release of the architecture permits independent sites to reproduce the concordance and calibration results on their own data.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the biomarkers remain stable across scanner vendors, they could serve as standardized endpoints in multi-center osteoarthritis trials.
  • The triage cascade logic could be adapted to other joints or to non-knee musculoskeletal conditions once similar fine-tuned models exist.
  • Longer follow-up beyond 48 months might reveal whether the same landmark features also predict slower-progressing disease trajectories.

Load-bearing premise

The heterogeneous musculoskeletal datasets used for fine-tuning and the downstream prediction models are assumed to be representative of real-world clinical populations and acquisition protocols.

What would settle it

A substantial drop in segmentation concordance or in prediction calibration when the same pipeline is applied to an independent, multi-site clinical MRI collection acquired under different protocols or scanners.

Figures

Figures reproduced from arXiv: 2501.13376 by Gabrielle Hoyer, Michelle W Tong, Rupsa Bhattacharjee, Sharmila Majumdar, Valentina Pedoia.

Figure 1
Figure 1. Figure 1: Musculoskeletal MRI segmentation study design. a) [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Summary of dataset composition, subject demographics, and imaging pro [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Baseline performance comparison of SAM, MedSAM, and SAM2 across [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Evaluation of finetuning strategies for SAM models in musculoskeletal MRI [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Analysis of MRI acquisition parameters and segmentation agreement for [PITH_FULL_IMAGE:figures/full_fig_p014_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Comparative evaluation of segmentation models using ground truth and [PITH_FULL_IMAGE:figures/full_fig_p016_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Downstream clinical tasks enabled by the AutoLabel [PITH_FULL_IMAGE:figures/full_fig_p017_7.png] view at source ↗
read the original abstract

Precision medicine in musculoskeletal imaging requires scalable measurement infrastructure. We developed a modular system that converts routine MRI into standardized quantitative biomarkers suitable for clinical decision support. Promptable foundation segmenters (SAM, SAM2, MedSAM) were fine-tuned across heterogeneous musculoskeletal datasets and coupled to automated detection for fully automatic prompting. Fine-tuned segmentations yielded clinically reliable measurements with high concordance to expert annotations across cartilage, bone, and soft tissue biomarkers. Using the same measurements, we demonstrate two applications: (i) a three-stage knee triage cascade that reduces verification workload while maintaining sensitivity, and (ii) 48-month landmark models that forecast knee replacement and incident osteoarthritis with favorable calibration and net benefit across clinically relevant thresholds. Our model-agnostic, open-source architecture enables independent validation and development. This work validates a pathway from automated measurement to clinical decision: reliable biomarkers drive both workload optimization today and patient risk stratification tomorrow, and the developed framework shows how foundation models can be operationalized within precision medicine systems.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript presents a modular system converting routine musculoskeletal MRI into quantitative biomarkers via fine-tuned promptable foundation segmenters (SAM, SAM2, MedSAM) on heterogeneous datasets, coupled with automated detection. It claims clinically reliable measurements with high concordance to expert annotations for cartilage, bone, and soft tissue biomarkers, and demonstrates two applications: a three-stage knee triage cascade reducing verification workload while maintaining sensitivity, and 48-month landmark models forecasting knee replacement and incident osteoarthritis with favorable calibration and net benefit. The architecture is model-agnostic and open-source.

Significance. If the reported concordance, calibration, and net benefit hold under external validation, the work could provide a practical pathway for scalable biomarker extraction to support clinical decision-making in musculoskeletal imaging, including workload reduction and risk stratification. The open-source framework is a strength for enabling independent validation and extension.

major comments (2)
  1. [Abstract] Abstract: the central claims of 'high concordance to expert annotations' and 'favorable calibration and net benefit' are stated without any quantitative metrics (e.g., Dice/IoU scores, ICC values, AUC, calibration slopes, or decision-curve net benefit numbers), error bars, or exclusion criteria. This prevents evaluation of whether the data support the assertions of clinical reliability and predictive utility.
  2. [Methods (datasets/fine-tuning)] Methods section on datasets and fine-tuning: no description is provided of how dataset heterogeneity (scanner vendors, field strengths, acquisition parameters, demographics) was quantified or sampled, nor is there an external validation cohort or ablation under distribution shift. The generalization assumption required for the triage cascade and 48-month models to translate to real-world populations therefore remains untested.
minor comments (2)
  1. [Abstract] Abstract: the phrase 'clinically reliable measurements' is used without an explicit definition or threshold (e.g., minimum ICC or Dice value) that would allow readers to judge the claim.
  2. [Abstract/Methods] The manuscript states the architecture is 'open-source' but does not provide a repository link or license details in the abstract or methods.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive comments. We address each major point below and indicate where revisions have been made.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claims of 'high concordance to expert annotations' and 'favorable calibration and net benefit' are stated without any quantitative metrics (e.g., Dice/IoU scores, ICC values, AUC, calibration slopes, or decision-curve net benefit numbers), error bars, or exclusion criteria. This prevents evaluation of whether the data support the assertions of clinical reliability and predictive utility.

    Authors: We agree that the abstract should contain quantitative support for the stated claims. The revised abstract now reports key metrics including mean Dice scores for segmentations, ICC values for biomarker concordance with experts, AUC values for the 48-month prediction models, calibration slopes, and decision-curve net benefit at clinically relevant thresholds, each accompanied by 95% confidence intervals and a statement of exclusion criteria. revision: yes

  2. Referee: [Methods (datasets/fine-tuning)] Methods section on datasets and fine-tuning: no description is provided of how dataset heterogeneity (scanner vendors, field strengths, acquisition parameters, demographics) was quantified or sampled, nor is there an external validation cohort or ablation under distribution shift. The generalization assumption required for the triage cascade and 48-month models to translate to real-world populations therefore remains untested.

    Authors: We have expanded the Methods section to quantify and report dataset heterogeneity, including explicit breakdowns by scanner vendor, field strength, acquisition parameters, and demographic variables, together with the sampling approach used across the heterogeneous collections. The study was designed around internal validation on these heterogeneous datasets rather than an external cohort or dedicated distribution-shift ablations; we have added an explicit limitations paragraph noting that external validation remains necessary to confirm generalizability to broader real-world populations. revision: partial

standing simulated objections not resolved
  • Absence of an external validation cohort and ablation experiments under distribution shift, as these analyses were not part of the original study design.

Circularity Check

0 steps flagged

No significant circularity detected; derivation chain is self-contained

full rationale

No load-bearing steps reduce by construction to inputs. The pipeline proceeds from fine-tuning foundation models on heterogeneous datasets to producing biomarker measurements, then applies those measurements to separate triage and 48-month outcome models. Concordance to expert annotations and calibration to clinical endpoints are presented as empirical results, not tautological fits. No equations, self-citations, or uniqueness theorems are invoked that would collapse the predictions into the training data. The central claims rest on falsifiable external benchmarks (expert annotations, knee replacement incidence) rather than internal redefinitions.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

Review performed on abstract only; full methods, data, and equations unavailable, so ledger entries are limited to those directly implied by the abstract text.

axioms (2)
  • domain assumption Fine-tuned foundation models will produce segmentations with high concordance to expert annotations on heterogeneous musculoskeletal MRI datasets.
    This premise underpins the claim of clinically reliable biomarkers.
  • domain assumption The derived biomarkers are sufficiently stable and informative to support both triage workload reduction and 48-month outcome prediction with favorable calibration.
    Required for the two downstream clinical applications to be valid.

pith-pipeline@v0.9.0 · 5715 in / 1366 out tokens · 21768 ms · 2026-05-23T05:41:01.166691+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

69 extracted references · 69 canonical work pages · 7 internal anchors

  1. [1]

    van der Laak, Bram van Ginneken, and Clara I

    G. Litjens, T. Kooi, B. E. Bejnordi, A. A. A. Setio, F. Ciompi, M. Ghafoorian, J. A. Van Der Laak, B. Van Ginneken, C. I. Sánchez, A survey on deep learning in medical image analysis, Medical Image Analysis 42 (2017) 60–88.doi:10.1016/j.media.2017.07.005

    work page doi:10.1016/j.media.2017.07.005 2017

  2. [2]

    S. K. Zhou, H. Greenspan, C. Davatzikos, J. S. Duncan, B. Van Ginneken, A. Madabhushi, J. L. Prince, D. Rueckert, R. M. Summers, A Review of Deep Learning in Medical Imaging: Imaging Traits, Technology Trends, Case Studies With Progress Highlights, and Future Promises, Proceedings of the IEEE 109 (5) (2021) 820–838.doi:10.1109/JPROC.2021.3054390

    work page doi:10.1109/jproc.2021.3054390 2021

  3. [3]

    U-Net: Convolutional Networks for Biomedical Image Segmentation

    O. Ronneberger, P. Fischer, T. Brox, U-Net: Convolutional Networks for Biomedical Image Segmenta- tion, version Number: 1 (2015).doi:10.48550/ARXIV.1505.04597

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1505.04597 2015

  4. [4]

    Milletari, N

    F. Milletari, N. Navab, S.-A. Ahmadi, V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation, in: 2016 Fourth International Conference on 3D Vision (3DV), 2016, pp. 565–571.doi:10.1109/3DV.2016.79

    work page doi:10.1109/3dv.2016.79 2016

  5. [5]

    D. Jha, P. H. Smedsrud, M. A. Riegler, D. Johansen, T. de Lange, P. Halvorsen, H. D. Johansen, ResUNet++: An Advanced Architecture for Medical Image Segmentation, version Number: 1 (2019). doi:10.48550/ARXIV.1911.07067

    work page doi:10.48550/arxiv.1911.07067 2019

  6. [6]

    Attention U-Net: Learning Where to Look for the Pancreas

    O. Oktay, J. Schlemper, L. L. Folgoc, M. Lee, M. Heinrich, K. Misawa, K. Mori, S. McDonagh, N. Y. Hammerla, B. Kainz, B. Glocker, D. Rueckert, Attention U-Net: Learning Where to Look for the Pancreas, version Number: 3 (2018).doi:10.48550/ARXIV.1804.03999

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1804.03999 2018

  7. [7]

    Isensee, P

    F. Isensee, P. F. Jaeger, S. A. A. Kohl, J. Petersen, K. H. Maier-Hein, nnU-Net: a self-configuring method for deep learning-based biomedical image segmentation, Nature Methods 18 (2) (2021) 203–

    work page 2021

  8. [8]

    doi:10.1038/s41592-020-01008-z

    work page doi:10.1038/s41592-020-01008-z

  9. [9]

    H. Cao, Y. Wang, J. Chen, D. Jiang, X. Zhang, Q. Tian, M. Wang, Swin-Unet: Unet-like Pure Transformer for Medical Image Segmentation, version Number: 1 (2021).doi:10.48550/ARXIV.2105. 05537

    work page doi:10.48550/arxiv.2105 2021

  10. [10]

    , author Jeyaseelan, L

    N. Tajbakhsh, L. Jeyaseelan, Q. Li, J. N. Chiang, Z. Wu, X. Ding, Embracing imperfect datasets: A review of deep learning solutions for medical image segmentation, Medical Image Analysis 63 (2020) 101693. doi:10.1016/j.media.2020.101693

    work page doi:10.1016/j.media.2020.101693 2020

  11. [11]

    W. Yan, L. Huang, L. Xia, S. Gu, F. Yan, Y. Wang, Q. Tao, MRI Manufacturer Shift and Adaptation: Increasing the Generalizability of Deep Learning Segmentation for MR Images Acquired with Different Scanners, Radiology: Artificial Intelligence 2 (4) (2020) e190195.doi:10.1148/ryai.2020190195

    work page doi:10.1148/ryai.2020190195 2020

  12. [12]

    H. Guan, M. Liu, Domain Adaptation for Medical Image Analysis: A Survey, IEEE Transactions on Biomedical Engineering 69 (3) (2022) 1173–1185.doi:10.1109/TBME.2021.3117407

    work page doi:10.1109/tbme.2021.3117407 2022

  13. [13]

    R. F. Loeser, S. R. Goldring, C. R. Scanzello, M. B. Goldring, Osteoarthritis: A disease of the joint as an organ, Arthritis & Rheumatism 64 (6) (2012) 1697–1707.doi:10.1002/art.34453

    work page doi:10.1002/art.34453 2012

  14. [14]

    A. J. Cruz-Jentoft, G. Bahat, J. Bauer, Y. Boirie, O. Bruyère, T. Cederholm, C. Cooper, F. Landi, Y. Rolland, A. A. Sayer, others, Sarcopenia: revised European consensus on definition and diagnosis, Age and ageing 48 (1) (2019) 16–31, publisher: Oxford University Press

    work page 2019

  15. [15]

    A. J. Haig, Paraspinal denervation and the spinal degenerative cascade, The Spine Journal 2 (5) (2002) 372–380.doi:10.1016/S1529-9430(02)00201-2

    work page doi:10.1016/s1529-9430(02)00201-2 2002

  16. [16]

    Segment Anything

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.-Y. Lo, P. Dollár, R. Girshick, Segment Anything, version Number: 1 (2023).doi:10.48550/ ARXIV.2304.02643

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  17. [17]

    N. Ravi, V. Gabeur, Y.-T. Hu, R. Hu, C. Ryali, T. Ma, H. Khedr, R. Rädle, C. Rolland, L. Gustafson, E. Mintun, J. Pan, K. V. Alwala, N. Carion, C.-Y. Wu, R. Girshick, P. Dollár, C. Feichtenhofer, SAM 2: Segment Anything in Images and Videos, arXiv preprint arXiv:2408.00714 (2024). URL https://arxiv.org/abs/2408.00714

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  18. [18]

    J. Ma, Y. He, F. Li, L. Han, C. You, B. Wang, Segment anything in medical images, Nature Commu- nications 15 (1) (2024) 654.doi:10.1038/s41467-024-44824-z

    work page doi:10.1038/s41467-024-44824-z 2024

  19. [19]

    M. A. Mazurowski, H. Dong, H. Gu, J. Yang, N. Konz, Y. Zhang, Segment anything model for medical 54 image analysis: An experimental study, Medical Image Analysis 89 (2023) 102918.doi:10.1016/j. media.2023.102918

    work page doi:10.1016/j 2023

  20. [20]

    S. He, R. Bao, J. Li, J. Stout, A. Bjornerud, P. E. Grant, Y. Ou, Computer-Vision Benchmark Segment- Anything Model (SAM) in Medical Images: Accuracy in 12 Datasets, version Number: 3 (2023). doi:10.48550/ARXIV.2304.09324

    work page doi:10.48550/arxiv.2304.09324 2023

  21. [21]

    Huang, X

    Y. Huang, X. Yang, L. Liu, H. Zhou, A. Chang, X. Zhou, R. Chen, J. Yu, J. Chen, C. Chen, S. Liu, H. Chi, X. Hu, K. Yue, L. Li, V. Grau, D.-P. Fan, F. Dong, D. Ni, Segment Anything Model for Medical Images?Publisher: arXiv Version Number: 7 (2023).doi:10.48550/ARXIV.2304.14660

    work page doi:10.48550/arxiv.2304.14660 2023

  22. [22]

    C. Chen, J. Miao, D. Wu, Z. Yan, S. Kim, J. Hu, A. Zhong, Z. Liu, L. Sun, X. Li, T. Liu, P.-A. Heng, Q. Li, MA-SAM: Modality-agnostic SAM Adaptation for 3D Medical Image Segmentation, version Number: 1 (2023). doi:10.48550/ARXIV.2309.08842

    work page doi:10.48550/arxiv.2309.08842 2023

  23. [23]

    arXiv preprint arXiv:2304.13785 (2023)

    K. Zhang, D. Liu, Customized Segment Anything Model for Medical Image Segmentation, version Number: 2 (2023). doi:10.48550/ARXIV.2304.13785

    work page doi:10.48550/arxiv.2304.13785 2023

  24. [24]

    H. Dong, H. Gu, Y. Chen, J. Yang, Y. Chen, M. A. Mazurowski, Segment anything model 2: an application to 2D and 3D medical images, version Number: 3 (2024). doi:10.48550/ARXIV.2408. 00756

    work page doi:10.48550/arxiv.2408 2024

  25. [25]

    J. Ma, S. Kim, F. Li, M. Baharoon, R. Asakereh, H. Lyu, B. Wang, Segment anything in medical images and videos: Benchmark and deployment (2024).arXiv:2408.03322. URL https://arxiv.org/abs/2408.03322

    work page arXiv 2024

  26. [26]

    Müller, I

    D. Müller, I. Soto-Rey, F. Kramer, Towards a guideline for evaluation metrics in medical image seg- mentation, BMC Research Notes 15 (1) (2022) 210.doi:10.1186/s13104-022-06096-y

    work page doi:10.1186/s13104-022-06096-y 2022

  27. [27]

    Hirling, E

    D. Hirling, E. Tasnadi, J. Caicedo, M. V. Caroprese, R. Sjögren, M. Aubreville, K. Koos, P. Horvath, Segmentation metric misinterpretations in bioimage analysis, Nature Methods (Jul. 2023).doi:10. 1038/s41592-023-01942-8

    work page 2023

  28. [28]

    The osteoarthritis initiative: report on the design rationale for the magnetic resonance imaging protocol for the knee 16 (12).doi:10.1016/j.joca.2008.06.016

    work page doi:10.1016/j.joca.2008.06.016 2008

  29. [29]

    Van Houwelingen, H

    H. Van Houwelingen, H. Putter, Dynamic Prediction in Clinical Survival Analysis, 0th Edition, CRC Press, 2011. doi:10.1201/b11311

    work page doi:10.1201/b11311 2011

  30. [30]

    Hoyer, K

    G. Hoyer, K. T. Gao, F. G. Gassert, J. Luitjens, F. Jiang, S. Majumdar, V. Pedoia, Foundations of a knee joint digital twin from qMRI biomarkers for osteoarthritis and knee replacement, npj Digital Medicine 8 (1), publisher: Springer Science and Business Media LLC (Feb. 2025). doi:10.1038/ s41746-025-01507-3

    work page 2025

  31. [31]

    A. J. Vickers, E. B. Elkin, Decision Curve Analysis: A Novel Method for Evaluating Prediction Models, Medical Decision Making 26 (6) (2006) 565–574, publisher: SAGE Publications. doi: 10.1177/0272989x06295361

    work page doi:10.1177/0272989x06295361 2006

  32. [32]

    Platt, Probabilistic outputs for support vector machines and comparisons to regularized likelihood methods 10 (3) (1999) 61–74

    J. Platt, Probabilistic outputs for support vector machines and comparisons to regularized likelihood methods 10 (3) (1999) 61–74

    work page 1999

  33. [33]

    A. A. Tolpadi, U. Bharadwaj, K. T. Gao, R. Bhattacharjee, F. G. Gassert, J. Luitjens, P. Giesler, J. N. Morshuis, P. Fischer, M. Hein, C. F. Baumgartner, A. Razumov, D. Dylov, Q. V. Lohuizen, S. J. Fransen, X. Zhang, R. Tibrewala, H. L. De Moura, K. Liu, M. V. W. Zibetti, R. Regatte, S. Majumdar, V. Pedoia, K2S Challenge: From Undersampled K-Space to Auto...

    work page doi:10.3390/bioengineering10020267 2023

  34. [34]

    Pedoia, C

    V. Pedoia, C. Russell, A. Randolph, X. Li, S. Majumdar, A.-A. Consortium, Principal component analysis-T1ρ voxel based relaxometry of the articular cartilage: a comparison of biochemical patterns in osteoarthritis and anterior cruciate ligament subjects, Quantitative Imaging in Medicine and Surgery 6 (6) (2016) 623–633.doi:10.21037/qims.2016.11.03

    work page doi:10.21037/qims.2016.11.03 2016

  35. [35]

    rep., Stryker Imorphics (2017)

    White Paper: Imorphics OA Knee MRI Measurements., Tech. rep., Stryker Imorphics (2017). URL www.imorphics.com

    work page 2017

  36. [36]

    M. Hess, B. Allaire, K. T. Gao, R. Tibrewala, G. Inamdar, U. Bharadwaj, C. Chin, V. Pedoia, M. Boux- sein, D. Anderson, S. Majumdar, Deep Learning for Multi-Tissue Segmentation and Fully Automatic Personalized Biomechanical Models from BACPAC Clinical Lumbar Spine MRI, Pain Medicine 24 (Sup- 55 plement_1) (2023) S139–S148.doi:10.1093/pm/pnac142

    work page doi:10.1093/pm/pnac142 2023

  37. [37]

    Thahakoya, Evaluating the relationship of proximal bone shape asymmetry with cartilage health and biomechanics in patients with hip OA., Toronto, Canada, 2023

    R. Thahakoya, Evaluating the relationship of proximal bone shape asymmetry with cartilage health and biomechanics in patients with hip OA., Toronto, Canada, 2023

    work page 2023

  38. [38]

    Gallo, C

    M. Gallo, C. Wyatt, V. Pedoia, D. Kumar, S. Lee, L. Nardo, T. Link, R. Souza, S. Majumdar, T1ρ and T2 relaxation times are associated with progression of hip osteoarthritis, Osteoarthritis and Cartilage 24 (8) (2016) 1399–1407.doi:10.1016/j.joca.2016.03.005

    work page doi:10.1016/j.joca.2016.03.005 2016

  39. [39]

    S. Lee, R. M. Lucas, D. A. Lansdown, L. Nardo, A. Lai, T. M. Link, R. Krug, C. B. Ma, Magnetic reso- nance rotator cuff fat fraction and its relationship with tendon tear severity and subject characteristics, Journal of Shoulder and Elbow Surgery 24 (9) (2015) 1442–1451.doi:10.1016/j.jse.2015.01.013

    work page doi:10.1016/j.jse.2015.01.013 2015

  40. [40]

    Nardo, D

    L. Nardo, D. C. Karampinos, D. A. Lansdown, J. Carballido-Gamio, S. Lee, R. Maroldi, C. B. Ma, T. M. Link, R. Krug, Quantitative assessment of fat infiltration in the rotator cuff muscles using water- fat MRI: Fat Infiltration in the Rotator Cuff Muscles, Journal of Magnetic Resonance Imaging 39 (5) (2014) 1178–1185.doi:10.1002/jmri.24278

    work page doi:10.1002/jmri.24278 2014

  41. [41]

    T. Baum, S. Inhuber, M. Dieckmeyer, C. Cordes, S. Ruschke, E. Klupp, P. M. Jungmann, R. Farlock, H. Eggers, H. Kooijman, E. J. Rummeny, A. Schwirtz, J. S. Kirschke, D. C. Karampinos, Association of Quadriceps Muscle Fat With Isometric Strength Measurements in Healthy Males Using Chemical Shift Encoding-Based Water-Fat Magnetic Resonance Imaging:, Journal ...

    work page doi:10.1097/rct.0000000000000374 2016

  42. [42]

    Bhattacharjee, K

    R. Bhattacharjee, K. Roach, Z. Akkaya, P. Giesler, R. Souza, V. Pedoia, S. Majumdar, Exploring Bilateral Thigh Normalized-Lean-Muscle And Fat Volume Associations With Knee Cartilage Thickness And Functional Parameters In Radiographic Hip Oa Patients, Osteoarthritis and Cartilage 31 (2023) S110–S111.doi:10.1016/j.joca.2023.01.062

    work page doi:10.1016/j.joca.2023.01.062 2023

  43. [43]

    M. J. Davison, M. R. Maly, P. J. Keir, S. M. Hapuhennedige, A. T. Kron, J. D. Adachi, K. A. Beattie, Lean muscle volume of the thigh has a stronger relationship with muscle power than muscle strength in women withkneeosteoarthritis, ClinicalBiomechanics 41(2017)92–97. doi:10.1016/j.clinbiomech. 2016.11.005

    work page doi:10.1016/j.clinbiomech 2017

  44. [44]

    Brett, C

    M. Brett, C. J. Markiewicz, M. Hanke, M.-A. Côté, B. Cipollini, P. McCarthy, D. Jarecka, C. P. Cheng, Y. O. Halchenko, M. Cottaar, E. Larson, S. Ghosh, D. Wassermann, S. Gerhard, G. R. Lee, H.-T. Wang, E. Kastman, J. Kaczmarzyk, R. Guidotti, O. Duek, J. Daniel, A. Rokem, C. Madison, B. Moloney, F. C. Morency, M. Goncalves, R. Markello, C. Riddell, C. Burn...

    work page doi:10.5281/zenodo.4295521 2020

  45. [45]

    B. C. Lowekamp, D. T. Chen, L. Ibáñez, D. Blezek, The Design of SimpleITK, Frontiers in Neuroin- formatics 7 (2013). doi:10.3389/fninf.2013.00045

    work page doi:10.3389/fninf.2013.00045 2013

  46. [46]

    Mason, Scaramallion, Mrbean-Bremen, Rhaxton, J

    D. Mason, Scaramallion, Mrbean-Bremen, Rhaxton, J. Suever, Vanessasaurus, D. P. Orfanos, G. Lemaitre, A. Panchal, A. Rothberg, M. D. Herrmann, J. Massich, J. Kerns, Korijn Van Golen, T. Ro- bitaille, S. Biggs, Moloney, C. Bridge, M. Shun-Shin, B. Conrad, Pawelzajdel, M. Mattes, Y. Lyu, F. C. Morency, Z. Baratz, T. Cogan, B. P. Sánchez, C. Clauss, H. Meine...

    work page doi:10.5281/zenodo.8034250 2023

  47. [47]

    Decoupled Weight Decay Regularization

    I. Loshchilov, F. Hutter, Decoupled Weight Decay Regularization, version Number: 3 (2017).doi: 10.48550/ARXIV.1711.05101

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1711.05101 2017

  48. [48]

    SGDR: Stochastic Gradient Descent with Warm Restarts

    I. Loshchilov, F. Hutter, SGDR: Stochastic Gradient Descent with Warm Restarts, version Number: 5 (2016). doi:10.48550/ARXIV.1608.03983

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1608.03983 2016

  49. [49]

    Nickolls, I

    J. Nickolls, I. Buck, M. Garland, K. Skadron, Scalable Parallel Programming with CUDA: Is CUDA 56 the parallel programming model that application developers have been waiting for?, Queue 6 (2) (2008) 40–53.doi:10.1145/1365490.1365500

    work page doi:10.1145/1365490.1365500 2008

  50. [50]

    Weiss, V

    F.Pedregosa, G.Varoquaux, A.Gramfort, V.Michel, B.Thirion, O.Grisel, M.Blondel, P.Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, E. Duchesnay, Scikit-learn: Machine Learning in Python, Journal of Machine Learning Research 12 (2011) 2825–2830

    work page 2011

  51. [51]

    Iriondo, F

    C. Iriondo, F. Liu, F. Calivà, S. Kamat, S. Majumdar, V. Pedoia, Towards understanding mechanis- tic subgroups of osteoarthritis: 8-year cartilage thickness trajectory analysis, Journal of Orthopaedic Research 39 (6) (2021) 1305–1317.doi:10.1002/jor.24849

    work page doi:10.1002/jor.24849 2021

  52. [52]

    Cummings, K

    J. Cummings, K. Gao, V. Chen, A. Morales Martinez, C. Iriondo, F. Caliva, S. Majumdar, V. Pedoia, The knee connectome: A novel tool for studying spatiotemporal change in cartilage thickness, Journal of Orthopaedic Research 42 (1) (2024) 43–53.doi:10.1002/jor.25637

    work page doi:10.1002/jor.25637 2024

  53. [53]

    A. G. Morales, J. J. Lee, F. Caliva, C. Iriondo, F. Liu, S. Majumdar, V. Pedoia, Uncovering associations between data-driven learned qMRI biomarkers and chronic pain, Scientific Reports 11 (1) (2021) 21989. doi:10.1038/s41598-021-01111-x

    work page doi:10.1038/s41598-021-01111-x 2021

  54. [54]

    Pedoia, J

    V. Pedoia, J. Lee, B. Norman, T. Link, S. Majumdar, Diagnosing osteoarthritis from T2 maps using deep learning: an analysis of the entire Osteoarthritis Initiative baseline cohort, Osteoarthritis and Cartilage 27 (7) (2019) 1002–1010.doi:10.1016/j.joca.2019.02.800

    work page doi:10.1016/j.joca.2019.02.800 2019

  55. [55]

    M. Han, R. Tibrewala, E. Bahroos, V. Pedoia, S. Majumdar, Magnetization-prepared spoiled gradient- echo snapshot imaging for efficient measurement of R2 -R 1ρ in knee cartilage, Magnetic Resonance in Medicine 87 (2) (2022) 733–745.doi:10.1002/mrm.29024

    work page doi:10.1002/mrm.29024 2022

  56. [56]

    Carballido-Gamio, G

    J. Carballido-Gamio, G. B. Joseph, J. A. Lynch, T. M. Link, S. Majumdar, Longitudinal analysis of MRI T 2 knee cartilage laminar organization in a subset of patients from the osteoarthritis initiative: A texture approach, Magnetic Resonance in Medicine 65 (4) (2011) 1184–1194.doi:10.1002/mrm.22693

    work page doi:10.1002/mrm.22693 2011

  57. [57]

    Iriondo, V

    C. Iriondo, V. Pedoia, S. Majumdar, Lumbar intervertebral disc characterization through quantitative MRI analysis: An automatic voxel-based relaxometry approach, Magnetic Resonance in Medicine 84 (3) (2020) 1376–1390.doi:10.1002/mrm.28210

    work page doi:10.1002/mrm.28210 2020

  58. [58]

    K. E. Roach, A. L. Bird, V. Pedoia, S. Majumdar, R. B. Souza, Automated evaluation of hip abductor muscle quality and size in hip osteoarthritis: Localized muscle regions are strongly associated with overall muscle quality, Magnetic Resonance Imaging 111 (2024) 237–245.doi:10.1016/j.mri.2024. 04.025

    work page doi:10.1016/j.mri.2024 2024

  59. [59]

    P. W. Hodges, L. Danneels, Changes in Structure and Function of the Back Muscles in Low Back Pain: Different Time Points, Observations, and Mechanisms, Journal of Orthopaedic & Sports Physical Therapy 49 (6) (2019) 464–476.doi:10.2519/jospt.2019.8827

    work page doi:10.2519/jospt.2019.8827 2019

  60. [60]

    Goubert, J

    D. Goubert, J. V. Oosterwijck, M. Meeus, L. Danneels, Structural Changes of Lumbar Muscles in Non-specific Low Back Pain: A Systematic Review, Pain Physician 19 (7) (2016) E985–E1000

    work page 2016

  61. [61]

    Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, St´ efan J

    P. Virtanen, R. Gommers, T. E. Oliphant, M. Haberland, T. Reddy, D. Cournapeau, E. Burovski, P. Pe- terson, W. Weckesser, J. Bright, S. J. Van Der Walt, M. Brett, J. Wilson, K. J. Millman, N. Mayorov, A. R. J. Nelson, E. Jones, R. Kern, E. Larson, C. J. Carey, I. Polat, Y. Feng, E. W. Moore, J. Vander- Plas, D. Laxalde, J. Perktold, R. Cimrman, I. Henriks...

    work page doi:10.1038/s41592-019-0686-2 2020

  62. [62]

    doi:10.21105/joss.01026

    R.Vallat, Pingouin: statisticsinPython, JournalofOpenSourceSoftware3(31)(2018)1026, publisher: The Open Journal. doi:10.21105/joss.01026

    work page doi:10.21105/joss.01026 2018

  63. [63]

    Perktold, S

    J. Perktold, S. Seabold, K. Sheppard, ChadFulton, K. Shedden, jbrockmendel, j grana6, P. Quacken- bush, V. Arel-Bundock, W. McKinney, I. Langmore, B. Baker, R. Gommers, yogabonito, s scherrer, Y. Zhurko, M. Brett, E. Giampieri, yl565, J. Millman, P. Hobson, Vincent, P. Roy, T. Augspurger, tvanzyl, alexbrc, T. Hartley, F. Perez, Y. Tamiya, Y. Halchenko, st...

    work page doi:10.5281/zenodo.10984387 2024

  64. [64]

    F. E. Harrell, Regression Modeling Strategies: With Applications to Linear Models, Logistic and Ordi- nal Regression, and Survival Analysis, Springer Series in Statistics, Springer International Publishing, Cham, 2015, iSSN: 0172-7397, 2197-568X.doi:10.1007/978-3-319-19425-7

    work page doi:10.1007/978-3-319-19425-7 2015

  65. [65]

    H. Uno, T. Cai, M. J. Pencina, R. B. D’Agostino, L. J. Wei, On the C-statistics for evaluating overall adequacy of risk prediction procedures with censored survival data, Statistics in Medicine 30 (10) (2011) 1105–1117, publisher: Wiley.doi:10.1002/sim.4154

    work page doi:10.1002/sim.4154 2011

  66. [66]

    Biewald, Experiment Tracking with Weights and Biases (2020)

    L. Biewald, Experiment Tracking with Weights and Biases (2020). URL https://www.wandb.com/

    work page 2020

  67. [67]

    PyTorch: An Imperative Style, High-Performance Deep Learning Library

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. Köpf, E. Yang, Z. DeVito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, S. Chintala, PyTorch: An Imperative Style, High-Performance Deep Learn- ing Library, version Number: 1 (2019).doi:10.48550/ARXIV.1912.01703

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1912.01703 2019

  68. [68]

    Diaz-Pinto, S

    A. Diaz-Pinto, S. Alle, A. Ihsani, M. Asad, V. Nath, F. Pérez-García, P. Mehta, W. Li, H. R. Roth, T. Vercauteren, D. Xu, P. Dogra, S. Ourselin, A. Feng, M. J. Cardoso, MONAI Label: A framework for AI-assisted interactive labeling of 3D medical images, Medical Image Analysis 95 (2024). doi: 10.1016/j.media.2024.103207

    work page doi:10.1016/j.media.2024.103207 2024

  69. [69]

    Jocher, J

    G. Jocher, J. Qiu, A. Chaurasia, Ultralytics YOLO (Jan. 2023). URL https://github.com/ultralytics/ultralytics 58

    work page 2023