Pith. sign in

REVIEW 2 major objections 2 minor 52 references

LM-CartSeg: Automated Segmentation of Lateral and Medial Cartilage and Subchondral Bone for Radiomics Analysis

T0 review · 2 major / 2 minor · reviewed 2026-05-17 · grok-4.3

Challenge this review Download PDF Read on arXiv ↗

Pith's one-line read An automatic pipeline segments knee cartilage and subchondral bone into stable lateral and medial compartments to extract radiomic features that classify osteoarthritis better than volume or thickness alone.

desk verdict LM-CartSeg gives a workable automatic pipeline for knee cartilage and bone ROIs plus radiomics that beats simple size measures on held-out data, but the geometric L/M rules need checking in advanced OA. read the letter →

arxiv 2512.03449 v3 submitted 2025-12-03 cs.CV

Tongxu Zhang , Zongpan Li , Aaron Kam Lun Leung , Siu Ngor Fu This is my paper · ORCID
classification cs.CV
keywords kneeosteoarthritiscartilagesegmentationradiomicsnnU-NetsubchondralboneMRIautomaticpipelinecompartmentalization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

LM-CartSeg combines two 3D nnU-Net models trained on separate knee MRI datasets and fuses their outputs at test time. Simple geometric rules then clean the segments, build 10 mm subchondral bone bands in physical space, and split the tibia into lateral and medial parts using PCA followed by k-means. The resulting ten regions of interest support extraction of 4,650 non-shape radiomic features per knee. On held-out test data the pipeline reaches DSC near 0.91 internally and 0.80 in zero-shot external evaluation, with quality-control signatures based on volume and thickness. Classification models built on these features reach AUC values of 0.91 and 0.83 on two separate cohorts, outperforming models that use only size-linked features.

What carries the argument

Fusion of two nnU-Net outputs refined by geometric post-processing rules consisting of 10 mm subchondral bone bands and a PCA plus k-means split for lateral/medial tibial compartmentalization.

What would settle it

On an external dataset from a different scanner or with more advanced osteoarthritis, the lateral/medial split produces frequent side swaps or the bone-band construction systematically misses cartilage tissue, causing the radiomic features to lose discriminative power relative to size measures.

Watch

Extended reading notes

Core claim

The paper shows that fusing two nnU-Net predictions and refining them with connected-component cleaning, 10 mm bone bands, and a data-driven PCA+k-means tibial split produces accurate, compartment-stable ROIs. From these ROIs the authors extract thousands of radiomic features of which only 6-12 percent correlate strongly with volume or thickness. Models using the full feature set classify OA versus non-OA cases with AUC up to 0.91 on the internal test set and 0.83 on an independent clinical cohort, exceeding the performance obtained when restricted to morphometric features alone.

Load-bearing premise

The geometric rules for bone-band construction and tibial split together with the nnU-Net fusion remain accurate and free of systematic bias on scanners, populations, and disease stages not seen during training.

Editorial extensions

If this is right

  • Post-processing raises macro ASSD from 2.63 mm to 0.36 mm and HD95 from 25.2 mm to 3.35 mm on the internal test set.
  • The geometric L/M rule yields compartments that stay consistent across datasets while a direct L/M nnU-Net exhibits domain-dependent side swaps.
  • Only 6-12 percent of the 4,650 features per ROI show strong correlation with volume or thickness.
  • Radiomics models using the full feature set achieve higher AUC for OA classification than models limited to size-linked features.
  • The pipeline supplies automatic, QC'd ROIs that can serve as a practical base for multi-centre knee OA radiomics work.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Reducing the need for manual ROI definition could enable radiomics analyses on much larger knee MRI collections than currently feasible.
  • Size-independent features may capture early tissue texture changes that precede visible morphometric alterations in osteoarthritis.
  • If the geometric rules prove robust, the same compartmentalization logic could be transferred to other joints or to CT imaging with only minor parameter adjustments.
  • Standardized automatic ROIs might help resolve reproducibility issues that have slowed adoption of radiomics in osteoarthritis clinical research.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 2 minor

Summary. The paper presents LM-CartSeg, a fully automatic pipeline that trains two 3D nnU-Net models on SKM-TEA (138 knees) and OAIZIB-CM (404 knees) for joint cartilage and subchondral bone segmentation, then applies geometric post-processing (connected-component cleaning, 10 mm subchondral bone bands in physical space, and PCA+k-means tibial L/M split) to produce QC'd ROIs. From these ROIs it extracts 4,650 non-shape radiomic features and demonstrates improved segmentation metrics (ASSD reduced from 2.63 mm to 0.36 mm, DSC ~0.91 on OAIZIB-CM test set; zero-shot DSC ~0.80 on SKI-10) together with OA classification AUCs of 0.91 (OAIZIB-CM) and 0.83 (Po-OA cohort) that exceed models using only volume/thickness features, with only 6-12% feature correlation to morphometry.

Significance. If the central claims hold, the work supplies a practical, reproducible foundation for multi-centre knee OA radiomics by automating anatomically meaningful ROIs with explicit QC signatures and showing that radiomic features carry discriminative information beyond simple morphometry. Strengths include held-out test evaluation, external cohort testing on SKI-10, and explicit reporting of low morphometric correlation; these elements support the claim of utility for larger studies.

major comments (2)
  1. [Results] Results section (and abstract): the geometric rules (10 mm subchondral bone band after connected-component cleaning and PCA+k-means tibial L/M split) are load-bearing for the claim of unbiased, anatomically meaningful ROIs across disease stages. No stratification of segmentation metrics, feature stability, or AUC performance by KL grade or deformity angle (varus/valgus) is reported, leaving open whether the reported gains and discriminative power partly reflect rule-induced artifacts in advanced OA rather than true tissue differences.
  2. [Methods] Methods section: training details, exclusion criteria, and any post-hoc tuning of the nnU-Net models or geometric parameters (free parameters include bone-band width and k-means cluster count) are insufficiently specified to allow full assessment of reproducibility and domain-shift robustness on unseen scanners or populations.
minor comments (2)
  1. [Abstract] Abstract: replace approximate values (DSC approx 0.91, zero-shot DSC approx 0.80) with exact figures or confidence intervals for precision.
  2. [Results] Results: clarify the exact radiomic feature classes contributing to the 4,650 features and the precise definition of the 'non-shape' subset used in the correlation and classification analyses.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and have revised the manuscript to improve clarity, reproducibility, and assessment of robustness where possible.

read point-by-point responses

  1. Referee: [Results] Results section (and abstract): the geometric rules (10 mm subchondral bone band after connected-component cleaning and PCA+k-means tibial L/M split) are load-bearing for the claim of unbiased, anatomically meaningful ROIs across disease stages. No stratification of segmentation metrics, feature stability, or AUC performance by KL grade or deformity angle (varus/valgus) is reported, leaving open whether the reported gains and discriminative power partly reflect rule-induced artifacts in advanced OA rather than true tissue differences.

    Authors: We acknowledge the referee's concern that stratification would provide stronger evidence for the generalizability of the geometric rules across disease stages. The OAIZIB-CM cohort contains a range of KL grades (primarily 0-3), and we will add a supplementary table stratifying DSC, ASSD, and radiomics-based AUC by KL subgroup in the revised manuscript. For deformity angles, these annotations are not available in either training or test cohorts, which limits direct stratification; however, the PCA+k-means split operates in physical space on the tibial plateau geometry and is inherently robust to varus/valgus rotation, as confirmed by stable L/M compartment volumes across the external SKI-10 dataset. We do not believe the rules introduce artifacts, given the large improvement in surface metrics and the low correlation (6-12%) of radiomic features with morphometry, but we will expand the Discussion to explicitly address this potential limitation and the need for future cohorts with deformity annotations. revision: partial

  2. Referee: [Methods] Methods section: training details, exclusion criteria, and any post-hoc tuning of the nnU-Net models or geometric parameters (free parameters include bone-band width and k-means cluster count) are insufficiently specified to allow full assessment of reproducibility and domain-shift robustness on unseen scanners or populations.

    Authors: We agree that the original Methods section lacked sufficient detail for full reproducibility. In the revised manuscript we will expand this section to report: complete nnU-Net training hyperparameters (patch size, batch size, optimizer, learning-rate schedule, and augmentation pipeline); explicit exclusion criteria applied to SKM-TEA and OAIZIB-CM (motion artifacts, incomplete FOV, severe metal artifacts); the anatomical rationale for the fixed 10 mm subchondral bone band (derived from typical cartilage-plus-subchondral thickness ranges in the literature); confirmation that k-means is deterministically set to k=2 with no post-hoc tuning beyond a single validation fold; and a brief sensitivity analysis of band width (8 mm, 10 mm, 12 mm) on a held-out subset. These additions will enable readers to assess domain-shift robustness more rigorously. revision: yes

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: all metrics derived from held-out data with independent geometric rules

full rationale

The paper trains two nnU-Net models on SKM-TEA and OAIZIB-CM, then applies deterministic post-processing (connected-component cleaning, fixed 10 mm subchondral bone band in physical space, and PCA+k-means tibial split) to produce ROIs on separate test sets (OAIZIB-CM test, SKI-10). Radiomic features are extracted from these ROIs and evaluated for correlation with volume/thickness (reported 6-12%) plus classification AUC on held-out OAIZIB-CM and Po-OA cohorts. No equation or self-citation reduces the final AUC or segmentation metrics to quantities defined by the same fitted parameters; the geometric rules are not optimized against the radiomics outcome, and all reported numbers come from unseen data. The derivation chain is therefore self-contained against external benchmarks.

Assumptions & free parameters 2 free parameters · 1 assumptions · 0 invented entities

The central claim rests on standard deep-learning assumptions plus two ad-hoc geometric choices whose justification is empirical rather than derived.

free parameters (2)
  • subchondral bone band width = 10 mm
    Fixed 10 mm width used to construct the bone region in physical space; chosen to capture relevant anatomy.
  • k-means cluster count for L/M split = 2
    Set to two clusters for the data-driven tibial lateral/medial division.
assumptions (1)
  • domain assumption The SKM-TEA and OAIZIB-CM training sets are sufficiently representative for zero-shot generalization to the test distributions.
    Invoked when the two nnU-Net models are applied to new data without retraining.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LM-CartSeg: Automated Segmentation of Lateral and Medial Cartilage and Subchondral Bone for Radiomics Analysis." pith.science (2026). https://pith.science/paper/2512.03449

@misc{pith2026251203449,
  author       = {Pith},
  title        = {Pith review of: LM-CartSeg: Automated Segmentation of Lateral and Medial Cartilage and Subchondral Bone for Radiomics Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2512.03449}},
  note         = {Machine review of arXiv:2512.03449}
}
abstract

Background and Objective: Radiomics of knee MRI requires robust, anatomically meaningful regions of interest (ROIs) that jointly capture cartilage and subchondral bone. Most existing work relies on manual ROIs and rarely reports quality control (QC). We present LM-CartSeg, a fully automatic pipeline for cartilage/bone segmentation, geometric lateral/medial (L/M) compartmentalization and radiomics analysis. Methods:Two 3D nnU-Net models were trained on SKM-TEA (138 knees) and OAIZIB-CM (404 knees). At test time, zero-shot predictions were fused and refined by simple geometric rules: connected-component cleaning,construction of 10mm subchondral bone bands in physical space, and a data-driven tibial L/M split based on PCA and $k$-means. Segmentation was evaluated on an OAIZIB-CM test set (103 knees) and on SKI-10 (100 knees). QC used volume and thickness signatures. From 10 ROIs we extracted 4,650 non-shape radiomic features to study inter-compartment similarity, dependence on ROI size, and OA vs. non-OA classification on OAIZIB-CM and a clinical Po-OA cohort (185 knees). Results: Post-processing improved macro ASSD on OAIZIB-CM from 2.63 to 0.36mm and HD95 from 25.2 to 3.35mm, with DSC approx 0.91; zero-shot DSC on SKI-10 was approx 0.80. The geometric L/M rule produced stable compartments across datasets, whereas a direct L/M nnU-Net showed domain-dependent side swaps. Only 6-12% of features per ROI were strongly correlated with volume or thickness. Radiomics-based models achieved AUC up to 0.91 (OAIZIB-CM) and 0.83 (Po-OA), clearly exceeding models restricted to size-linked features. Conclusions: LM-CartSeg yields automatic, QC'd ROIs and radiomic features that carry discriminative information beyond simple morphometry, providing a practical foundation for multi-centre knee OA radiomics studies.

Figures

Figures reproduced from arXiv: 2512.03449 by the authors.

Figure 1
Figure 1. Overview of the proposed pipeline: LM-CartSeg. A 3D knee MR volume is processed [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Data-driven tibial compartment definition: (A) Segmented cartilage and subchon [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Macro-averaged DSC, ASSD and HD95 on the OAIZIB-CM test set, comparing [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Zero-shot macro-averaged DSC, ASSD and HD95 on the SKI-10 dataset, comparing [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: Accuracy of the learned tibial L/M partition across datasets. Rows (top to bot [PITH_FULL_IMAGE:figures/full_fig_p015_5.png]
Figure 6
Figure 6. Figure 6: Anatomical QC of 10 mm cartilage bands across datasets. Rows (top to bottom) show OAIZIB-CM, SKI-10 and Po-OA; columns (left to right) show tibial cartilage volume (LT vs. MT), tibial mean thickness, medial/lateral tibial and femoral thickness ratios, femoral cartilage…
Figure 7
Figure 7. Figure 7: Radiomic similarity between knee compartments across datasets. Chord diagrams [PITH_FULL_IMAGE:figures/full_fig_p018_7.png]
Figure 8
Figure 8. Figure 8: Number of radiomic features strongly correlated with ROI size. For each ROI and [PITH_FULL_IMAGE:figures/full_fig_p019_8.png]
Figure 9
Figure 9. Figure 9: OA vs non-OA classification ROC curves (5-fold CV) on OAIZIB-CM (top) and Po [PITH_FULL_IMAGE:figures/full_fig_p021_9.png]
Figure 10
Figure 10. Figure 10: Correlation of LASSO-selected radiomic features with ROI size. Heatmaps show [PITH_FULL_IMAGE:figures/full_fig_p022_10.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 52 canonical work pages

  1. [1]

    E. R. Vina, C. K. Kwoh, Epidemiology of osteoarthritis: literature update, Current opinion in rheumatology 30 (2018) 160–167

    work page 2018

  2. [2]

    J. D. Steinmetz, G. T. Culbreth, L. M. Haile, Q. Rafferty, J. Lo, K. G. Fukutaki, J. A. Cruz, A. E. Smith, S. E. Vollset, P. M. Brooks, et al., Global, regional, and national burden of osteoarthritis, 1990–2020 and pro- jections to 2050: a systematic analysis for the global burden of disease study 2021, The Lancet Rheumatology 5 (2023) e508–e522

    work page 1990

  3. [3]

    B. A. de Vries, S. J. Breda, B. Sveinsson, E. J. McWalter, D. E. Meuffels, G. P. Krestin, B. A. Hargreaves, G. E. Gold, E. H. Oei, Detection of knee synovitis using non-contrast-enhanced qdess compared with contrast- enhanced mri, Arthritis Research & Therapy 23 (2021) 55

    work page 2021

  4. [4]

    Y. Wang, A. E. Wluka, G. Jones, C. Ding, F. M. Cicuttini, Use magnetic resonance imaging to assess articular cartilage, Therapeutic advances in musculoskeletal disease 4 (2012) 77–97

    work page 2012

  5. [5]

    A. H. Gomoll, H. Yoshioka, A. Watanabe, J. C. Dunn, T. Minas, Preop- erative measurement of cartilage defects by mri underestimates lesion size, Cartilage 2 (2011) 389–393. 25

    work page 2011

  6. [6]

    G. Cao, S. Gao, B. Xiong, Application of quantitative t1, t2 and t2* map- ping magnetic resonance imaging in cartilage degeneration of the shoulder joint, Scientific Reports 13 (2023) 4558

    work page 2023

  7. [7]

    Lambin, R

    P. Lambin, R. T. Leijenaar, T. M. Deist, J. Peerlings, E. E. De Jong, J. Van Timmeren, S. Sanduleanu, R. T. Larue, A. J. Even, A. Jochems, et al., Radiomics: the bridge between medical imaging and personalized medicine, Nature reviews Clinical oncology 14 (2017) 749–762

    work page 2017

  8. [8]

    J. E. Van Timmeren, D. Cester, S. Tanadini-Lang, H. Alkadhi, B. Baessler, Radiomics in medical imaging—“how-to” guide and critical reflection, In- sights into imaging 11 (2020) 91

    work page 2020

Show all 52 references
  1. [9]

    Hirvasniemi, S

    J. Hirvasniemi, S. Klein, S. Bierma-Zeinstra, M. W. Vernooij, D. Schiphof, E. H. Oei, A machine learning approach to distinguish between knees without and with osteoarthritis using mri-based radiomic features from tibial bone, European radiology 31 (2021) 8513–8521

    2021

  2. [10]

    Z. Xue, L. Wang, Q. Sun, J. Xu, Y. Liu, S. Ai, L. Zhang, C. Liu, Radiomics analysis using mr imaging of subchondral bone for identification of knee osteoarthritis, Journal of Orthopaedic Surgery and Research 17 (2022) 414

    2022

  3. [11]

    T. Cui, R. Liu, Y. Jing, J. Fu, J. Chen, Development of machine learning models aiming at knee osteoarthritis diagnosing: an mri radiomics analysis, Journal of Orthopaedic Surgery and Research 18 (2023) 375

    2023

  4. [12]

    J. Fu, L. Mu, D. Dong, M. Li, Z. Miao, X. Huai, Y. Zheng, H. Zhang, An mri-based radiomics framework for early identification and progression stratification in knee osteoarthritis: data from the osteoarthritis initiative, BMC Musculoskeletal Disorders 26 (2025) 1–10

    2025

  5. [13]

    T. Chen, J. Chen, H. Liu, Z. Liu, B. Yu, Y. Wang, W. Zhao, Y. Peng, J. Li, Y.Yang, etal., Integrationoflongitudinalload-bearingtissuemriradiomics and neural network to predict knee osteoarthritis incidence, Journal of Orthopaedic Translation 51 (2025) 187–197

    2025

  6. [14]

    Angelone, F

    F. Angelone, F. K. Ciliberti, H. Jónsson, M. K. Gíslason, M. Romano, A. Franco, F. Amato, P. Gargiulo, Knee cartilage degradation in the me- dial and lateral anatomical compartments: a radiomics study, in: 2024 IEEE international conference on metrology for eXtended reality, a...

    2024

  7. [15]

    Angelone, F

    F. Angelone, F. K. Ciliberti, G. P. Tobia, H. Jónsson Jr, A. M. Ponsiglione, M. K. Gislason, F. Tortorella, F. Amato, P. Gargiulo, Innovative diagnostic approaches for predicting knee cartilage degeneration in osteoarthritis pa- tients: a radiomics-based study, Information Sys...

    2025

  8. [16]

    T. Lin, S. Peng, S. Lu, S. Fu, D. Zeng, J. Li, T. Chen, T. Fan, C. Lang, S. Feng, et al., Prediction of knee pain improvement over two years for knee osteoarthritis using a dynamic nomogram based on mri-derived radiomics: a proof-of-concept study, Osteoarthritis and cartilage ...

    2023

  9. [17]

    Y. Xie, Y. Dan, H. Tao, C. Wang, C. Zhang, Y. Wang, J. Yang, G. Yang, S. Chen, Radiomics feature analysis of cartilage and subchondral bone in differentiating knees predisposed to posttraumatic osteoarthritis after ante- rior cruciate ligament reconstruction from healthy knees...

    2021

  10. [18]

    Isensee, P

    F. Isensee, P. F. Jaeger, S. A. Kohl, J. Petersen, K. H. Maier-Hein, nnu- net: a self-configuring method for deep learning-based biomedical image segmentation, Nature methods 18 (2021) 203–211

    2021

  11. [19]

    C. G. Chadoulos, D. E. Tsaopoulos, S. Moustakidis, N. L. Tsakiridis, J. B. Theocharis, A novel multi-atlas segmentation approach under the semi- supervisedlearningframework: Applicationtokneecartilagesegmentation, Computer Methods and Programs in Biomedicine 227 (2022) 107208

    2022

  12. [20]

    S. Khan, B. Azam, Y. Yao, W. Chen, Deep collaborative network with alpha matte for precise knee tissue segmentation from mri, Computer Methods and Programs in Biomedicine 222 (2022) 106963

    2022

  13. [21]

    Isensee, T

    F. Isensee, T. Wald, C. Ulrich, M. Baumgartner, S. Roy, K. Maier-Hein, P. F. Jaeger, nnu-net revisited: A call for rigorous validation in 3d medical image segmentation, in: International Conference on Medical Image Com- puting and Computer-Assisted Intervention, Springer, 2024...

    2024

  14. [22]

    W. Lv, J. Peng, J. Hu, Y. Lu, Z. Zhou, H. Xu, K. Xing, X. Zhang, L. Lu, Lmsst-gcn: Longitudinal mri sub-structural texture guided graph convo- lution network for improved progression prediction of knee osteoarthritis, Computer Methods and Programs in Biomedicine 261 (2025) 108600

    2025

  15. [23]

    Ambellan, A

    F. Ambellan, A. Tack, M. Ehlke, S. Zachow, Automated segmentation of knee bone and cartilage combining statistical shape knowledge and convo- lutional neural networks: Data from the osteoarthritis initiative, Medical image analysis 52 (2019) 109–118

    2019

  16. [24]

    Y. Yao, J. Zhong, L. Zhang, S. Khan, W. Chen, Cartimorph: A frame- work for automated knee articular cartilage morphometrics, Medical Image Analysis 91 (2024) 103035

    2024

  17. [25]

    Wirth, O

    W. Wirth, O. Benichou, C. K. Kwoh, A. Guermazi, D. Hunter, R. Putz, F. Eckstein, O. Investigators, Spatial patterns of cartilage loss in the me- dial femoral condyle in osteoarthritic knees: data from the osteoarthritis initiative, Magnetic resonance in medicine 63 (2010) 574–581. 27

    2010

  18. [26]

    R. K. Surowiec, E. P. Lucas, E. K. Fitzcharles, B. M. Petre, G. J. Dornan, J. E. Giphart, R. F. LaPrade, C. P. Ho, T2 values of articular cartilage in clinically relevant subregions of the asymptomatic knee, Knee Surgery, Sports Traumatology, Arthroscopy 22 (2014) 1404–1414

    2014

  19. [27]

    Panfilov, A

    E. Panfilov, A. Tiulpin, M. T. Nieminen, S. Saarakkala, V. Casula, Deep learning-based segmentation of knee mri for fully automatic subregional morphological assessment of cartilage tissues: data from the osteoarthritis initiative, Journal of Orthopaedic Research®40 (2022) 1113–1124

    2022

  20. [28]

    A. D. Desai, A. M. Schmidt, E. B. Rubin, C. M. Sandino, M. S. Black, V. Mazzoli, K. J. Stevens, R. Boutin, C. Ré, G. E. Gold, et al., Skm-tea: A dataset for accelerated mri reconstruction with dense image labels for quantitative clinical evaluation, arXiv preprint arXiv:2203.0...

    2022

  21. [29]

    Carré, G

    A. Carré, G. Klausner, M. Edjlali, M. Lerousseau, J. Briend-Diop, R. Sun, S. Ammari, S. Reuzé, E. Alvarez Andres, T. Estienne, et al., Standardiza- tion of brain mr images across machines and protocols: bridging the gap for mri-based radiomics, Scientific reports 10 (2020) 12340

    2020

  22. [30]

    Zwanenburg, M

    A. Zwanenburg, M. Vallières, M. A. Abdalah, H. J. Aerts, V. Andrearczyk, A. Apte, S. Ashrafinia, S. Bakas, R. J. Beukinga, R. Boellaard, et al., The image biomarker standardization initiative: standardized quantitative radiomics for high-throughput image-based phenotyping, Rad...

    2020

  23. [31]

    Dovrou, K

    A. Dovrou, K. Nikiforaki, D. Zaridis, G. C. Manikis, E. Mylona, N. Tachos, M. Tsiknakis, D. I. Fotiadis, K. Marias, A segmentation-based method improving the performance of n4 bias field correction on t2weighted mr imaging data of the prostate, Magnetic Resonance Imaging 101 (...

    2023

  24. [32]

    Hotelling, Analysis of a complex of statistical variables into principal components., Journal of educational psychology 24 (1933) 417

    H. Hotelling, Analysis of a complex of statistical variables into principal components., Journal of educational psychology 24 (1933) 417

    1933

  25. [33]

    MacQueen, Multivariate observations, in: Proceedings ofthe 5th Berke- ley Symposium on Mathematical Statisticsand Probability, volume 1, 1967, pp

    J. MacQueen, Multivariate observations, in: Proceedings ofthe 5th Berke- ley Symposium on Mathematical Statisticsand Probability, volume 1, 1967, pp. 281–297

    1967

  26. [34]

    Hafezi-Nejad, A

    N. Hafezi-Nejad, A. Guermazi, F. W. Roemer, D. J. Hunter, E. B. Dam, B. Zikria, C. K. Kwoh, S. Demehri, Prediction of medial tibiofemoral com- partment joint space loss progression using volumetric cartilage measure- ments: data from the fnih oa biomarkers consortium, European...

    2017

  27. [35]

    Heimann, B

    T. Heimann, B. J. Morrison, M. A. Styner, M. Niethammer, S. Warfield, Segmentation of knee images: a grand challenge, in: Proc. MICCAI Work- shop on Medical Image Analysis for the Clinic, volume 1, Beijing, China, 2010. 28

    2010

  28. [36]

    L. R. Dice, Measures of the amount of ecologic association between species, Ecology 26 (1945) 297–302

    1945

  29. [37]

    Heimann, B

    T. Heimann, B. Van Ginneken, M. A. Styner, Y. Arzhaeva, V. Aurich, C. Bauer, A. Beck, C. Becker, R. Beichel, G. Bekes, et al., Comparison and evaluation of methods for liver segmentation from ct datasets, IEEE transactions on medical imaging 28 (2009) 1251–1265

    2009

  30. [38]

    Fournel, A

    J. Fournel, A. Bartoli, D. Bendahan, M. Guye, M. Bernard, E. Rauseo, M. Y. Khanji, S. E. Petersen, A. Jacquier, B. Ghattas, Medical image segmentation automatic quality control: A multi-dimensional approach, Medical Image Analysis 74 (2021) 102213

    2021

  31. [39]

    J. Guo, P. Yan, Y. Qin, M. Liu, Y. Ma, J. Li, R. Wang, H. Luo, S. Lv, Automated measurement and grading of knee cartilage thickness: a deep learning-based approach, Frontiers in Medicine 11 (2024) 1337993

    2024

  32. [40]

    Specktor-Fadida, L

    B. Specktor-Fadida, L. Ben-Sira, D. Ben-Bashat, L. Joskowicz, Segqc: a segmentation network-based framework for multi-metric segmentation quality control and segmentation error detection in volumetric medical im- ages, Medical Image Analysis (2025) 103638

    2025

  33. [41]

    Peuna, J

    A. Peuna, J. Hekkala, M. Haapea, J. Podlipská, A. Guermazi, S. Saarakkala, M. T. Nieminen, E. Lammentausta, Variable angle gray level co-occurrence matrix analysis of t2 relaxation time maps reveals degener- ative changes of cartilage in knee osteoarthritis: Oulu knee osteoart...

    2018

  34. [42]

    B. R. Kim, H. J. Yoo, H.-D. Chae, S. H. Hong, J.-Y. Choi, Fat-suppressed t2 mapping of human knee femoral articular cartilage: comparison with conventional t2 mapping, BMC musculoskeletal disorders 22 (2021) 662

    2021

  35. [43]

    J. J. Van Griethuysen, A. Fedorov, C. Parmar, A. Hosny, N. Aucoin, V. Narayan, R. G. Beets-Tan, J.-C. Fillion-Robin, S. Pieper, H. J. Aerts, Computational radiomics system to decode the radiographic phenotype, Cancer research 77 (2017) e104–e107

    2017

  36. [44]

    Duron, D

    L. Duron, D. Balvay, S. Vande Perre, A. Bouchouicha, J. Savatovsky, J.-C. Sadik, I. Thomassin-Naggara, L. Fournier, A. Lecler, Gray-level discretiza- tion impacts reproducible mri radiomics texture features, PLoS One 14 (2019) e0213459

    2019

  37. [45]

    Tibshirani, Regression shrinkage and selection via the lasso, Journal of the Royal Statistical Society Series B: Statistical Methodology 58 (1996) 267–288

    R. Tibshirani, Regression shrinkage and selection via the lasso, Journal of the Royal Statistical Society Series B: Statistical Methodology 58 (1996) 267–288

    1996

  38. [46]

    Rovetta, Raiders of the lost correlation: a guide on using pearson and spearman coefficients to detect hidden correlations in medical sciences, Cureus 12 (2020)

    A. Rovetta, Raiders of the lost correlation: a guide on using pearson and spearman coefficients to detect hidden correlations in medical sciences, Cureus 12 (2020). 29

    2020

  39. [47]

    Cover, P

    T. Cover, P. Hart, Nearest neighbor pattern classification, IEEE transac- tions on information theory 13 (1967) 21–27

    1967

  40. [48]

    Schölkopf, A

    B. Schölkopf, A. J. Smola, Learning with kernels: support vector machines, regularization, optimization, and beyond, MIT press, 2002

    2002

  41. [49]

    T. Chen, C. Guestrin, Xgboost: A scalable tree boosting system, in: Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining, 2016, pp. 785–794

    2016

  42. [50]

    R. F. Shah, A. M. Martinez, V. Pedoia, S. Majumdar, T. P. Vail, S. A. Bini, Variation in the thickness of knee cartilage. the use of a novel machine learningalgorithmforcartilagesegmentationofmagneticresonanceimages, The Journal of arthroplasty 34 (2019) 2210–2215

    2019

  43. [51]

    M.P.Jansen, S.C.Mastbergen, J.W.MacKay, T.D.Turmezei, F.Lafeber, Knee joint distraction results in mri cartilage thickness increase up to 10 years after treatment, Rheumatology 61 (2022) 974–982

    2022

  44. [52]

    Giurazza, C

    G. Giurazza, C. Caria, S. Campi, E. Franceschetti, G. F. Papalia, S. Bas- ciani, A. Zampoli, P. Gregori, R. Papalia, A. Marinozzi, Femoral cartilage thickness measured on mri varies among individuals: Time to deepen one of the principles of kinematic alignment in total knee ar...

    2025

Pith tools

Reviewed May 17, 2026 · model on record in the stance chip above.