FREE SHIPPING over $40
OVER 9000 FIVE-STAR REVIEWS
5% OFF ALL SUBSCRIPTIONS
100% HAPPINESS GUARANTEE
Analyze Diet
0
Home / Equine Research Bank / J Magn Reson Imaging / Machine Learning Prediction of Collagen Fiber Orientation an...
Journal of magnetic resonance imaging : JMRI2022; 57(4); 1056-1068; doi: 10.1002/jmri.28353

Machine Learning Prediction of Collagen Fiber Orientation and Proteoglycan Content From Multiparametric Quantitative MRI in Articular Cartilage.

Abstract: Machine learning models trained with multiparametric quantitative MRIs (qMRIs) have the potential to provide valuable information about the structural composition of articular cartilage. To study the performance and feasibility of machine learning models combined with qMRIs for noninvasive assessment of collagen fiber orientation and proteoglycan content. Retrospective, animal model. An open-source single slice MRI dataset obtained from 20 samples of 10 Shetland ponies (seven with surgically induced cartilage lesions followed by treatment and three healthy controls) yielded to 1600 data points, including 10% for test and 90% for train validation. A 9.4 T MRI scanner/qMRI sequences: T , T , adiabatic T and T , continuous-wave T and relaxation along a fictitious field (T ) maps. Five machine learning regression models were developed: random forest (RF), support vector regression (SVR), gradient boosting (GB), multilayer perceptron (MLP), and Gaussian process regression (GPR). A nested cross-validation was used for performance evaluation. For reference, proteoglycan content and collagen fiber orientation were determined by quantitative histology from digital densitometry (DD) and polarized light microscopy (PLM), respectively. Normality was tested using Shapiro-Wilk test, and association between predicted and measured values was evaluated using Spearman's Rho test. A P-value of 0.05 was considered as the limit of statistical significance. Four out of the five models (RF, GB, MLP, and GPR) yielded high accuracy (R  = 0.68-0.75 for PLM and 0.62-0.66 for DD), and strong significant correlations between the reference measurements and predicted cartilage matrix properties (Spearman's Rho = 0.72-0.88 for PLM and 0.61-0.83 for DD). GPR algorithm had the highest accuracy (R  = 0.75 and 0.66) and lowest prediction-error (root mean squared [RMSE] = 1.34 and 2.55) for PLM and DD, respectively. Multiparametric qMRIs in combination with regression models can determine cartilage compositional and structural features, with higher accuracy for collagen fiber orientation than proteoglycan content. 2 TECHNICAL EFFICACY: Stage 2.
© 2022 The Authors. Journal of Magnetic Resonance Imaging published by Wiley Periodicals LLC on behalf of International Society for Magnetic Resonance in Medicine.
Get Full Text
Publication Date: 2022-07-21 PubMed ID: 35861162DOI: 10.1002/jmri.28353Google Scholar: Lookup
The Equine Research Bank provides access to a large database of publicly available scientific literature. Inclusion in the Research Bank does not imply endorsement of study methods or findings by Mad Barn.
LinkedInCopy
  • Journal Article
  • Research Support
  • Non-U.S. Gov't

Summary

This research summary has been generated with artificial intelligence and may contain errors and omissions. Refer to the original study to confirm details provided. Submit correction.

Research Overview: This study investigates the efficacy of machine learning models combined with multiparametric quantitative MRIs (qMRIs) in predicting the structural composition of articular cartilage, focusing specifically on collagen fiber orientation and proteoglycan content. The results suggest these models can leverage qMRI data to infer cartilage features with high accuracy, particularly collagen fiber orientation.

Research Design and Methodology

  • The researchers employed a retrospective, animal model study approach, using a single slice MRI dataset from 20 samples of 10 Shetland ponies. Seven of these ponies had surgical-induced cartilage lesions followed by treatment, and three were healthy controls.
  • The dataset was split into 10% for testing and 90% for training and validation. The data was captured using a 9.4 T MRI scanner/qMRI sequences: T , T, adiabatic T and T, continuous-wave T, and relaxation along a fictitious field (T) maps.
  • Five machine learning regression models were deployed: random forest (RF), support vector regression (SVR), gradient boosting (GB), multilayer perceptron (MLP), and Gaussian process regression (GPR). These models’ performance was evaluated using nested cross-validation.
  • As a reference standard for the study, the proteoglycan content and collagen fiber orientation were determined by quantitative histology using digital densitometry (DD) and polarized light microscopy (PLM) respectively. Statistical testing included the Shapiro-Wilk test for normality and Spearman’s Rho test to assess the association between the predicted and measured values.

Main Findings

  • Four out of the five models (RF, GB, MLP, and GPR) showed high accuracy in predicting collagen fiber orientation (R=0.68-0.75) and proteoglycan content (R=0.62-0.66).
  • Strong and statistically significant correlations between the reference measurements and predicted cartilage matrix properties were observed. For example, Spearman’s Rho values stood at 0.72-0.88 for PLM and 0.61-0.83 for DD.
  • GPR algorithm demonstrated the highest accuracy (R=0.75 and 0.66) and lowest prediction-error for both PLM and DD.

Conclusion and Implications

  • The results suggest that the combination of multiparametric qMRIs and machine learning regression models can accurately predict cartilage compositional and structural features.
  • The accuracy was found to be higher for collagen fiber orientation than proteoglycan content. Further investigation may be needed to improve the predictive accuracy for proteoglycan content.
  • The study demonstrates the potential for noninvasive methods to predict complex characteristics within the composition of articular cartilage with considerable success. This could shape future efforts in understanding and managing diseases related to cartilage degradation, such as osteoarthritis.

Cite This Article

APA
Mirmojarabian SA, Kajabi AW, Ketola JHJ, Nykänen O, Liimatainen T, Nieminen MT, Nissi MJ, Casula V. (2022). Machine Learning Prediction of Collagen Fiber Orientation and Proteoglycan Content From Multiparametric Quantitative MRI in Articular Cartilage. J Magn Reson Imaging, 57(4), 1056-1068. https://doi.org/10.1002/jmri.28353

Publication

Journal of magnetic resonance imaging : JMRI
ISSN: 1522-2586
NlmUniqueID: 9105850
Country: United States
Language: English
Volume: 57
Issue: 4
Pages: 1056-1068

Researcher Affiliations

Mirmojarabian, Seyed Amir
  • Research Unit of Medical Imaging, Physics and Technology, University of Oulu, Oulu, Finland.
Kajabi, Abdul Wahed
  • Center for Magnetic Resonance Research, University of Minnesota, Minneapolis, MN, US.
Ketola, Juuso H J
  • Research Unit of Medical Imaging, Physics and Technology, University of Oulu, Oulu, Finland.
Nykänen, Olli
  • Department of Applied Physics, University of Eastern Finland, Kuopio, Finland.
Liimatainen, Timo
  • Research Unit of Medical Imaging, Physics and Technology, University of Oulu, Oulu, Finland.
  • Department of Diagnostic Radiology, Oulu University Hospital, Oulu, Finland.
Nieminen, Miika T
  • Research Unit of Medical Imaging, Physics and Technology, University of Oulu, Oulu, Finland.
  • Department of Diagnostic Radiology, Oulu University Hospital, Oulu, Finland.
  • Medical Research Center, University of Oulu and Oulu University Hospital, Oulu, Finland.
Nissi, Mikko J
  • Research Unit of Medical Imaging, Physics and Technology, University of Oulu, Oulu, Finland.
  • Department of Applied Physics, University of Eastern Finland, Kuopio, Finland.
Casula, Victor
  • Research Unit of Medical Imaging, Physics and Technology, University of Oulu, Oulu, Finland.
  • Medical Research Center, University of Oulu and Oulu University Hospital, Oulu, Finland.

MeSH Terms

  • Animals
  • Horses
  • Cartilage, Articular / pathology
  • Proteoglycans
  • Retrospective Studies
  • Magnetic Resonance Imaging
  • Machine Learning
  • Collagen

References

This article includes 35 references
  1. Guermazi A, Alizai H, Crema MD, Trattnig S, Regatte RR, Roemer FW. Compositional MRI techniques for evaluation of cartilage degeneration in osteoarthritis.. Osteoarthr Cartil 2015;23:1639-1653.
  2. Xia Y, Moody JB, Alhadlaq H. Orientational dependence of T2 relaxation in articular cartilage: A microscopic MRI (μMRI) study.. Magn Reson Med 2002;48:460-469.
  3. Rautiainen J, Nissi MJ, Liimatainen T, Herzog W, Korhonen RK, Nieminen MT. Adiabatic rotating frame relaxation of MRI reveals early cartilage degeneration in a rabbit model of anterior cruciate ligament transection.. Osteoarthr Cartil 2014;22:1444-1452.
  4. Nissi MJ, Salo E-N, Tiitu V. Multi-parametric MRI characterization of enzymatically degraded articular cartilage.. J Orthop Res 2016;34:1111-1120.
  5. Kajabi AW, Casula V, Ojanen S. Multiparametric MR imaging reveals early cartilage degeneration at 2 and 8 weeks after ACL transection in a rabbit model.. J Orthop Res 2020;38:1974-1986.
  6. Lin P-C, Reiter DA, Spencer RG. Sensitivity and specificity of univariate MRI analysis of experimentally degraded cartilage.. Magn Reson Med 2009;62:1311-1318.
  7. Nissi MJ, Rieppo J, Töyräs J. Estimation of mechanical properties of articular cartilage with MRI - dGEMRIC, T2 and T1 imaging in different species with variable stages of maturation.. Osteoarthr Cartil 2007;15:1141-1148.
  8. Englund M, Roos EM, Lohmander LS. Impact of type of meniscal tear on radiographic and symptomatic knee osteoarthritis: A sixteen-year followup of meniscectomy with matched controls.. Arthritis Rheum 2003;48:2178-2187.
  9. Sophia Fox AJ, Bedi A, Rodeo SA. The basic science of articular cartilage: Structure, composition, and function.. Sports Health 2009;1:461-468.
  10. Laasanen M, Töyräs J, Korhonen R. Biomechanical properties of knee articular cartilage.. Biorheology 2003;40:133-140.
  11. Saarakkala S, Julkunen P, Kiviranta P, Mäkitalo J, Jurvelin JS, Korhonen RK. Depth-wise progression of osteoarthritis in human articular cartilage: Investigation of composition, structure and biomechanics.. Osteoarthr Cartil 2010;18:73-81.
  12. Lin P-C, Reiter DA, Spencer RG. Classification of degraded cartilage through multiparametric MRI analysis.. J Magn Reson 2009;201:61-71.
  13. Reiter DA, Roque RA, Lin P-C. Mapping proteoglycan-bound water in cartilage: Improved specificity of matrix assessment using multiexponential transverse relaxation analysis.. Magn Reson Med 2011;65:377-384.
  14. Nykänen O, Sarin JK, Ketola JH. T2* and quantitative susceptibility mapping in an equine model of post-traumatic osteoarthritis: Assessment of mechanical and structural properties of articular cartilage.. Osteoarthr Cartil 2019;27:1481-1490.
  15. Reiter DA, Irrechukwu O, Lin P-C. Improved MR-based characterization of engineered cartilage using multiexponential T2 relaxation and multivariate analysis.. NMR Biomed 2012;25:476-488.
  16. Lin P-C, Irrechukwu O, Roque R, Hancock B, Fishbein KW, Spencer RG. Multivariate analysis of cartilage degradation using the support vector machine algorithm.. Magn Reson Med 2012;67:1815-1826.
  17. Chea P, Mandell JC. Current applications and future directions of deep learning in musculoskeletal radiology.. Skeletal Radiol 2020;49:183-197.
  18. Prasoon A, Petersen K, Igel C, Lauze F, Dam E, Nielsen M. Deep feature learning for knee cartilage segmentation using a triplanar convolutional neural network.. In: Mori K, Sakuma I, Sato Y, Barillot C, Navab N, editors. Med image Comput Comput Interv - MICCAI 2013. Berlin, Heidelberg: Springer; 2013. p 246-253.
  19. Xue Y, Zhang R, Deng Y, Chen K, Jiang T. A preliminary examination of the diagnostic value of deep learning in hip osteoarthritis.. PLoS One 2017;12:e0178992.
  20. Linka K, Itskov M, Truhn D, Nebelung S, Thüring J. T2 MR imaging vs. computational modeling of human articular cartilage tissue functionality.. J Mech Behav Biomed Mater 2017;74:477-487.
  21. Linka K, Thüring J, Rieppo L. Machine learning-augmented and microspectroscopy-informed multiparametric MRI for the non-invasive prediction of articular cartilage composition.. Osteoarthr Cartil 2021;29:592-602.
  22. Thüring J, Linka K, Itskov M. Multiparametric MRI and computational modelling in the assessment of human articular cartilage properties: A comprehensive approach.. Biomed Res Int 2018;2018:9460456.
  23. Sarin JK, te Moller NCR, Mancini IAD. Arthroscopic near infrared spectroscopy enables simultaneous quantitative evaluation of articular cartilage and subchondral bone in vivo.. Sci Rep 2018;8:13409.
  24. Sarin JK, Nykänen O, Tiitu V. Arthroscopic determination of cartilage proteoglycan content and collagen network structure with near-infrared spectroscopy.. Ann Biomed Eng 2019;47:1815-1826.
  25. Kajabi AW, Casula V, Sarin JK. Evaluation of articular cartilage with quantitative MRI in an equine model of post-traumatic osteoarthritis.. J Orthop Res 2021;39:63-73.
  26. Witschey WRT, Borthakur A, Elliott MA. Artifacts in T1ρ-weighted imaging: Compensation for B1 and B0 field imperfections.. J Magn Reson 2007;186:75-85.
  27. Michaeli S, Sorce D, Garwood M. T2ρ and T1ρ adiabatic relaxations and contrasts.. Curr Anal Chem 2008;4:8-25.
  28. Liimatainen T, Sorce DJ, O'Connell R, Garwood M, Michaeli S. MRI contrast from relaxation along a fictitious field (RAFF).. Magn Reson Med 2010;64:983-994.
  29. Kajabi AW, Casula V, Nissi MJ. Assessment of meniscus with adiabatic T1ρ and T2ρ relaxation time in asymptomatic subjects and patients with mild osteoarthritis: A feasibility study.. Osteoarthr Cartil 2018;26:580-587.
  30. Kiviranta I, Jurvelin J, Säämänen A-M, Helminen HJ. Microspectrophotometric quantitation of glycosaminoglycans in articular cartilage sections stained with safranin O.. Histochemistry 1985;82:249-255.
  31. Rieppo J, Hallikainen J, Jurvelin JS, Kiviranta I, Helminen HJ, Hyttinen MM. Practical considerations in the use of polarized light microscopy in the analysis of the collagen network in articular cartilage.. Microsc Res Tech 2008;71:279-287.
  32. Krstajic D, Buturovic LJ, Leahy DE, Thomas S. Cross-validation pitfalls when selecting and assessing regression and classification models.. J Chem 2014;6:10.
  33. Linka K, Schäfer A, Hillgärtner M. Towards patient-specific computational modelling of articular cartilage on the basis of advanced multiparametric MRI techniques.. Sci Rep 2019;9:7172.
  34. Akella SVS, Regatte RR, Wheaton AJ, Borthakur A, Reddy R. Reduction of residual dipolar interaction in cartilage by spin-lock technique.. Magn Reson Med 2004;52:1103-1109.
  35. Schulz E, Speekenbrink M, Krause A. A tutorial on Gaussian process regression: Modelling, exploring, and exploiting functions.. J Math Psychol 2018;85:1-16.

Citations

This article has been cited 1 times.
  1. Zibetti MVW, Menon RG, de Moura HL, Zhang X, Kijowski R, Regatte RR. Updates on Compositional MRI Mapping of the Cartilage: Emerging Techniques and Applications.. J Magn Reson Imaging 2023 Jul;58(1):44-60.
    doi: 10.1002/jmri.28689pubmed: 37010113google scholar: lookup
Subscribe to our newsletter
Join 99,344 horse owners like you who receive our email updates on horse health and nutrition.