Diagnosing osteoarthritis from T2 maps using deep learning: an analysis of the entire Osteoarthritis Initiative baseline cohort

V Pedoia; J Lee; B Norman; T M Link; S Majumdar

doi:10.1016/j.joca.2019.02.800

Diagnosing osteoarthritis from T₂ maps using deep learning: an analysis of the entire Osteoarthritis Initiative baseline cohort

Osteoarthritis Cartilage. 2019 Jul;27(7):1002-1010. doi: 10.1016/j.joca.2019.02.800. Epub 2019 Mar 21.

Authors

V Pedoia¹, J Lee², B Norman³, T M Link⁴, S Majumdar⁵

Affiliations

¹ Department of Radiology and Biomedical Imaging, University of California, San Francisco, USA; Center of Digital Health Innovation (CDHI), USA. Electronic address: valentina.pedoia@ucsf.edu.
² Department of Radiology and Biomedical Imaging, University of California, San Francisco, USA. Electronic address: Jinhee.Lee@ucsf.edu.
³ Department of Radiology and Biomedical Imaging, University of California, San Francisco, USA. Electronic address: berknorman@me.com.
⁴ Department of Radiology and Biomedical Imaging, University of California, San Francisco, USA. Electronic address: Thomas.Link@ucsf.edu.
⁵ Department of Radiology and Biomedical Imaging, University of California, San Francisco, USA; Center of Digital Health Innovation (CDHI), USA. Electronic address: Sharmila.Majumdar@ucsf.edu.

Abstract

Objective: We aim to study to what extent conventional and deep-learning-based T₂ relaxometry patterns are able to distinguish between knees with and without radiographic osteoarthritis (OA).

Methods: T₂ relaxation time maps were analyzed for 4,384 subjects from the baseline Osteoarthritis Initiative (OAI) Dataset. Voxel Based Relaxometry (VBR) was used for automatic quantification and voxel-based analysis of the differences in T₂ between subjects with and without radiographic OA. A Densely Connected Convolutional Neural Network (DenseNet) was trained to diagnose OA from T₂ data. For comparison, more classical feature extraction techniques and shallow classifiers were used to benchmark the performance of our algorithm's results. Deep and shallow models were evaluated with and without the inclusion of risk factors. Sensitivity and Specificity values and McNemar test were used to compare the performance of the different classifiers.

Results: The best shallow model was obtained when the first ten Principal Components, demographics and pain score were included as features (AUC = 77.77%, Sensitivity = 67.01%, Specificity = 71.79%). In comparison, DenseNet trained on raw T₂ data obtained AUC = 83.44%, Sensitivity = 76.99%, Specificity = 77.94%. McNemar test on two misclassified proportions form the shallow and deep model showed that the boost in performance was statistically significant (McNemar's chi-squared = 10.33, degree of freedom (DF) = 1, P-value = 0.0013).

Conclusion: In this study, we presented a Magnetic Resonance Imaging (MRI)-based data-driven platform using T₂ measurements to characterize radiographic OA. Our results showed that feature learning from T₂ maps has potential in uncovering information that can potentially better diagnose OA than simple averages or linear patterns decomposition.

Keywords: Convolutional neural network; Deep learning; Quantitative MRI; T(2) relaxation times; Voxel based relaxometry.

Publication types

Research Support, N.I.H., Extramural
Review

MeSH terms

Aged
Algorithms
Cartilage, Articular / pathology
Case-Control Studies
Deep Learning*
Disease Progression*
Female
Humans
Magnetic Resonance Imaging / methods*
Male
Middle Aged
Osteoarthritis, Knee / diagnostic imaging*
Osteoarthritis, Knee / epidemiology
Osteoarthritis, Knee / pathology*
Pain Measurement*
Sensitivity and Specificity
Severity of Illness Index

Abstract

Publication types

MeSH terms

Grants and funding