ABSTRACT
BACKGROUND: Articular cartilage degeneration is the hallmark change of osteoarthritis, a severely disabling disease with high prevalence and considerable socioeconomic and individual burden. Early, potentially reversible cartilage degeneration is characterized by distinct changes in cartilage composition and ultrastructure, while the tissue's morphology remains largely unaltered. Hence, early degenerative changes may not be diagnosed by clinical standard diagnostic tools. METHODS: Against this background, this study introduces a novel method to determine the tissue composition non-invasively. Our method involves quantitative MRI parameters (i.e., T1, T1ρ, T2 and [Formula: see text] maps), compositional reference measurements (i.e., microspectroscopically determined local proteoglycan [PG] and collagen [CO] contents) and machine learning techniques (i.e., artificial neural networks [ANNs] and multivariate linear models [MLMs]) on 17 histologically grossly intact human cartilage samples. RESULTS: Accuracy and precision were higher in ANN-based predictions than in MLM-based predictions and moderate-to-strong correlations were found between measured and predicted compositional parameters. CONCLUSION: Once trained for the clinical setting, advanced machine learning techniques, in particular ANNs, may be used to non-invasively determine compositional features of cartilage based on quantitative MRI parameters with potential implications for the diagnosis of (early) degeneration and for the monitoring of therapeutic outcomes.
Subject(s)
Cartilage, Articular/diagnostic imaging , Machine Learning , Multiparametric Magnetic Resonance Imaging , Osteoarthritis, Knee/diagnostic imaging , Spectroscopy, Fourier Transform Infrared , Adult , Aged , Aged, 80 and over , Cartilage, Articular/metabolism , Cartilage, Articular/pathology , Female , Humans , Male , Middle Aged , Osteoarthritis, Knee/metabolism , Osteoarthritis, Knee/pathologyABSTRACT
Quantitative magnetic resonance imaging (qMRI) is a promising approach to detect early cartilage degeneration. However, there is no consensus on which cartilage component contributes to the tissue's qMRI signal properties. T1, T1ρ, and T2â maps of cartilage samples (n = 8) were generated on a clinical 3.0-T MRI system. All samples underwent histological assessment to ensure structural integrity. For cross-referencing, a discretized numerical model capturing distinct compositional and structural tissue properties, that is, fluid fraction (FF), proteoglycan (PG) and collagen (CO) content and collagen fiber orientation (CFO), was implemented. In a pixel-wise and region-specific manner (central versus peripheral region), qMRI parameter values and modelled tissue parameters were correlated and quantified in terms of Spearman's correlation coefficient ρs. Significant correlations were found between modelled compositional parameters and T1 and T2â, in particular in the central region (T1: ρs ≥ 0.7 [FF, CFO], ρs ≤ -0.8 [CO, PG]; T2â: ρs ≥ 0.67 [FF, CFO], ρs ≤ -0.71 [CO, PG]). For T1ρ, correlations were considerably weaker and fewer (0.16 ≤ ρs ≤ -0.15). QMRI parameters are characterized in their biophysical properties and their sensitivity and specificity profiles in a basic scientific context. Although none of these is specific towards any particular cartilage constituent, T1 and T2â reflect actual tissue compositional features more closely than T1ρ.