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

BACKGROUND: Machine learning models trained with multiparametric quantitative MRIs (qMRIs) have the potential to provide valuable information about the structural composition of articular cartilage. PURPOSE: To study the performance and feasibility of machine learning models combined with qMRIs for noninvasive assessment of collagen fiber orientation and proteoglycan content. STUDY TYPE: Retrospective, animal model. 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. FIELD STRENGTH/SEQUENCE: A 9.4 T MRI scanner/qMRI sequences: T1 , T2 , adiabatic T1ρ and T2ρ , continuous-wave T1ρ and relaxation along a fictitious field (TRAFF ) maps. ASSESSMENT: 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. STATISTICAL TESTS: 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. RESULTS: Four out of the five models (RF, GB, MLP, and GPR) yielded high accuracy (R2  = 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 (R2  = 0.75 and 0.66) and lowest prediction-error (root mean squared [RMSE] = 1.34 and 2.55) for PLM and DD, respectively. DATA CONCLUSION: Multiparametric qMRIs in combination with regression models can determine cartilage compositional and structural features, with higher accuracy for collagen fiber orientation than proteoglycan content. EVIDENCE LEVEL: 2 TECHNICAL EFFICACY: Stage 2.

Myocardium Assessment by Relaxation along Fictitious Field, Extracellular Volume, Feature Tracking, and Myocardial Strain in Hypertensive Patients with Left Ventricular Hypertrophy.

Background: Previous research has shown impaired global longitudinal strain (GLS) and slightly elevated extracellular volume fraction (ECV) in hypertensive patients with left ventricular hypertrophy (HTN LVH). Up to now, only little attention has been paid to interactions between macromolecules and free water in hypertrophied myocardium. Purpose: To evaluate the feasibility of relaxation along a fictitious field with rank 2 (RAFF2) in HTN LVH patients. Study Type. Single institutional case control. Subjects: 9 HTN LVH (age, 69 ± 10 years) and 11 control subjects (age, 54 ± 12 years). Field Strength/Sequence. Relaxation time mapping (T 1, T 1ρ , and T RAFF2 with 11.8 µT maximum radio frequency field amplitude) was performed at 1.5 T using a Siemens Aera (Erlangen, Germany) scanner equipped with an 18-channel body array coil. Assessment. ECV was calculated using pre- and postcontrast T 1, and global strains parameters were assessed by Segment CMR (Medviso AB Co, Sweden). The parametric maps of T 1ρ and T RAFF2 were computed using a monoexponential model, while the Bloch-McConnell equations were solved numerically to model effect of the chemical exchange during radio frequency pulses. Statistical Tests. Parametric maps were averaged over myocardium for each subject to be used in statistical analysis. Kolmogorov-Smirnov was used as the normality test followed by Student's t-test and Pearson's correlation to determine the difference between the HTN LVH patients and controls along with Hedges' g effect size and the association between variables, respectively. Results: T RAFF2 decreased statistically (83 ± 2 ms vs 88 ± 6 ms, P < 0.031), and global longitudinal strain was impaired (GLS, -14 ± 3 vs - 18 ± 2, P < 0.002) in HTN LVH patients compared to the controls, respectively. Also, significant negative correlation was found between T RAFF2 and GLS (r = -0.53, P < 0.05). Data Conclusion. Our results suggest that T RAFF2 decrease in HTN LVH patients may be explained by gradual collagen accumulation which can be reflected in GLS changes. Most likely, it increases the water proton interactions and consequently decreases T RAFF2 before myocardial scarring.