Improving Accuracy and Reproducibility of Cartilage T₂ Mapping in the OAI Dataset Through Extended Phase Graph...

Area of Science:

• Biomedical Imaging
• Quantitative MRI
• Osteoarthritis Research

Background:

• Osteoarthritis Initiative (OAI) uses Multi-Echo Spin-Echo (MESE) for knee cartilage T2 mapping.
• Current mono-exponential T2 fitting in OAI has limitations, neglecting stimulated echoes and B1 inhomogeneities.
• Extended Phase Graph (EPG) modeling offers a more robust approach but hasn't been applied to OAI data.

Purpose of the Study:

• To evaluate the impact of different T2 fitting methods on cartilage T2 accuracy and reproducibility within the OAI dataset.
• To compare EPG-based modeling against traditional exponential-based approaches for T2 mapping.

Main Methods:

• Retrospective analysis of 100 subjects (50 OA, 50 healthy) from the OAI dataset.
• T2 mapping using 3D, 2D MESE sequences at 3-Tesla.
• Comparison of three EPG-based methods (nonlinear least squares, dictionary matching, deep learning) and three mono-exponential methods (linear least squares, nonlinear least squares, noise-corrected exponential).

Main Results:

• EPG-based methods showed superior T2 accuracy (mean absolute error < 0.5 msec at SNR > 100) versus mono-exponential methods (error > 7 msec).
• EPG approaches demonstrated enhanced reproducibility with narrower limits of agreement (1.5-5 msec).
• EPG methods yielded systematically lower T2 values (10-17 msec difference) compared to mono-exponential fitting.

Conclusions:

• EPG modeling significantly enhances the agreement and reproducibility of knee cartilage T2 mapping in the OAI cohort.
• This advanced modeling technique provides more accurate T2 measurements for osteoarthritis research.