Same-model and cross-model variability in knee cartilage thickness measurements using 3D MRI systems

Area of Science:

• Biomedical Engineering
• Radiology
• Orthopedics

Background:

• Automated three-dimensional (3D) analysis of knee cartilage using Magnetic Resonance Imaging (MRI) is advancing.
• Scanner variability is a critical challenge for multi-center clinical applications of MRI-based cartilage measurements.
• Quantifying this variability is essential for reliable longitudinal studies and multi-center collaborations.

Purpose of the Study:

• To quantify and compare same-model variability (intra-scanner) and cross-model variability (inter-scanner) in knee cartilage thickness measurements.
• To assess variability across MRI scanners from five different manufacturers (Canon, Fujifilm, GE, Philips, Siemens).
• To evaluate the impact of a specific 3D volume analysis software on measurement variability.

Main Methods:

• Ten healthy volunteers underwent two 3T MRI scans of their right knee on systems from five manufacturers.
• Standard imaging protocols included fat-suppressed spoiled gradient echo and proton density weighted sequences.
• A deep learning-based 3D volume analysis software automatically segmented knee cartilage into 7 subregions.

Main Results:

• Same-model variability: 82% of measurements showed ≤0.10 mm variability; 98% showed ≤0.20 mm.
• Cross-model variability: 51% of measurements showed ≤0.10 mm variability; 84% showed ≤0.20 mm.
• Mean same-model variability (0.06 ± 0.05 mm) was significantly lower than cross-model variability (0.11 ± 0.09 mm) (p < 0.001).

Conclusions:

• Knee cartilage thickness measurements show significantly higher variability across different MRI systems compared to repeated scans on the same system.
• This variability is software-dependent and has critical implications for multi-center studies and longitudinal assessments.
• Standardization or careful consideration of scanner differences is crucial for accurate MRI-based knee cartilage analysis.