Abstract
OBJECTIVE: Quantitative magnetic resonance imaging (qMRI) parameters such as T1ρ and T2* characterize changes in cartilage matrix composition prior to cartilage loss. However, T1ρ and T2* measurement reliability in phantoms that mimic in vivo knee size and position within the scanner bore (left or right side of isocenter) is unknown. This study aimed to quantify sources of variance, and the systematic differences between left and right positions in T1ρ and T2* in small vials of a commercially available phantom and a knee-sized phantom. METHODS: Phantoms were imaged 100 mm to the left and right of isocenter on five days using 3D MAPSS (T1ρ) and 3D gradient echo (T2*) sequences at 3T. Variance component analysis estimated the variability attributable to slice, side, vial, and day. Measurement error was quantified with 95% confidence intervals (CI). Paired t-tests evaluated systematic differences between the left and right imaging locations (p<.05). RESULTS: Averaged across days the left-to-right phantom positions produced differences of 0.1ms (p<.0001) for T1ρ and -0.8ms (p<.0001) for T2* in the knee-sized phantoms, and differences ranged from -1.5±0.3ms (p=0.0004) to -0.4±0.4ms (p=0.045) for T1ρ and -3.3±2.5ms (p<.0001) to 2.3±1.5ms (p<.0001) for T2* in the small vial phantoms. The total variance for T2* was much less for the knee-sized phantom (0.9) than the small-vial pairs with the same concentration (7.0). CONCLUSIONS: This study revealed a position and phantom size dependence on qMRI parameters not reported previously. Future studies and quality control acquisitions should consider position within the MR scanner and phantom size in their design.