Table 1.
Imaging parameters for the MRI sequences.
Fig 1.
MR images of the knee and cartilage thickness mapping obtained using MRI systems from different manufacturers.
(a) Sagittal MR images of the knee taken with MRI machines from five different companies. Representative examples were selected from each company. (b) MRI cartilage thickness mapping and regions of the knee. The cartilage thickness was provided using a color scale, with thicker areas indicated in white and thinner areas indicated in red. The region of interest is indicated by a red line, and the femoral cartilage is divided into four regions by halving it vertically and horizontally. PMF, posterior medial femoral; PLF, posterior lateral femoral; AMF, anterior medial femoral; ALF, anterior lateral femoral; MT, medial tibial; LT, lateral tibial; P, patellar.
Fig 2.
Combination of same-model variability and cross-model variability.
For one knee, MRI scans were taken twice using the same MRI model. These scans were repeated for the five different MRI models. The absolute difference in cartilage thickness between the first and second scans on the same model was defined as same-model variability. This resulted in five combinations for one region of one knee. The absolute difference between measurements on one model and measurements on other models was defined as the cross-model variability. This resulted in 40 combinations for one region of one knee.
Fig 3.
Cartilage thickness measured at the first and second time points.
MRI scans were performed on 10 subjects using five different MRI models. Each subject was represented by a standardized symbol. Each MRI model was displayed in its own corporate color. Measurements were shown for seven knee regions. The data were plotted separately for the first and second scans. This yielded a total of 350 plots.
Fig 4.
Scatter plots of same-model variability and cross-model variability.
(a) Scatter plot of same-model variability. Each subject was represented by a standardized symbol. Each MRI model was displayed in its own corporate color. Measurements are shown for seven regions. There were five combinations for each region and subject, resulting in 50 plots per region. The seven knee regions gave a total of 350 plots. (b) Scatter plot of cross-model variability. For each region, the cross-model variability was plotted as a single vertical bar, without distinguishing among the 10 subjects and five MRI models. This yielded 40 combinations for each region and subject, resulting in 400 plots per region. The seven knee regions gave a total of 2,800 plots.
Fig 5.
Distribution of same-model variability and cross-model variability.
The number of measurements is shown above each bar in the graph. For example, a variability of “(0.08, 0.10) mm” indicates 0.08 mm < variability ≤ 0.10 mm. For the same-model variability, the variability was ≤ 0.10 mm in 84% of the 350 measurements and ≤ 0.20 mm in 98%. For the cross-model variability, the variability was ≤ 0.10 mm in 67% of the 2800 measurements and ≤ 0.20 mm in 83%.