Abstract
Introduction
It was shown that MRI T1ρ images could be used to assess proteoglycan loss; therefore, it can be employed as a quantitative measure of intervertebral discs (IVD) degeneration, especially in early stages. To use T1ρ as a diagnostic or prognostic tool, however, one needs to assess the reproducibility of these measurements. Moreover, earlier studies showed diurnal changes in the IVDs during rest and loading conditions. Therefore, the goal of this work was to assess the reproducibility and diurnal changes in T1ρ measurements in the IVDs.
Material and Methods
This study was approved by the IRB and written consents were obtained from 12 adults (Age: Median [min–max]: 28.5 [20–41] years; F: 5, M: 7). Images were acquired using 3T MRI with CTL-spine coil. 3D MAPSS pulse sequence with spin-lock times (TSL) = [0, 20, 40, 60] ms, spin-lock amplitude = 400 Hz, and TR/TE = 6.5/1.6 ms, FOV = 310 mm and 3 mm/16-sagittal slices was used. T1ρ values were calculated using monoexponential fitting. For reproducibility analysis, the follow-up scans were acquired with a median delay of 112 days and at a different time of day (ToD). For analysis, an operator drew regions of interest (ROI) for nucleus pulposus (NP) and calculated the average (Fig. 1). Data from 59 lumbar discs were used (one disc excluded due to a local artifact). Diurnal changes in T1ρ measurements in the NPs were analyzed using SPSSv21 with α = 0.05. Two variables were defined for analysis: (1) ToD difference between the two scans in hours (Median [min, max]: 2.2 [−3.6, 6.0] h). (2) T1ρ difference between two scans in milliseconds (mean ± SD: −0.093 ± 6.83 ms). Generalized estimating equations (GEE) were employed. To account for the day difference between the repeated scans, it was included in the model. To investigate reproducibility: (1) Bland–Altman plot was generated, (2) repeatability coefficient, RC (1.96·sqrt[2· σ2]); coefficient of variation, CV (σ/µ); and intraclass correlation coefficient, ICC (τ2/[τ2 + σ2]) were calculated. σ2 is the variability attributed to measurement error, µ is the grand average of measurements, and τ2 is the between subject disc variability.
Results
The time of day was a significant factor (p = 0.035) in T1ρ measurements. However, the number of days between the repeated scans was not significant (p = 0.852). Fig. 2 shows the difference between the repeated T1ρ measurements plotted against the number of days between the scans. Interestingly, T1ρ values in the NP increased during the day. The Bland–Altman plot is shown in Fig. 3. Calculated repeatability coefficients are RC = 13.4 ms, CV = 0.057, and ICC = 0.90.
Conclusion
Diurnal changes were a small but significant factor in T1ρ measurements. Moreover, there were no significant changes in T1ρ measurements in the repeated scans since the intervals between the longitudinal scans were relatively short compared with the typical time scale of typical disc degeneration processes. The CV of Bland–Altman test showed that the standard deviation of repeated T1ρ measurements was approximately 5.7% of the average T1ρ value. Moreover, ICC parameter showed that the variation in repeated T1ρ measurements was small compared with between subject variations.
Acknowledgment
This study is supported by AOSpine research grant.
Flow diagram.
ΔT1ρ versus ΔToD.
T1ρ Bland–Altman plot.