Xinran Zhong^{1,2}, Sepideh Shakeri^{1}, Dapeng Liu^{1}, James Sayre^{1}, Steven S. Raman^{1}, Holden H. Wu^{1,2}, and Kyunghyun Sung^{1,2}

Reliable pre-contrast T_{1} estimation is crucial for quantitative DCE MRI. Variable flip angle is widely used for pre-contrast T_{1} measurements and is sensitive to B_{1}^{+} inhomogeneity. Although various B_{1}^{+} techniques have been proposed, the application of B_{1}^{+} compensation is not widely accepted yet. In this study, by evaluating T_{1} intra-scanner and inter-scanner consistency with and without B_{1}^{+} compensation, we confirmed the necessity to perform B_{1}^{+} compensation and a B_{1}^{+} estimation method named reference region variable flip angle (RR-VFA) is recommended due to its consistent T_{1} estimation and wide availability.

With IRB approval, twenty-one volunteers (27±4 years old) were prospectively recruited to assess intra- and inter-scanner variabilities. Each subject was scanned twice on two 3.0T scanners (Skyra – Scanner 1 and Prisma – Scanner 2, Siemens Healthineers) resulting in total 84 scans, as summarized in Fig. 1. The imaging protocol consisted of 2D T2-weighted (T2W) turbo spin echo (TSE) sequence, 2D satTFL sequence for B_{1}^{+} estimation and 3D VFA T_{1} weighted sequences with a dual-echo readout for both T_{1} estimation and RR-VFA B_{1}^{+} estimation. After processing, each scan resulted in one anatomical T2W image, two B_{1}^{+} maps and three T_{1} images. We mainly evaluated the three T_{1} maps, including VFA T_{1} map (T_{1,VFA}), T_{1} map with satTFL B_{1}^{+} compensation (T_{1,satTFL}) and T_{1} map with RR-VFA B_{1}^{+} compensation (T_{1,RR-VFA}).

To account for the position mismatch between the scans for each volunteer, rigid registration was performed for each volunteer based on T2W images and applied to corresponding B_{1}^{+} and T_{1} maps. An experienced radiologist delineated volumetric regions of interest (ROIs) on T2W images in the prostate, left pelvic muscle and right pelvic muscle. The ROIs were transferred to T_{1} maps and the average T_{1} values within the ROIs were recorded for later comparison.

Lin’s Concordance Correlation Coefficient (CCC) was used to evaluate both intra- and inter-scanner T_{1} consistency. There were in total two pairs of intra-scanner comparisons and four pairs of inter-scanner comparisons. For each pair of comparison, CCC was calculated and then averaged based on intra- or inter-scanner comparison respectively. To visualize the T_{1} estimation consistency, linear regression plots as well as Bland-Altman plots were shown for one representative pair of comparison, and the Pearson’s correlation and limits of agreements were recorded.

One example slice of T_{1} within the ROIs overlaid on T2W images were shown in Fig. 2. The figure first demonstrated that the image registration between scans achieved reasonable results, making applying the same set of ROIs on four scans at the same time feasible. More importantly, T_{1, VFA} within prostate ROI exhibited large variance across four scans and T_{1,VFA} within the left and right pelvic muscles ROIs had inconsistent T_{1} values for the same scan. T_{1,satTFL} and T_{1,RR-VFA} on the other hand achieved more consistent T_{1} estimation both between scans and within the same scan.

Linear regression plots in Fig. 3 and Bland Altman plots in Fig. 4 also confirmed this observation. In Fig. 3, T_{1,RR-VFA} had the regression slope closer to 1 and larger Pearson’s correlation square r^{2} compare to T_{1,VFA }and T_{1,satTFL}. Similarly, the Bland Altman plots in Fig. 4 showed that the T_{1} estimation for each tissue becomes more consistent for B_{1}^{+} corrected T_{1} maps (T_{1,satTFL} and T_{1,RR-VFA}) compared to T_{1,VFA}. The 95% limits of agreement of B_{1}^{+} corrected T_{1} were narrower than those in T_{1,VFA}.

The averaged CCC and standard deviation were reported in Table 1. T_{1, RR-VFA} had higher CCC (0.914 for intra-scanner and 0.897 for inter-scanner comparison) compared to T_{1,satTFL} (0.906 for intra-scanner and 0.880 for inter-scanner comparison) and T_{1,VFA} (0.881 for intra-scanner and 0.795 for inter-scanner comparison).

1. Alonzi R, Padhani AR, Allen C. Dynamic contrast enhanced MRI in prostate cancer. European Journal of Radiology. 2007;63(3):335–350.

2. Huang W, Chen Y, Fedorov A, Li X, Jajamovich GH, Malyarenko DI, Aryal MP, LaViolette PS, Oborski MJ, O’Sullivan F, et al. The Impact of Arterial Input Function Determination Variations on Prostate Dynamic Contrast-Enhanced Magnetic Resonance Imaging Pharmacokinetic Modeling: A Multicenter Data Analysis Challenge. Tomography. 2016;2(1):56–66.

3. Tsai W-C, Kao K-J, Chang K-M, Hung C-F, Yang Q, Lin C-YE, Chen C-M. B1 Field Correction of T1 Estimation Should Be Considered for Breast Dynamic Contrast-enhanced MR Imaging Even at 1.5 T. Radiology. 2017;282(1):55–62.

4. Chung S, Kim D, Breton E, Axel L. Rapid B1+ mapping using a preconditioning RF pulse with turboFLASH readout. Magnetic Resonance in Medicine. 2010;64(2):439–446.

5. Sung K, Saranathan M, Daniel BL, Hargreaves BA. Simultaneous T 1 and B 1 + Mapping Using Reference Region Variable Flip Angle Imaging. Magnetic Resonance in Medicine. 2013;70(4):954–961.

6. Rangwala NA, Dregely I, Wu HH, Sung K. Optimization and evaluation of reference region variable flip angle (RR-VFA) B1+ and T 1 Mapping in the Prostate at 3T. Journal of Magnetic Resonance Imaging. 2017;45(3):751–760.

Figure 1. Experiment design summary. Each volunteer was scanned four times on two 3.0 T scanners. The volunteers were repositioned between two scans on the same scanner, and the time interval between scans on two scanners varies from same-day to 70 days. Each scan consists of three different sequences and after post-processing, we obtained one T2W image, two B_{1}^{+} maps and three T_{1} maps to analyze.

Figure 2. T_{1} value within ROIs overlaid on T2W image after registration on one representative slice. Each row represents one T_{1} map, and each column represents one scan. After registration, four scans were aligned, and the same set of ROIs can be applied to four scans at the same time. With B_{1}^{+} compensation, T_{1,satTFL} and T_{1,RR-VFA} had more consistent T_{1} value in the prostate across different scans, and the T_{1} value of left and right pelvic muscles for each scan became more uniform compared to T_{1,VFA}.

Figure 3. Linear Regression and squared Pearson’s Correlation (r^{2}) for both intra-scanner comparison between Scan 1 and Scan 2 on Scanner 1 (a-c) and inter-scanner comparison between Scan 1 on Scanner 1 and Scan 1 on Scanner 2 (d-f) of average T_{1} within each ROI. Each encoded color indicates one ROI. T_{1,RR-VFA} has the highest r^{2}.

Figure 4. Bland Altman plots for both intra-scanner comparison between Scan 1 and Scan 2 on Scanner 1 (a-c) and inter-scanner comparison between Scan 1 on Scanner 1 and Scan 1 on Scanner 2 (d-f) of average T_{1} within each ROI. Each encoded color indicates one ROI. T_{1,satTFL} and T_{1,RR-VFA} have a more uniform T_{1} value for each tissue, and a narrower limits of agreement compared to T_{1,VFA}.

Table 1. Average CCC and standard deviation for intra-scanner and inter-scanner comparison for T_{1,VFA}, T_{1,satTFL} and T_{1,RR-VFA}. T_{1,RR-VFA} had the highest CCC for both intra-scanner and inter-scanner comparison.