To evaluate the accuracy and precision of variable flip angle (VFA) T₁ estimates in the breast before and after correcting for B₁ inhomogeneities using the Bloch-Siegert shift.¹ Because the VFA technique yields rapid, high-resolution T₁ maps², it is often used in clinical applications of quantitative DCE-MRI data of the breast.³ Unfortunately, the accuracy of the VFA-derived T₁ values is affected by B₁ transmit inhomogeneities², which can be substantial in breast imaging at 3T.⁴ We previously reported preliminary data suggesting the Bloch-Siegert B₁ mapping technique improved the accuracy of VFA T₁ measurements in phantoms and in the breast of a healthy volunteer.⁵ We have extended the study to further test the hypothesis that Bloch-Siegert B₁ mapping improves the accuracy of VFA T₁ measurements of the breast in a cohort of healthy volunteers. To test the hypothesis that Bloch-Siegert B₁ mapping improved measurement precision, the same cohort was enrolled in a test-retest study. Test-retest MRI sessions were performed using a 3T Philips Achieva MR scanner equipped with a two-channel body coil and a 16-channel receive double-breast coil (MammoTrak, Philips Healthcare, Best, The Netherlands) on 16 women (mean: 44 years, range: 25-67) with no history of breast disease. Each scanning session lasted approximately 30 minutes with a 10-minute rest period in between. T₁ was measured from 3D spoiled gradient echo images with multiple flip angles (10 flip angles = 2,4,6…20°; matrix = 192×192; FOV = 256×256×50 mm³; 10 slices, TR/TE = 7.9/4.6 ms). B₁ field variations were measured using the Bloch-Siegert method with a 2 ms frequency-swept B₁ phase encoding pulse⁶ with matched slices (RMS B₁ = 2.29 mT; matrix = 104×102; TR/TE = 491/5.4 ms). As a gold standard, T₁ was also measured using a single-slice inversion recovery (IR) sequence (12 inversion times between 25-10,000 ms; matrix = 128×91; FOV = 256×256 mm²). Fat and fibroglandular tissue (FGT) were segmented from the IR image where the signal intensity of the FGT was null, and mean T₁ values from both tissue regions were calculated for each imaging technique and scan session. The effect of B₁ correction on accuracy was evaluated by calculating the percent error (%err), concordance correlation coefficient (CCC), and bootstrap 95% confidence interval⁷ (CI) for the mean differences in absolute deviation between IR- and VFA-derived T₁ values (with and without B₁ correction). For reproducibility, the 95% CI of the mean difference, within-subject standard deviation (wSD), and repeatability coefficient (r) of the test-retest sessions were calculated.Figure 1 is a representative test-retest set of T₁ maps generated from IR (left column), uncorrected VFA (center column), and B₁-corrected VFA (right column). Large spatial variations in T₁ values of the FGT are observed in the uncorrected VFA maps, which are minimized after B₁ correction. After B₁ correction, %err decreased from 20% to 9% and CCC increased from 0.55 to 0.83. Similar trends in accuracy were observed in the fat (see Table 1). The bootstrap 95% CIs for FGT and fat are 57.8 ms to 139 ms and 17.2 ms to 42.2 ms, respectively; both values for each CI are positive which indicates that the absolute deviation from IR is smaller after B₁ correction for both ROIs. Figure 2 shows Bland-Altman plots for each imaging technique and tissue. After B₁ correction, the 95% CI, wSD, and r decreased from ±94 ms to ±38 ms, 100 ms to 40 ms, 276 ms to 111 ms, respectively. Similar trends were again observed in the fat (see Table 1). The Bloch-Siegert method of B₁ mapping improves the accuracy of T₁ measurements derived from VFA data in the breast. Previous simulation results reported large errors in DCE-MRI parameters (e.g., K^trans) due to inaccuracies in T₁.⁸ Therefore, improvements in accuracy are imperative in order to lower the error associated with the precontrast T₁ measurement in quantitative analyses of DCE-MRI data. Reproducibility experiments are important in order to define the associated variability in a measurement so that future studies can be designed and powered appropriately. Our data show that Bloch-Siegert B₁ mapping improves the reproducibility of VFA-derived T₁ measurements in the breast thereby improving measurement precision. These data, combined with other preliminary reports^5,9, indicate that B₁ mapping using the Bloch-Siegert method is an attractive option for accurate and precise VFA-derived measurements of T₁ in the breast at 3T. Furthermore, our protocol acquires 3D T₁ and B₁ maps in less than three minutes which is ideal for clinical applications. Future work includes employing the B₁ mapping method described herein in an ongoing clinical study³ of breast cancer patients undergoing neoadjuvant chemotherapy.