Impact of uncertainty in longitudinal T1 measurements on quantification of dynamic contrast‐enhanced MRI

Abstract

The objective of this study was to assess the uncertainty in T1 measurement, by estimating the repeatability coefficient (RC) from two repeated scans, in normal appearing brain tissues employing two different T1 mapping methods. All brain MRI scans were performed on a 3 T MR scanner in 10 patients who had low grade/benign tumors and partial brain radiation therapy (RT) without chemotherapy, at pre-RT, 3 weeks into RT, end RT (6 weeks) and 11, 33, and 85 weeks after RT. T1-weighted images were acquired using (1) a spoiled gradient echo sequence with two flip angles (2FA: 5° and 15°) and (2) a progressive saturation recovery sequence (pSR) with five different TR values (100-2000 ms). Manually drawn volumes of interest (VOIs) included left and right normal putamen and thalamus in gray matter, and frontal and parietal white matter, which were distant from tumors and received a total of accumulated radiation doses less than 5 Gy at 3 weeks. No significant changes or even trends in mean T1 from pre-RT to 3 weeks into RT in these VOIs (p ≥ 0.11, Wilcoxon sign test) allowed us to calculate the repeatability statistics of between-subject means of squares, within-subject means of squares, F-score, and RC. The 2FA method produced RCs in the range of (9.7-11.7)% in gray matter and (12.2-14.5)% in white matter; while the pSR method led to RCs ranging from 10.9 to 17.9% in gray matter and 7.5 to 10.3% in white matter. The overall mean (±SD) RCs produced by the two methods, 12.0 (±1.6)% for 2FA and 12.0 (±3.8)% for pSR, were not significantly different (p = 0.97). A similar repeatability in T1 measurement produced by the time efficient 2FA method compared with the time consuming pSR method demonstrates that the 2FA method is desirable to integrate into dynamic contrast-enhanced MRI for rapid acquisition.