Comparison of test-retest reproducibility of DESPOT and 3D-QALAS for waterT1andT2mapping

Abstract

AbstractPurposeRelaxometry, specificallyT1andT2mapping, has become an essential technique for assessing the properties of biological tissues related to various physiological and pathological conditions. Many techniques are being used to estimateT1andT2relaxation times, ranging from the traditional inversion or saturation recovery and spin-echo sequences to more advanced methods. Choosing the appropriate method for a specific application is critical since the precision and accuracy ofT1andT2measurements are influenced by a variety of factors including the pulse sequence and its parameters, the inherent properties of the tissue being examined, the MRI hardware, and the image reconstruction. The aim of this study is to evaluate and compare the test-retest reproducibility of two advanced MRI relaxometry techniques (Driven Equilibrium Single Pulse Observation ofT1andT2, DESPOT, and 3D Quantification using an interleaved Look-Locker acquisition Sequence with aT2preparation pulse, QALAS), forT1andT2mapping in a healthy volunteer cohort.Methods10 healthy volunteers underwent brain MRI at 1.3 mm3isotropic resolution, acquiring DESPOT and QALAS data (∼11.8 and ∼5 minutes duration, including field maps, respectively), test-retest with subject repositioning, on a 3.0 Tesla Philips Ingenia Elition scanner. To reconstruct theT1andT2maps, we used an equation-based algorithm for DESPOT and a dictionary-based algorithm that incorporates inversion efficiency andB1-field inhomogeneity for QALAS. The test-retest reproducibility was assessed using the coefficient of variation (CoV), intraclass correlation coefficient (ICC) and Bland-Altman plots.ResultsOur results indicate that both the DESPOT and QALAS techniques demonstrate good levels of test-retest reproducibility forT1andT2mapping across the brain. Higher whole-brain voxel-to-voxel ICCs are observed in QALAS forT1(0.84 ± 0.039) and in DESPOT forT2(0.897 ± 0.029). The Bland-Altman plots show smaller bias and variability ofT1estimates for QALAS (mean of -0.02 s, and upper and lower limits of -0.14 and 0.11 s, 95% CI) than for DESPOT (mean of -0.02 s, and limits of -0.31 and 0.27 s). QALAS also showed less variability (mean 1.08 ms, limits –1.88 to 4.04 ms) forT2compared to DESPOT (mean of 2.56 ms, and limits -17.29 to 22.41 ms). The within-subject CoVs for QALAS range from 0.6% (T2in CSF) to 5.8% (T2in GM), while for DESPOT they range from 2.1% (T2in CSF) to 6.7% (T2in GM). The between-subject CoVs for QALAS range from 2.5% (T2in GM) to 12% (T2in CSF), and for DESPOT they range from 3.7% (T2in WM) to 9.3% (T2in CSF).ConclusionOverall, QALAS demonstrated better reproducibility forT1andT2measurements than DESPOT, in addition to reduced acquisition time.