Accuracy and reproducibility of four T1 mapping sequences: a head-to-head comparison of MOLLI, ShMOLLI, SASHA, and SAPPHIRE

Background Quantitative myocardial T 1 mapping provides in-vivo tissue characterization for assessment of cardiomyo-pathies. Pre and post-contrast T 1 maps can be used to calculate the extracellular volume fraction (ECV) to detect diffuse myocardial fibrosis. Several imaging approaches have recently been proposed for measuring T 1 values [1-4], but no head-to-head comparison has been reported to cross-examine their accuracy and reproducibility. In this study, we compared both T 1 maps and ECV measurements from the following tech-niques: Modified Look-Locker Inversion Recovery (MOLLI) [1], Shortened MOLLI (ShMOLLI) [2], Saturation recovery single-shot acquisition (SASHA) [3], and SAturation Pulse Prepared Heart rate independent Inversion-REcovery sequence (SAPPHIRE) [4]. on a 1.5 T Phillips scanner using a b-SSFP readout (TR/TE/ = ms/70°, FOV 360 × 337 mm2, voxel 1.9 × 2.5 mm2, slice thickness 8 mm, SENSE factor In a phantom experiment, the four were each repeated 10 times and were compared to the gold standard T 1 measurements obtained using spin echo acquisitions (15 inversion times from 100 ms to 3000 ms). In-vivo analysis experiments was performed in 8 healthy subjects (38 ± 19 y, 4 m), and in 10 patients (56 ± 14 y, 6 m). Pre-contrast imaging was performed twice with

Background Quantitative myocardial T 1 mapping provides in-vivo tissue characterization for assessment of cardiomyopathies. Pre and post-contrast T 1 maps can be used to calculate the extracellular volume fraction (ECV) to detect diffuse myocardial fibrosis. Several imaging approaches have recently been proposed for measuring T 1 values [1][2][3][4], but no head-to-head comparison has been reported to cross-examine their accuracy and reproducibility. In this study, we compared both T 1 maps and ECV measurements from the following techniques: Modified Look-Locker Inversion Recovery (MOLLI) [1], Shortened MOLLI (ShMOLLI) [2], Saturation recovery single-shot acquisition (SASHA) [3], and SAturation Pulse Prepared Heart rate independent Inversion-REcovery sequence (SAPPHIRE) [4].

Methods
The four T 1 mapping methods were implemented on a 1.5 T Phillips scanner using a b-SSFP readout (TR/TE/α = 3.1/1.5 ms/70°, FOV = 360 × 337 mm2, voxel size = 1.9 × 2.5 mm2, slice thickness = 8 mm, SENSE factor = 2). In a phantom experiment, the four methods were each repeated 10 times and were compared to the gold standard T 1 measurements obtained using spin echo acquisitions (15 inversion times from 100 ms to 3000 ms). In-vivo analysis experiments was performed in 8 healthy subjects (38 ± 19 y, 4 m), and in 10 patients (56 ± 14 y, 6 m). Precontrast imaging was performed twice with the four methods. Healthy subjects were removed from the bore between the two pre-contrast scans to simulate a separate exam. Post-contrast T 1 mapping was performed twice at 15 and 30 mins post-injection. T 1 maps were reconstructed offline using an in-house platform and were analyzed by a blinded observer. In all T 1 maps, the septum and the blood pool were manually delineated, and an ECV value was then computed from each pre and post-contrast T 1 map pair. For each method, T 1 measurement variations between the two sets of pre-contrast images and ECV measurement variations generated from the second pre-contrast T 1 and each of the two post-contrast T 1 data were examined.

Results
SASHA and SAPPHIRE were more accurate but less reproducible than MOLLI and ShMOLLI for T 1 mapping in phantom experiments. MOLLI was more reproducible than ShMOLLI and SAPPHIRE was more reproducible than SASHA. There was a trend for MOLLI and ShMOLLI to be more reproducible than SASHA and SAPPHIRE for pre-contrast T 1 mapping in all subjects. There was no statistical significant difference in ECV measurement reproducibility among the four methods in both healthy subjects (One-way ANOVA, p = 0.51) and patients (p = 0.35). However, MOLLI and ShMOLLI yielded large errors in the derived ECV values due to error propagation of T 1 measurements.

Conclusions
Both SASHA and SAPPHIRE T 1 sequences yield excellent accuracy, but with lower reproducibility compare to MOLLI and ShMOLLI. Reproducibility of ECV measurements is similar with all methods, but MOLLI and ShMOLLI demonstrated large systematic errors.
Funding NIH R01EB008743-01A2 Figure 1 Reproducibility of T 1 measurements in phantom containing T 1 samples from 300 ms to 1450 ms. MOLLI and ShMOLLI were less accurate and more reproducible than SASHA and SAPPHIRE. SAPPHIRE was also more reproducible than SASHA while having similar accuracy.