Characterization of T1 bias from lipids in MOLLI and SASHA pulse sequences
Journal of Cardiovascular Magnetic Resonance volume 17, Article number: W10 (2015)
Increased myocardial T1 values are associated with fibrosis and edema, while decreased values in Fabry disease have been attributed to the short T1 of infiltrative lipids [1, 2]. The relationship between lipid concentration (LC) and best-fit T1 values is unknown. This study aims to determine the dependence of MOLLI and SASHA T1 values on LC.
MOLLI and SASHA T1 mapping sequences are based on the bSSFP acquisition. bSSFP signal yield as a function of off-resonance frequency is well characterized [4, 5], with phase inversion in sequential "bSSFP bands" and a profile shift as a function of resonance frequency, resulting in constructive/destructive interference between water and fat [6, 7]. We hypothesized that lipids may decrease or increase T1 values as a function of off-resonance frequency.
Bloch equation simulations of MOLLI and SASHA for 0:2:10% LC incorporated exact pulse sequence parameters including slice profiles and an accurate fat spectral line shape.
MOLLI and SASHA acquisitions (identical to simulations) were repeated 50 times, spanning 450 Hz of off-resonance (1.25 bSSFP bands) in both phantoms (LC of 0.5-10%), and in-vivo in three calf muscle regions with different LC . Acquisition Parameters: 1.5T Siemens Sonata, single-shot bSSFP, 1.35/2.7ms TE/TR, 192x72 matrix, 360x270 mm FOV, 70° SASHA flip, 35° MOLLI flip . T1 values were calculated using standard Look-Locker correction (MOLLI) or 2 and 3 parameter exponential models (SASHA) at each frequency increment across the bSSFP band.
MOLLI and SASHA T1 values have an asymmetric relationship with off-resonance, with larger positive and negative biases with larger LC (Fig. 1). Over a small ±45 Hz range, a 1% LC gives rise to a T1 bias ranging from -39 to +26ms (MOLLI), -19 to +16ms (SASHA 3p) and -44 to +25ms (SASHA 2p). The location of the cross-over point is a function of field-strength and TR; these findings are specific to 1.5T and TR=2.7ms. MOLLI T1 values have an additional intrinsic dependence on off-resonance resulting in an underlying domed shape .
Relatively low LC results in clinically relevant negative or positive shifts in tissue T1 over a narrow range of off-resonance frequencies with MOLLI and SASHA. Thus, increased or decreased native T1 values can potentially be ascribed to lipids, which can confound underlying increased water T1 values ascribed to fibrosis or edema and complicate the use of T1 mapping for indirect identification of lipids via reduced T1 values [1, 2].
Sado DM: Circ Cardiovasc Imag. 2013, 6: 392-10.1161/CIRCIMAGING.112.000070.
Thompson RB: Circ Cardiovasc Imag. 2013, 6: 637-10.1161/CIRCIMAGING.113.000482.
Chow K: Magn Reson Med. 2014, 71: 2082-10.1002/mrm.24878.
Carr H: Phys Rev. 1958, 112: 1693-10.1103/PhysRev.112.1693.
Miller KL: Magn Reson Med. 2010, 63: 385-10.1002/mrm.22212.
Aquaro GD: J Magn Reson Imag. 2014, 40: 126-10.1002/jmri.24335.
Bley TA: J Magn Reson Imag. 2010, 31: 4-10.1002/jmri.21895.
Machann J: Magn Reson Imag. 2003, 17: 350-10.1002/jmri.10255.
Chow K: JCMR. 2014, 16: M9-
Kellman P: J Cardiovasc Magn Reson. 2013, 22: 15-
About this article
Cite this article
Thiesson, S.B., Thompson, R.B. & Chow, K. Characterization of T1 bias from lipids in MOLLI and SASHA pulse sequences. J Cardiovasc Magn Reson 17 (Suppl 1), W10 (2015). https://doi.org/10.1186/1532-429X-17-S1-W10