Characterization of T₁ bias from lipids in MOLLI and SASHA pulse sequences

Background

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[3] 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.

Methods

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 [8]. 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 [9]. 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.

Results

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 [10].

Figure 1

SASHA 2p, SASHA 3p, and MOLLI T₁ values vs off-resonance for Simulations, Phantoms, and In-Vivo Data. Black vertical lines on the SASHA 2p simulation plot span 90Hz.

Conclusions

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].