T₂-dependent errors in MOLLI T₁ values: simulations, phantoms, and in-vivo studies

Background

Diffuse myocardial fibrosis occurs in various cardiomyopathies and can be indirectly assessed with blood and myocardial T₁ mapping at baseline and after gadolinium administration. The widely used MOdified Look-Locker Inversion-recovery (MOLLI) [1] sequence is known to underestimate myocardial T₁ at higher heart rates, but its dependence on T₂ has not been explored. We investigate MOLLI’s T₁ accuracy in phantoms and confirm with simulations and in-vivo studies. T₁ values are further compared with a saturation-recovery T₁ mapping sequence [2].

Methods

Phantoms

14 NiCl₂ agarose phantoms with a broad range of T₁ and T₂ values were imaged with a gold-standard inversion-recovery spin-echo (IR-SE) sequence, MOLLI, and a new SAturation-recovery single-SHot Acquisition (SASHA) technique (Siemens Avanto 1.5T). IR-SE: 16 TIs 100-5000ms, TE=11ms, TR>5s, 90° flip. MOLLI: 2 inversion sets of 3 and 5 images, 75% partial Fourier, TImin=110ms with 80ms increment, 35° flip, TE/TR=1.3/2.9ms, simulated HR=60bpm. SASHA: single-shot SSFP images from 10 consecutive heartbeats with incremented TI spanning the RR interval in the last 9 images (no saturation in the first image), 70° flip, TE/TR=1.3/2.6ms, full k-space, simulated HR=60bpm. T₂: spin-echo (SE) with 7 TEs 11-200ms. Simulations: Bloch equation simulations of MOLLI and SASHA were performed in MATLAB using actual acquisition and physiology parameters and SE measured T₁ and T₂ values.

In-Vivo

For 10 healthy volunteers (5 male, 28.8±6.6yrs), blood and myocardial T₁s were measured using MOLLI and SASHA (parameters as above) in a mid-ventricular short-axis slice at baseline and 20 minutes following 0.1mmol/kg Magnevist.

Results

In blood-like phantoms with long T₂ (179-196ms), SASHA and MOLLI T₁s agree well with IR-SE (0.7±0.5% and 2.2±1.8% absolute difference respectively), while shorter T₂ (46-76ms) tissue-like phantoms have greater underestimation with MOLLI (8.4±3.5%) than SASHA (0.9±0.6%) (Fig. 1). MOLLI simulations predict underestimated T₁s, with 1.3±0.9% absolute difference from observed values (vertical lines, Fig. 1). SASHA simulations also agree well with observations (0.8±0.5%, not shown). In healthy volunteers (63.3±8.4bpm), MOLLI T₁s also show greater underestimation compared to SASHA in tissue than blood, although the difference is larger than observed in phantoms or predicted by simulations in all cases (Table 1).

Figure 1

Error in MOLLI and SASHA T₁ values compared to gold standard inversion-recovery spin-echo (IR-SE) in "tissue-like" phantoms (T₁s 339-1145ms) and "blood-like" phantoms (T₁s 275-1452ms). A Bloch equation simulation of MOLLI using actual acquisition and physiology timing parameters is also shown, with the difference between simulated and actual results indicated with a vertical line for each phantom.

Table 1 Comparison of MOLLI and SASHA T₁ values in 10 healthy volunteers prior to and 20 minutes following 0.1 mmol/kg Magnevist. All comparisons between MOLLI and SASHA are significant (p<0.01, two-tailed, paired Student’s t-test).

Conclusions

MOLLI significantly underestimates T₁s in shorter T₂ tissue-like phantoms but less so in longer T₂ blood-like phantoms, as predicted by simulations. Similar trends were observed in-vivo with MOLLI, although with greater T₁ underestimation (compared to SASHA) than predicted. SASHA had good agreement with IR-SE T₁ phantom measurements and simulations and can be acquired in less time than MOLLI.

Funding

N/A