Introduction

Magnetic resonance imaging (MRI) and spectroscopy (MRS) have gained increasing favor in recent years as non-invasive techniques for measuring and monitoring hepatic fat in non-alcoholic fatty liver disease (NAFLD^1,2). One such technique, high-speed T2-corrected multi-echo ¹H spectroscopy (HISTO³), employs a specialized stimulated echo acquisition mode spectroscopy localization sequence to acquire data at multiple echo times within a single breath hold. When combined with a moderate repetition time (TR), HISTO allows measurement of the proton density fat fraction (PDFF) with minimal relaxation weighting. Because current MR liver fat techniques require a breath hold – HISTO included – these techniques have limited application in patients where breath holding might be particularly challenging, such as with the elderly and children. While the effects of respiratory motion in abdominal imaging are well understood, its effects in abdominal spectroscopy have been little studied. The purpose of this study was to evaluate the efficacy of HISTO during free breathing.

Methods

All studies were approved by the IRB and performed on a 1.5 T MRI system (Siemens Healthineers, Erlangen, Germany) using a 16-channel body array coil in conjunction with a spine coil. Breath-hold (BH) and free-breathing (FB) liver spectra were acquired from 18 participants (10/8 female/male, age range: 9-19 yrs, mean: 15.8±2.3 yrs) using a commercial HISTO sequence. Repeatability data was collected on 7 of the participants (3/4 female/male): participants were removed from the magnet, taken off the table, placed back on the table and repositioned for the second set of measurements. Voxels were placed away from major vessels, the left liver lobe, and the edges of the liver. No attempt was made to position the voxel in the same region for repeat measurements. The HISTO sequence was run with a 3000-ms TR, a 10-ms mixing time (TM), and five echo times: 12, 24, 36, 48 and 72-ms (Figure 1). Signals consisted of 1024 complex points acquired with a 1200 Hz spectral width from a 3x3x3 cm3 voxel. Liver fat data sets were acquired during a 15 s free-breathing period, then repeated for a 15 s breath hold period. Liver PDFFs were obtained from automatically generated HISTO reports. Water line widths were measured using AMARES in jMRUI^4,5. The relation between BH and FB liver PDFF was examined through linear regression. To examine repeatability, we employed linear regression and measured differences in PDFF in the sub group of 7 participants for each FB-BH pair, and for the first and second set of FB data, and first and second set of BH data.

Results

Agreement between BH and FB liver fat measurements is excellent (Figure 2) with a coefficient of determination r² of 0.997, and a regression curve with a slope of 0.997 (p < 0.00001) and an intercept of only 0.095. For repeated measurements, all r² values were greater than 0.990 s (Table 1). Differences between the BH and FB fat measurements for all 25 data sets ranged from -1.32 to 0.96 percentage points. Surprisingly, but not significantly (p=0.306, N=18, Signed Rank Test), line widths were in general wider for BH than FB measurements. Correlation analysis of all repeated measurements yields values greater than 0.995 for all possible combinations, an indication that differences between repeated sets are not significant.

Discussion

With careful voxel placement, a 15 s HISTO sequence can be performed without a breath hold. This makes the sequence very useful in pediatrics, sedated patients, and others with breathing and respiratory challenges. Furthermore, these results indicate that multiple liver regions can be spectroscopically interrogated without additional stress to patients. Although differences between successive FB and BH measurements are on average less than one percentage point, the largest difference occurs when the BH spectral line width is more than 100% greater than that of the FB line width. Visual verification of spectral quality might thus be required for longitudinal measurements when tracking small changes – less than 2-3 percentage points. Current results strongly support free-breathing spectroscopic liver PDFF measurements during multi-echo acquisitions.

Acknowledgements

Thanks to all the MR Technologists at Children's for their help with this study.