Synopsis

Keywords: Muscle, Muscle

We explored the feasibility of T₁ and T₂ mapping of normal tissue (masseter muscle) in the head and neck (HN) region using magnetic resonance fingerprinting (MRF) and evaluated its repeatability. Our study included fingerprinting data from four healthy volunteers and four repeats each, along with acquiring B₀ maps. The T₁ and T₂ estimation of masseter muscle by MRF is similar to previous literature values. We report wCV<4.3% for T₁ measurements and wCV<7% for T₂ measurements for all subjects. No significant B₀ variations were observed in the masseter muscle regions.

Introduction

MRI is commonly used for imaging tissue in the head and neck (HN) region. While T₁ and T₂ metrics could be useful biomarkers for HN tumor morphology, only a limited number of studies investigates the relaxometric mapping of HN region ¹. Clinical relaxometry mapping is restricted by long acquisition times.
In this study, we explored the feasibility of T₁ and T₂ mapping of normal tissue masseter muscle in the HN region using magnetic resonance fingerprinting (MRF) and evaluated its reliability and robustness by conducting a repeatability study on four healthy volunteers with four repeats each. We are the first to demonstrate normal tissue masseter muscle T₁ and T₂ mapping using MRF to the best of our knowledge.

Methods

Volunteer selection
Four healthy volunteers participated in this study. Written informed consent was obtained from all the volunteers. The median age of the four volunteers was 32 years (range, 24-34 years; all males). None of the volunteers have a patient history related to masseter muscles. The subject-specific information is presented in Table 1.
MRF implementation
The TR/FA scheme of the MRF sequence used parameters reported in previous literature ^{2, 3}. The sequence parameters were: Field of View (FOV) of 25 cm × 25 cm, matrix size=225×225, sampling bandwidth=±250 kHz, TE=2.3 ms, slice thickness=5 mm, number of slices=25. The acquisition time of 25 slices is 7:26 (min:sec). The raw data were reconstructed using singular value decomposition (SVD) based dictionary matching in MATLAB (MathWorks, Nattick, MA). The reconstruction time of a 25 slices dataset is ~3 minutes.
HN MRI protocol
All of the volunteers' scans were performed on a 3T, 70 cm bore scanner (SIGNA Premier, General Electric Healthcare (GEHC,Waukesha, WI, USA), using an HN coil for imaging. The HN imaging protocol consists of a vendor-supplied T₁ spin echo (SE) sequence, a T₂ fat-suppressed SE sequence, a MRF sequence, and a B₀ mapping sequence.
The T₁ and T₂ SE sequence were only scanned once in the first repeat for subject 1. The B₀ mapping sequence was added at the end of the protocol for all subjects except subject 1. We repeated the MRF sequence four times for each subject in a coffee break scan setting. The total scan time of one repeat was ~15 minutes.
A vendor sequence (fast GRE based B₀ mapping) was run to acquire the B₀ maps. The parameters of the B₀ mapping sequence were: TR=250 ms, FOV=250 mm, slice thickness=5 mm, and FA=30 degree. The total acquisition time for the sequence was 2:12 (min:sec).
Region of interest (ROI) and statistical analysis
All ROIs were selected using ITK-SNAP (www.itksnap.org) ⁴. Two regions (left and right masseter muscle) were selected for each slice where masseter muscles were presented. The size of ROIs were adjusted based on the size of the masseter muscles. We used within-case coefficients of variation (wCV) to assess the repeatability of T₁ and T₂ values estimated using MRF. For repeatability, wCV is defined as the ratio of the standard deviation (SD) to the mean T₁ and T₂ values over four repeats.

Results/Discussion

Figure 1 presents the T₁w and T₂w images acquired using vendor supplied T₁ SE and T₂ fat-suppressed SE sequence and T₁ and T₂ maps acquired using MRF for all subjects. One representative slice was used. The orange arrows indicate the location of the masseter muscle. ROIs were selected for all slices where masseter muscles were presented.
Table 2 summarizes the HN volunteers’ T₁ and T₂ value for normal-appearing masseter muscle for all volunteers. The mean T₁ and T₂ values were 982.0 ms and 75.1 ms for left masseter muscle and 926.2 ms and 69.2 ms for right masseter muscle respectively. These values are comparable to previous literature values ¹. The T₁ and T₂ values of the left masseter muscle are slightly larger than the right masseter muscle. This may be caused by B₀ inhomogeneity, shown in Figure 3.
Figure 2 shows the T₁ and T₂ values of MRF for four datasets. MRF estimates have less than 7% variations for all subjects. The wCV of T₂ estimations is larger than the T₁ estimations. This may be because the MRF dictionary used in this study has resolution of 10 ms in the small T₂ range. A dictionary with finer resolution will likely improve the wCV of T₂ estimations.
Figure 3 presents the B₀ maps for subject 2, 3, and 4. We did not acquire B₀ map for subject 1. The left side of the masseter muscle experiences larger B₀ variations compared to the right side of the masseter muscle. This may contribute to the difference in mean T₁ and T₂ between left and right masseter muscle shown in Table 2. B₀ inhomogeneity correction methods may be applied in future work to improve the T₁ and T₂ values of masseter muscles.

Conclusion

In this study, we investigated feasibility of relaxometric mapping of normal tissue masseter muscle using MRF and evaluated its repeatability. We conducted repeatability study on four healthy volunteers, with four repeats for each volunteer. B₀ maps were acquired at the end of the protocol. We report T₁ and T₂ estimates by MRF similar to previous studies, and wCV<7% for all volunteers.

Acknowledgements

This work was supported, in part, by GE-Columbia research partnership grant and also performed at Zuckerman Mind Brain Behavior Institute MRI Platform, a shared resource, and Columbia MR Research Center site.