Multi-site, multi-vendor comparison of T1 measurement using ISMRM/NIST system phantom
Kathryn E Keenan1, Karl F Stupic1, Michael A Boss1, Stephen E Russek1, Tom L Chenevert2, Pottumarthi V Prasad3, Wilburn E Reddick4, Kim M Cecil5, Jie Zheng6, Peng Hu7, Edward F Jackson8, and Ad Hoc Committee for Standards in Quantitative MR9

1Physical Measurement Laboratory, National Institute of Standards and Technology, Boulder, CO, United States, 2University of Michigan, Ann Arbor, MI, United States, 3NorthShore University Health System, Evanston, IL, United States, 4St. Jude Children's Research Hospital, Memphis, TN, United States, 5Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States, 6Washington University in St. Louis, St. Louis, MO, United States, 7University of California Los Angeles, Los Angeles, CA, United States, 8University of Wisconsin, Madison, WI, United States, 9ISMRM, Berkeley, CA, United States

Synopsis

We used the ISMRM/NIST system phantom to assess variations of T1 measurements across MRI systems at 1.5 T and 3 T, to determine the repeatability and reproducibility of the T1 measurements. This study demonstrates that T1 variations from NMR-measured value are correlated site-to-site within a vendor and by position within the head coil. The deviation from the NMR-measured values is greater at 3 T than at 1.5 T. The VFA data has a larger variation than IR; B1 inhomogeneity could contribute to the larger systematic error in VFA measurements. The ISMRM/NIST system phantom is an excellent tool for evaluating multi-site MRI acquisition protocols.

Purpose

Accurate T1 relaxation time measurement is critical for many quantitative MRI applications. An MRI system phantom (Figure 1 A) was developed through collaboration between the ISMRM Ad Hoc Committee on Standards for Quantitative MR and the National Institute of Standards and Technology (NIST).1 The system phantom has SI-traceable components and is being monitored by NIST for stability and accuracy. We used the system phantom to assess variations of T1 measurements across MRI systems at 1.5 T and 3 T. The purpose of this study was to determine the accuracy, precision, repeatability and reproducibility of the T1 measurements.

Methods

Two system phantoms were imaged at multiple sites for three vendors across two field strengths using head coils with 8 to 32 channels (Table 1 in Figure 1 C). For each vendor, one location performed repeatability analysis (n=3 measurements) at 3 T; repeatability was assessed with the coefficient of variation.2 The study included T1 measurement by inversion recovery (IR) using 2D FSE-IR and variable flip angle using 3D FSPGR. The T1 data was fit using custom software (Phantom Viewer, developed at NIST). Details regarding the phantom and data acquisition protocol are available online: https://collaborate.nist.gov/mriphantoms/bin/view/MriPhantoms/RecommendedImagingProtocols. NiCl2 was used to dope deionized water and create a range of T1 values from approximately 20 to 2000 ms. Across all measurements, reported temperature ranged from 17.1 °C to 23.3 °C; temperature was reported either of the MRI room or of the bulk water in the phantom.

Results

The deviations from the reference NMR T1 measurements at 20 °C for each of the vendors and both methods are shown for 1.5 T in Figure 2 and 3 T in Figure 3. The repeatability study coefficients of variation are presented in Figure 4. The CV for Vendor B, VFA is three times the CV of the other vendors and methods; Vendor B, VFA also has the greatest deviation from the NMR-measured values (Figure 3 B). T1 VFA deviation from reference values for representative data sets vary by position within the head coil (Figure 5).

Discussion

Systematic differences in T1 relaxation time measurement can be identified as a function of measurement method (IR or VFA), sample position and scanner type. As measured in our lab, the temperature variation of the NiCl2 T1 relaxation time is non-linear and increases by 7.2% over the range of bore temperatures (17.1 °C to 23.3 °C). Thus, the deviation from reference T1 cannot be attributed only to temperature variation. The deviation from NMR-measured value is greater at 3 T than at 1.5 T. The VFA data has a larger variation than IR; B1 inhomogeneity could contribute to the larger systematic error in VFA measurements. Unlike 3 T, the 1.5 T data do not exhibit an obvious pattern of deviation from the NMR-measured values, with the exception of Vendor B, which consistently underestimates T1 with VFA. However, additional data is needed for vendors A and B. This study demonstrates that T1 variations from NMR-measured value are correlated site-to-site within a vendor and by position within the head coil. The ISMRM/NIST system phantom is an excellent tool for evaluating multi-site MRI acquisition protocols.

Acknowledgements

We appreciate the efforts of all those who completed scans of the phantom, without them this study would not be possible. The SQMR committee members are listed here: http://www.ismrm.org/members-only/committee-directory/#SQMR.

References

1. Russek SE, Boss MA, Jackson EF, et al. Characterization of NIST/ISMRM MRI System Phantom. Proceedings of the 20th meeting of the ISMRM, Melbourne, Australia, 2012. 2. Sullivan DC, Obuchowski NA, Kessler LG, et al. Metrology Standards for Quantitative Imaging Biomarkers. Radiology. 2015.

Figures

ISMRM/NIST system phantom diagram (A); MRI image of T1 array (B); and Number of sites, by vendor and field included in this study (C).

Deviation of MRI measured values from NMR measured values at 1.5 T for both IR and VFA methods. (A) IR deviation is larger at short T1 times, due in part to model difficulty in fitting the data. (B) VFA deviation does not have an obvious pattern; more data is needed.

Deviation of MRI measured values from NMR measured values at 3 T for both inversion recovery and variable flip angle methods. (A) IR deviation has outliers at low T1 times, which can be difficult for the model to fit. (B) VFA deviation patterns are repeated for measurements within a vendor.

Repeatability assessment using the coefficient of variation at 3 T for both IR and VFA. For each vendor, repeatability was assessed at one location with three measurements.

Positional dependency of the deviation from NMR value for VFA measurements at 3 T. One representative data set is used from each vendor. A face is drawn on the plot to demonstrate position of the phantom within the head coil.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3290