The aim of this study was to examine the variation in signal intensity ratio (SIR) values in Eurospin gel phantoms and healthy volunteer (HV) brain images in response to different magnetic resonance imaging hardware and software settings.
Gel phantoms with T1 relaxation times similar to the dentate nucleus (DN), pons (P), globus palladus (GP), and thalamus (Th) were scanned using a T1-weighted 2-dimensional spin-echo sequence on 2 magnetic resonance imaging scanners (3 T and 1.5 T). Imaging was performed by sequentially altering selected magnetic resonance (MR) parameters relative to a default pulse sequence, and the protocol was implemented repeatedly over 3 months. The experiment was also repeated on a cohort of 15 young HVs. Calculations of DN/P and GP/Th SIR values were derived for the images of the gels (GelDN/P and GelGP/Th) and the HVs (HVDN/P and HVGP/Th).
For the default sequence, the mean SIR values of GelDN/P and GelGP/Th varied by ±2.20% and ±0.75%, respectively, when measured over multiple imaging sessions (3 T). Within a single imaging session, these variations were smaller (±0.17% for GelDN/P and ±0.15% for GelGP/Th). At 1.5 T, the equivalent SIR variations for GelDN/P were ±1.41% (multiple sessions) and ±0.41% (single session), and that for GelGP/Th were ±0.47% (multiple sessions) and ±0.33% (single session).
Sequential changes to the MR sequence parameters resulted in gel SIR variations as follows: 14.07% ± 2.43% (with/without normalization filters), −7.80% ± 0.28% (different echo times, TE), and −5.06% ± 0.59% (selective activation of RF coil elements). The largest variations were noted when the gels were positioned below the scanner isocenter, where SIR measurements were different by 22%.
For the HVs, the SIR values were found to be consistently within 0.64% (single session) for the default sequence. Sequential changes to the MR sequence parameters resulted in SIR variations of −24.47% ± 2.47% (with/without normalization filters), −15.32% ± 7.71% (different echo times, TE), and −2.90% ± 0.78% (selective activation of RF coil elements).
This study has demonstrated that SIR percentage changes from baseline of a similar magnitude to brain gadolinium contrast agent signal hyperintensities can be replicated in phantom models and HVs by altering common MR acquisition parameters and hardware.
From the *Division of Molecular and Clinical Medicine, School of Medicine, University of Dundee; Departments
†Medical Physics, and
‡Clinical Radiology, NHS Tayside, Ninewells Hospital and Medical School, Dundee, United Kingdom.
Received for publication August 8, 2018; and accepted for publication, after revision, September 12, 2018.
Conflicts of interest and sources of funding: Financial support for the study was provided via the Renal Dialysis for Venous Access (ReDVA) project (a European Union's Seventh Framework Program grant agreement no. 324487). The authors would like to state that we have received funding from Guerbet. L.Y. is currently receiving PhD sponsorship, while S.M. and G.H. have previously received research and educational support grants. Guerbet, however, were not involved in the study or publication processes.
Correspondence to: J. Graeme Houston, MD, FRCR, Division of Molecular and Clinical Medicine, School of Medicine, University of Dundee, Ninewells Hospital and Medical School, James Arrnott Drive Dundee DD1 9SY, United Kingdom. E-mail: firstname.lastname@example.org.