The Impact of Altering MRI Equipment and Scanning Parameters on Phantom Signal Measurements Possible implications for interpreting Gadolinium signal changes within the brain Laura Kate Young1, Shona Matthew1, Stephen Gandy1,2, Lukasz Priba2, John Graeme Houston1,2 1Division
of Molecular & Clinical Medicine, University of Dundee, Ninewells Hospital & Medical School, Dundee, UK 2Medical Physics, NHS Tayside, Ninewells Hospital & Medical School, Dundee, UK, 3Clinical Radiology, NHS Tayside, Ninewells Hospital & Medical School, Dundee, UK
“
Retrospective studies have correlated increases in brain signal intensity ratios (SIR) on MR images with gadolinium deposition from some MR contrast agents. This phantom work has demonstrated that simple alterations to common MR parameters can result in SIR fluctuations of similar magnitude to those due to gadolinium deposition. In this context, historical clinical brain images should therefore be interpreted carefully and pulse sequence parameters should be standardised wherever possible.
Gadolinium-based Contrast Agents (GdCAs) used in MR imaging with the detection of brain hyper-signal intensities1 (HSI), which indicate gadolinium retention2.
• Signal intensity ratios (SIR) measured on retrospective images –
commonly the dentate nucleus-to-pons (DN/P) and globus pallidus-tothalamus (GP/Th). A HSI is typically defined as a change in SIR greater than 4%3. Whilst the impact different acquisition parameters, sequences and equipment used may have on SIR have been discussed4, it has never been thoroughly investigated.
• •
• AIM: to obtain and compare phantom signal intensity ratio (SIR)
Mean % Change in SIR
measurements acquired on MR images in which scanning parameters and equipment were systematically altered.
UK) was centrally loaded with 4 gel-filled tubes of closely matching intrinsic T1 to those of the DN, P, GP and Th and were tested for suitability. Imaging was performed on a 3.0T PrismaFIT machine (Siemens, Germany). Scanning: A T1-weighted 2D spin echo sequence was used as “baseline” (TR, 700ms; TE, 12ms; FA, 90°; slice thickness, 5mm; field-of-view (FoV) 250mm; bandwidth, 130Hz/pixel; no. of slices, 1; no. of averages, 1; 18-channel head/neck phased array coil; no image filters). Baseline parameters were systematically varied one at a time. Alterations were included TR, TE, image filters and no. of RF coil channels. The protocol was repeated on 10 different occasions over the course of two months. The same protocol was also tested on a 1.5T Signa HDxt machine (GE Healthcare, USA). Image analysis: The percentage change in SIR from baseline was measured (as an average over the ten repeats) and compared with the threshold for hyper-signal.
•
• • •
Results
GelDN/P
4.00
• Phantom: A Eurospin gel ‘TO5’ phantom (Diagnostic Sonar Ltd, Livingston,
• PHANTOM: 3 out of the 4 gels were chosen for SIR replication. T1 at 3.0T of
0.00
GelDN = 977ms (in vivo; 1018.5ms5), GelGP = 804ms (951ms6) and GelP and GelTh = 1053ms (1054.1ms and 1074.8ms, respectively5).
-4.00
Filter A + SCIC - ON
PURE - ON
SCIC - ON
Slice Uniformity - OFF
Filter A + PURE - ON
Filter A + SCIC - ON
PURE - ON
SCIC - ON
Slice Uniformity - OFF
References 1. Kanda, T. et al. High Signal Intensity in Dentate Nucleus and Globus Pallidus on Unenhanced T1-weighted MR Images: Relationship with Increasing Cumulative Dose of a Gadolinium-based Contrast Material. Radiology 270, 834–841 (2014). 2. McDonald, R. J. et al. Intracranial Gadolinium Deposition after Contrast-enhanced MR Imaging. Radiology 275, 772–782 (2015). 3. Ramalho, J. et al. T1 signal-intensity increase in the dentate nucleus after multiple exposures to Gadodiamide: Intraindividual comparison between 2 commonly used sequences. Am. J. Neuroradiol. 37, 1427–1431 (2016). 4. Adin, M. E. et al. Hyperintense Dentate Nuclei on T1-Weighted MRI: Relation to Repeat Gadolinium Administration. Am. J. Neuroradiol. 36, 1859–65 (2015). 5. Badve, C. et al. Simultaneous T1 and T2 Brain Relaxometry in Asymptomatic Volunteers using Magnetic Resonance Fingerprinting. Tomography 1, 136–144 (2015). 6. Mädler, B. et al. 3D-Relaxometry - Quantitative T1 and T2 Brain Mapping at 3T. Proc. Intl. Soc. Mag. Reson. Med. (2006).
Filter A + PURE - ON
Each RF coil channel detects local regions of signal which are then combined to form the final image. Inconsistent use of RF coil elements across longitudinal scan series may result in apparent SIR differences being detected due to SNR variations.
Filter A - ON
•
The TR and TE times help to define when the MR signal is sampled. Inconsistent use of these parameters across 1 longitudinal scan series may result inPage apparent SIR differences being detected due to the different tissue T1 and T2 values.
Filter A - ON
•
Normalisation filters are commonly used to improve Page 1 the uniformity of the signal across the FOV, and the B1 filter adjusts for field inhomogeneities caused by the scanned object. Inconsistent implementation of these filters across longitudinal scan series may result in apparent SIR differences being detected.
4.00
Ex. Dyn. Range - ON
•
GelGP/Th 12.00
Ex. Dyn. Range - ON
Discussion
TE = 36
array uniformity enhancement (PURE) and surface coil intensity correction (SCIC) filters and changes in TE and TR.
TE = 36
Head 1-2 - ONLY
Head 3-4 - ONLY
Head 1-4 - ONLY
TE = 36
TE = 24
TR = 400
TR = 1000
Image filter med - ON
B1 filter med - ON
Norm + Dist. corr. - ON
Dist corr.- ON
Normalise - ON
Prescan Norm. - ON
Fig 1. Mean % change in SIR for GelDN/P and GelGP/Th scanned on the Siemens scanner. Dotted lines = ±4% HSI threshold.
• GE: Av. baselines: GelDN/P = 0.920 ± 0.013 and GelGP/Th = 1.069 ± 0.005. • Both GelDN/P and GelGP/Th were mainly affected by application of phased-
TE = 24
-8.00
-12.00 TE = 24
•
-4.00
-4.00
TR = 400
•
0.009. GelDN/P was mainly affected by application of normalise filter, change in number of receiver channels and change in TE. GelGP/Th was mainly affected by B1 filter (medium), change in number of receiver channels and change in TE and TR.
4.00
TR = 400
• SIEMENS: Av. baselines: GelDN/P = 1.044 ± 0.023 and GelGP/Th = 1.208 ±
12.00
TR = 1000
GelGP/Th
GelDN/P
TR = 1000
Head 1-2 - ONLY
Head 3-4 - ONLY
Head 1-4 - ONLY
TE = 36
TE = 24
TR = 400
TR = 1000
Image filter med - ON
B1 filter med - ON
Norm + Dist. corr. - ON
Dist corr.- ON
0.00
Normalise - ON
4.00
Prescan Norm. - ON
Mean % Change in SIR
-8.00
Mean % Change in SIR
• Recent studies have linked the repeat administration of some
Mean % Change in SIR
Introduction
” Method
-4.00
-12.00
Fig 2. Mean % change in SIR for GelDN/P and GelGP/Th scanned on the GE scanner. Dotted lines = ±4% HSI threshold.
Conclusion
• These results show that SIR fluctuations of similar magnitude to HSI
Page 1
can be achieved by altering these common MR acquisition parameters: -> Image filters -> TR and TE -> No. of receiver coil channels
• There also seems to be a dependency related to the use of different scanner hardware and software – which may have implications for comparing longitudinal scan series across a number of years.
Page 1
It is recommended that for future brain MR studies involving GdCA’s, MR sequence parameters are standardised, where possible, and the effects are carefully considered when drawing conclusions about the significance of hyper signal intensities. Conflicts of Interest This project has received funding from the European Union’s Seventh Framework Programme for research; technological development and demonstration under grant agreement no 324487 and from Guerbet and SINAPSE. Contact Information Laura K Young
[email protected]
