Patrick M. Lehmann1, Mads Andersen2, Anina Seidemo1, Xiang Xu3,4, Xu Li4,5, Nirbhay Yadav4,5, Ronnie Wirestam1, Frederik Testud6, Patrick A. Liebig7, Pia C. Sundgren8,9, Peter C. M. van Zijl4,5, and Linda Knutsson1,4
1Department of Medical Radiation Physics, Lund University, Lund, Sweden, 2Philips Healthcare, Copenhagen, Denmark, 3BioMedical Engineering and Imaging Institute, Icahn School of Medicine at Mount Sinai, New York, NY, United States, 4Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins University School of Medicine, Baltimore, MD, United States, 5F.M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, MD, United States, 6Siemens Healthcare AB, Malmö, Sweden, 7Siemens Healthcare GmbH, Erlangen, Germany, 8Department of Radiology, Lund University, Lund, Sweden, 9Lund University Bioimaging Centre, Lund University, Lund, Sweden
Synopsis
D-glucose was recently suggested as a CEST-based biodegradable alternative
to gadolinium-based contrast agents. Dynamic glucose-enhanced (DGE) MRI can retrieve
information about glucose uptake, determined by tissue perfusion, transport,
and metabolism. Motion artefacts in DGE-MRI can be mistaken for CEST effects,
while motion correction may erroneously alter true DGE signal. A digital human
head phantom based on a realistic glucose infusion protocol was developed to
analyse motion artefacts and validate rigid-head retrospective motion
correction. This phantom can be used for testing different correction
approaches using various motion patterns and contrast responses to better understand
these effects in vivo.
Introduction
Dynamic glucose-enhanced (DGE) MRI uses the phenomenon of chemical exchange saturation transfer (CEST) or exchange-enhanced relaxation to study D-glucose uptake, which is determined by tissue perfusion, transport and metabolism 1-5. However, at clinical field strengths, the CEST effect is small (~1% of the water signal), and analysing DGE data in human brain is associated with challenges such as artefacts due to motion 6. These artefacts can be mistaken for CEST effects, while motion correction may erroneously alter true DGE signal. In this study, a digital human head phantom was developed to simulate the DGE contrast at 3 T. This phantom can be used for motion artefact analysis and optimization of different motion correction approaches.Methods
An MPRAGE brain image from a tumour patient (glioblastoma WHO IV) was used to segment grey matter (GM), white matter (WM), other soft tissue, cerebrospinal fluid (CSF), and bone. For further differentiation, regions of interest (ROIs) were drawn to select veins (V), arteries (A), lateral ventricles (LV), and tumour tissue (T). GlucoCEST effect sizes (normalized intensities I) for V, A, GM, and WM at 3 T were derived using the Bloch-McConnell equations for a B1 = 2.0 µT (IV= 0.49 %, IA = 1.05 %, IGM = 0.19 %, IWM = 0.06 %), based on concentration changes of 10 mM in blood and ¼ of that in tissue 7. CSF was characterized by ICSF = 0.63 %, 60 % of IA 8, and for high-grade glioma IT = 0.73 % was assumed based on increased vascular contribution 9 and leakage into extravascular extracellular space. GlucoCEST curves were simulated based on DGE response curves observed at 3 T using a prototype sequence (MAGNETOM, Prisma, Siemens Healthcare, Erlangen, Germany) and 7 T (Achieva, Philips, Best, The Netherlands) 3, 10. GlucoCEST contrast curves and tissue segmentations were combined creating a glucoCEST time series, representing the ground truth. A dynamic time series with an acquisition time of 690 s and a temporal resolution of 5 s was simulated, including (non-saturated) data acquisition (20 s), baseline (20-140 s), D-glucose infusion (140-380 s), and signal rise and decay (380-690 s). One representative rigid-body motion pattern was applied to this time series (Figs.1, A, B). LVs can show a relative volume change of 1 % due to cardiac pulsation 11. D-glucose loading can lead to ventricular dilatation by about 2.4 % 12. Such non-rigid LV movement was therefore simulated by adding an affine transformation with a time-dependent scaling coefficient to the LV segmentation (Fig. 1, C). The glucoCEST time series were downsampled to (2×2×3) mm3 and Rician noise was added. Three simulated glucoCEST experiments were investigated: without motion and with infusion (ground truth), motion without infusion, and with motion and infusion illustrated in Fig. 2. Rigid-head-motion correction was performed using Elastix 13. DGE signal given by
$$S_{DGE} (t)=\frac{S_{base}-S(t)}{S_{0}},$$
was
calculated using the mean pre-infusion
(baseline) signal $$$S_{base}$$$, the signal time course $$$S(t)$$$, and the non-saturated water signal $$$S_{0}$$$. ROIs were drawn for WM and tumour for extracting contrast response
curves. Averaged area under curve (AUC) maps were created according to
$$AUC_{mean}=\frac{\sum_{0}^{t} S_{DGE}(t)}{N},$$
for pre-infusion (115 s), during infusion (245 s) and two post-infusion
time intervals (45 s before, 160 s after peak) showing signal rise and decay. N indicates the number of averaged image time frames. Results and Discussion
The simulated dynamic signal ratio $$$S_{sat}/S_{0}$$$ time
series -and $$$S_{DGE}(t)$$$ are shown in Figs. 3A and 3B, respectively. The ground
truth AUC maps (Fig. 4E) show a clear dynamic tissue response. Added motion
(Figs. 4, A, B) introduced strong artefacts (pseudo-CEST contrast) in terms of
hyperintensities at tissue interfaces and hypointensities in LVs. Even without
D-glucose infusion, this caused apparent effects that have similar or higher effect
size in the AUC maps as the ground truth (Fig. 4, A) and e. g. ~ - 0.8 % in tumour
tissue (Fig. 5, A). Similar findings have been reported in vivo 6, 14.
GlucoCEST contrast is more reliable after
retrospective motion correction (Figs. 4, C, D; Figs. 5, B, D), removing the
most severe artefacts, but not fully recovering original image series quality. Especially
LVs and tissue interfaces show residual AUC effects in form of hyper- and
hypointensities (Figs. 4, C, D). Pseudo- and true glucoCEST contrast could not
be fully disentangled, which was attributed to smoothing and interpolation
errors related to the motion correction algorithm. The signal without infusion but
motion showed a strong and broad negative signal, where the signal with
infusion is reduced (Figs.5, B, D). The signal drop corresponds to a sudden
movement (255 s, Figs. 1, A, B). The measured pseudo-CEST effect depends
strongly on the motion pattern.
Conclusion
A digital human head
phantom simulating a glucoCEST experiment in combination with rigid-body
movement and dynamic LV dilatation/contraction was developed. Motion induced
pseudo-CEST effect was partially alleviated by rigid motion correction.
Residual motion effects in tissue boundaries were tangled with true CEST
contrast complicating image interpretation. This can lead to data misinterpretation.
Artefacts similar to in vivo observations
could be reproduced. The phantom presented here will be useful in developing
proper analysis methods that can separate true signal effects from motion
artefacts.Acknowledgements
No acknowledgement found.References
- Chan KW, McMahon MT, Kato
Y, et al. Natural D-glucose as a biodegradable MRI contrast agent for detecting
cancer. Magn Reson
Med 2012;68(6):1764–1773.
-
Walker-Samuel S, Ramasawmy R, Torrealdea F, et al. In vivo imaging of glucose uptake and
metabolism in tumors.
Nat Med. 2013;19:1067-1072.
- Xu X, Chan KWY, Knutsson L, et
al. Dynamic
glucose enhanced (DGE) MRI for combined imaging of blood–brain barrier break
down and increased blood volume in brain cancer. Magnetic Resonance in Medicine 2015; 74(6): 1556-1563.
- Paech D, Schuenke P, Koehler C, et al. T1ρ-weighted Dynamic Glucose-enhanced MR Imaging in the
Human Brain. Radiology
2017;285(3): 914–922.
- Schuenke P, Paech D, Koehler
C, et al. Fast and
quantitative T1ρ-weighted
dynamic glucose enhanced MRI. Sci Rep 2017;7:42093
- Zaiss M, Herz K, Deshmane A, et al. Possible
artifacts in dynamic CEST MRI due to motion and field alterations. J Magn Reson
2019;298:16-22.
- Xu X, Sehgal AA, Yadav NN, et al. d‐glucose weighted
chemical exchange saturation transfer (glucoCEST)‐based dynamic glucose
enhanced (DGE) MRI at 3T: early experience in healthy volunteers and brain
tumor patients. MRM 2020;84:247-262.
- Karen R, Principles of Neurologic Infectious Diseases. McGraw Hill Higher
Education. 2005;p.4:978-0-07-140816-5.
- La Violette PS, Cohen AD, Prah MA, et al. Vascular change measured with independent component analysis of
dynamic susceptibility contrast MRI predicts bevacizumab response in high-grade
glioma.Neuro-Oncology
2013;15:442-450.
- Knutsson L, Seidemo A, Rydhög
Scherman A, et al. Arterial Input Functions and Tissue Response Curves in
Dynamic Glucose-Enhanced (DGE) Imaging: Comparison Between glucoCEST and Blood
Glucose Sampling in Humans. Tomography 2018;2379-1381.
- Zhu D, Xenos M, Linninger AA, et al. Dynamics of lateral
ventricle and cerebrospinal fluid in normal and hydrocephalic brains. JMRI
2006;24:756-770.
- Puri B, Lewis H, Saeed N, et al. Volumetric change of the
lateral ventricles in the human brain following glucose loading. Exp Phy
1999;84(1):223-226.
- Klein S, Staring M, Murphy K, et al. elastix:
a toolbox for intensity‐based medical image registration. IEEE Trans Med
Imaging. 2010;29:196–205.
- Herz K, Lindig T, Deshmane A, et al. T1ρ‐based dynamic
glucose‐enhanced (DGEρ) MRI at 3 T: method development and early clinical experience in the
human brain. Magn Reson Med. 2019;00:1-16.