EMIM 2018 ControlCenter

Online Program Overview Session: PW-25

To search for a specific ID please enter the hash sign followed by the ID number (e.g. #123).

MRI and Hyperpolarization Technology

Session chair: Arnaud Comment - Cambridge, UK; Arend Heerschap - Nijmegen, The Netherlands
Shortcut: PW-25
Date: Friday, 23 March, 2018, 11:30 AM
Room: Banquet Hall | level -1
Session type: Poster Session


Click on an contribution preview the abstract content.

# 216

Using Pulsed Excitation Waveforms for Quantitative Relaxation Mapping in Magnetic Particle Imaging (#459)

D. Hensley1, 2, Z. W. Tay1, N. Oude Booijink1, P. W. Goodwill2, B. Zheng1, S. M. Conolly1, 3

1 University of California, Berkeley, Bioengineering, Berkeley, California, United States of America
2 Magnetic Insight, Inc., Alameda, California, United States of America
3 University of California, Berkeley, Electrical Engineering and Computer Sciences, Berkeley, California, United States of America


Magnetic particle imaging (MPI) is an emerging radiation-free tracer imaging modality [1, 2]. In MPI, a sensitive magnetic field region is rastered across the imaging field of view to query for the presence of magnetic nanoparticle tracers. Current continuous wave, sinusoidal excitation methods ignore or are hindered by the dynamic physical process by which the tracer rotates in response to the sensitive region trajectory.  Here, we present the first application of pulsed excitation waveforms which allow us to directly quantify relaxation dynamics and construct relaxation maps.


We used our previously reported arbitrary waveform relaxometer (AWR) [3] to implement square wave and steady-state recovery sequences as depicted in Fig. 1. We also implemented new reconstruction algorithms to grid the raw data to a 1D magnetic field domain as shown in Fig. 1 (c—e) and Fig. 2. 40 uL samples of various MPI tracers (10 -- 25 mg/mL) were used to test the new pulsed MPI (pMPI) methods. In square wave excitation, the time domain signal associated with each square wave half period corresponds to a relaxation impulse response which can be directly quantified and gridded to the output domain. In the case of a steady-state acquisition, the inter-pulse period required to induce steady-state can be measured directly from raw time-domain data.


Fig. 2 shows 1D experimental results using pMPI waveforms. Fig. 2 (a) shows high resolution relaxation maps of monodisperse tracers with different core sizes.  As expected, larger tracers are associated with longer time constants (38 us peak for 32 nm tracer, 4 us peak for 21 nm tracer). Fig. 2 (b—d) show plots of measured peak time constants using our steady-state recovery pulse sequence. In Fig. 2 (b), peak time constant as a function of viscosity across a physiologically relevant range is shown. As the theory predicts, we observe a linear relationship between viscosity and time constant. In Fig. 2 (c), peak time constant changes as a function of pH and temperature are demonstrated and in Fig. 2 (d), changes in peak time constant before and after addition of biotinylated albumin to streptavidin coated tracer are shown. Larger tracers that physically rotate show a large change in time constant while smaller tracers that do not rotate show no statistically significant change.


In this work, we show that the magnetic relaxation physics of MPI tracers can sensitively encode physiologic variables of interest and that we can use pulsed excitation waveforms to encode and quantify these processes in ways not possible with canonical continuous wave excitation. In the future, we believe pMPI techniques can open new frontiers in MPI molecular imaging by unlocking flexible and well-posed methods of imaging parameters such as viscosity, pH, binding state, labeled cell metabolic state, and kinetics.


[1]       B. Gleich and J. Weizenecker. Tomographic imaging using the nonlinear response of magnetic particles. Nature, 435(7046):1214-1217, 2005. doi: 10.1038/nature03808.

[2]       P.W. Goodwill and S. M. Conolly. The X-space formulation of the magnetic particle imaging process: 1-D signal, resolution, bandwidth, SNR, SAR, and magnetostimulation. IEEE transactions on medical imaging, 29.11: 1851–1859, 2010.

[3]       Z. W. Tay, P. W. Goodwill, D. W. Hensley, L. A. Taylor, B. Zheng, and S. M. Conolly. A high-throughput, arbitrary-waveform, MPI spectrometer and relaxometer for comprehensive magnetic particle optimization and characterization. Scientific reports, 6, p.34180, 2016.


We would like to acknowledge funding support from NIH 5R01EB019458-03, NIH 5R24MH106053-03, UC Discovery Grant 29623, W. M. Keck Foundation Grant 009323, and NSF GRFP.

MPI Relaxation Encoding and Pulsed Excitation
(a) Depiction of square and steady-state recovery pMPI sequences. (b) We can use pulsed waveforms to remove detrimental relaxation blurring effects by ensuring steady-state conditions are reached when scanning. (c,d) We can also easily measure these same relaxation effects. Depiction of creating a 1D pMPI relaxation map with square wave excitation. (e) Simulated 1D relaxation image.
Experimental Pulsed MPI Relaxation Mapping Results

Pulsed techniques allow for high sensitivity quantification of MPI relaxation and associated parameters. (a) 1D relaxation images of MPI tracers with different core sizes. (b) Peak relaxation time as a function of viscosity. (c) pMPI detection of relaxation changes with pH and temperature. (d) pMPI detection of binding events, but only in an MPI tracer that can physically rotate.

Keywords: magnetic particle imaging, pulse sequence design, pulsed excitation, relaxation mapping, signal encoding
# 217

In Vivo Sensitive and Radiation-Free Ventilation/Perfusion 3D Lung Imaging with Magnetic Particle Imaging (#328)

X. Y. Zhou1, 2, Z. W. Tay1, 2, P. Chandrasekharan2, B. Zheng2, T. Vu3, P. Nahid4, S. M. Conolly2, 5

1 UC Berkeley - UCSF Graduate Program in Bioengineering, Bioengineering, Berkeley, California, United States of America
2 UC Berkeley, Bioengineering, Berkeley, California, United States of America
3 UCSF, Radiology and Biomedical Imaging, San Francisco, California, United States of America
4 UCSF, Pulmonary and Critical Care Medicine, San Francisco, California, United States of America
5 UC Berkeley, Electrical Engineering and Computer Sciences, Berkeley, California, United States of America


Lung ventilation/perfusion (V/Q) scintigraphy is often used to diagnose pulmonary embolism (PE) [1]. While V/Q scans have been supplanted by CTPA for many patients, V/Q is needed for patients who cannot tolerate iodine due to poor kidney function [2]. Unfortunately, due to the use of radionuclides that must be prepared when a V/Q scan is called, patients can face a 2+ h wait for V/Q [3]. Magnetic particle imaging (MPI) is a sensitive medical imaging technique with zero ionizing radiation and long persistence for the kidney-safe SPIO tracers [4]. We show MPI V/Q time course scans in vivo rats.


For ventilation imaging, female Fisher rats under isoflurane anesthesia were intubated and ventilated with aerosolized SPIOs (Kent Scientific Aeroneb, <4.0 micron droplets of 5 mg/ml Micromod Perimag), described in Fig 1a. For perfusion imaging, 700,000 macroaggregated albumin SPIO conjugates (MAA-SPIOs) as described in [5] and visualized in Fig. 1c were administered via IV tail vein injection. Rats were imaged on the Berkeley custom-built 3D 7 T/m field free point MPI scanner described in [6] for Fig. 1d, or on the UCB custom-built 6.3 T/m field free line projection MPI scanner described in [7] for Fig. 1b.  CT imaging was performed on a RS9-80 Micro CT scanner (GE). 


The SPIO ventilation procedure with aerosolized SPIOs allows MPI visualization of the rat lung ventilation in vivo, comprising the “V” scan, while an IV injection of MAA-SPIO allows MPI visualization of the rat lung vasculature in vivo, comprising the “Q” scan. These scans can be performed on the same rat sequentially analogous to a clinical V/Q scintigraphy scan (data not shown), or time course V or Q scans can be taken to monitor V or Q function over time, as shown in Fig. 1b and Fig. 1d. 


High sensitivity and high contrast 3D V/Q scans in vivo rats demonstrate that MPI may be uniquely suited for ionizing radiation-free and iodine-free lung imaging, such as for diagnosis of PE. Moreover, because the magnetic tracer exhibits no radioactive decay, MPI V or Q imaging can be used to monitor V or Q function over time, such as for tracking drug deposition to the lung [8]. MPI V/Q imaging could offer completely iodine and radiation-free, robust 3D lung perfusion images with speed, convenience, cost and safety that are superior to conventional V/Q.  


[1] Neumann et al., Semin Nucl Med, 1980

[2] Thompson et al., UpToDate, 2016

[3] Freeman and Haramati, Eur J Nucl Med Mol Imaging, 2009

[4] Goodwill et al., Adv Mater, 2012

[5] Zhou et al., Phys Med Biol, 2017

[6] Zheng et al., Theranostics, 2016

[7] Yu et al., ACS Nano, 2017

[8] Dubsky and Fouras, Adv Drug Deliv Rev, 2015


We would like to acknowledge National Institutes of Health and National Science Foundation Graduate Research Fellowship Program funding, as well as Mike Wendland for assistance with CT scans at the Berkeley Preclinical Imaging Center.  

Fig. 1
(a) Ventilation set up. The intubated rat receives nebulized SPIOs. (b) MPI lung ventilation time course. MPI lung ventilation images show delivery to lung airways and clearance over days. (c) Perfusion set up. MAA-SPIOs are injected through the tail vein. (d) MPI lung perfusion time course. MPI lung perfusion images show delivery to lung vasculature and clearance through liver and spleen.
Keywords: pulmonary, ventilation, perfusion, magnetic particle imaging, lung
# 218

Reproducible 13C-hyperpolarization of Succinate to 11% in an MRI using SAMBADENA (#532)

S. Berner1, 2, 3, A. B. Schmidt1, 4, M. Zimmermann1, S. Knecht1, 4, A. Pravdivtzev4, J. Hennig1, D. von Elverfeldt1, J. - B. Hövener4

1 Department of Radiology, Medical Physics, Medical Center—University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Baden-Württemberg, Germany
2 German Cancer Consortium (DKTK), Freiburg, Baden-Württemberg, Germany
3 German Cancer Research Center (DKFZ), Heidelberg, Baden-Württemberg, Germany
4 Department of Radiology and Neuroradiology, Section Biomedical Imaging, MOIN CC University Medical Center Schleswig-Holstein, University of Kiel, Kiel, Schleswig-Holstein, Germany


The sensitivity of magnetic resonance (MR) is sufficient for the imaging of 1H-signals in vivo but too low to map 13C metabolites with high resolution in vivo. Hyperpolarization (HP) boosts the sensitivity by overcoming the low thermal polarization, paving the way towards monitoring of metabolism non-invasively. Here, we present higher and more reproducible SAMBADENA1 -HP of the biomolecule succinate (SUC) than before. Note that we already presented similar results2 suffering from poor reproducibility. A change in the experimental workflow increased the reproducibility and polarization yield.


Aqueous 1-13C, 2,3-2H2-SUC (5mM, 700 µL) was formed by the pairwise addition of parahydrogen (pH2) to fumarate using a Rh-bisphospine catalyst directly in the bore of a preclinical 7T MRI (Biospec, Bruker, Germany)1. After injecting >90% pH2 into the precursor solution (t1= 2 s) and waiting (t2= 3 s), Goldman’s sequence3 was played out. In order to preserve the I1I2-para-spin-order Goldman’s sequence was designed for, heteronuclear MLEV decoupling4 was either played out during t2 or t1+t2. Polarization (P) was quantified with respect to a thermally polarized reference. For systematic analysis the effect of varying durations of MLEV pulses on HP was investigated. Moreover, the impact of non-resonant MLEV pulses was examined.


Low and non-reproducible P was obtained when MLEV was played out during t1+t2 (Fig. 1a). When MLEV was applied during t2 alone, P increased significantly (Fig. 1b): A maximum average of P ≈(11±0.6)% was obtained for D90 = 700µs, corresponding to a signal enhancement of 20.000 at 7T. For shorter and much longer D90, P dropped significantly, likely because of power limitation of the coil and low excitation band width, respectively (Fig. 1b). Likewise, P was much reduced when an off-resonance was added to MLEV (Fig. 2). The loss of P when MLEV is applied during t1 may be attributed to field inhomogeneities caused by the bubbles. This hypothesis is supported by the drop in P that is observed for long D90 (Fig. 1b) and when off-resonances are induced on purpose (Fig. 2). This mechanism may explain as well the previously reported but non-reproducible P ≈ 7%, where the synchronization of injection and CW-decoupling was poor: In few cases, decoupling may have been applied accidently during t2.


This investigation of succinate HP with SAMBADENA led to the identification of critical parameters that caused strong loss in polarization yield. By controlling these parameters, namely the hydrogenation process and decoupling, high and reproducible P ≈ 11 % was achieved routinely. Broader r.f. pulses, i.e. shorter pulses along with a more powerful resonator, may further increase the polarization.


1. Schmidt, A. B, Berner, S. et al. Liquid-state carbon-13 hyperpolarization generated in an MRI system for fast imaging. Nat. Commun. 8, ncomms14535 (2017).

2.Stephan Berner, Andreas Schmidt, Schimpf Waldemar & Jan-Bernd Hövener. Hyperpolarization of a biomolecule to 7% in an MRI using SAMBADENA. Proceedings of EMIM 2017

3. Goldman, M. & Jóhannesson, H. Conversion of a proton pair para order into 13C polarization by rf irradiation, for use in MRI. Comptes Rendus Phys. 6, 575–581 (2005).

4. Levitt, M. H., Freeman, R. & Frenkiel, T. Broadband heteronuclear decoupling. J. Magn. Reson. 1969 47, 328–330 (1982).

Fig. 1: P as a function of MLEV-pulse duration with (a) and without (b) decoupling during injection

a) Low P was obtained if the spin system was decoupled during the entire hydrogenation reaction (t1 and t2). Each data point corresponds to a single experiment.

b) High and reproducible P was observed if decoupling was applied after pH2 injection (t2). A maximum P was observed for D90 = 700 µs (n = 3 experiments for D90 > 600µs, n = 1 for D90 = 500µs and 600µs; mean and std.dev shown).

Fig. 2: 13C-HP of succinate as function of off-resonance of the decoupling pulses

P decreased significantly when an off-resonance was added to the MLEV pulses, e.g. for ± 1ppm from P = 9.5 % to ≈ 2 % (D90 = 1000 µs). Note that the spin order transfer sequence was always on resonant, centered on the chemical shift of 1H and 13C. Each datapoint shows the mean value and standard deviation of 3 experiments.


Keywords: MRI, Hyperpolarization, parahydrogen, biomolecule
# 219

Magnetic Resonance Imaging of protein load in biocatalisys reactors (#300)

A. Egimendia1, D. Grajales2, 3, J. C. Mateos2, D. Padro1, F. López-Gallego2, 4, P. Ramos-Cabrer1, 4

1 CIC biomaGUNE, Magnetic Resonance Imaging, Donostia San Sebastián, Spain
2 CIC biomaGUNE, Heterogeneous Biocatalysis Group, Donostia San Sebastián, Spain
3 CIATEJ A.C., Industrial Biotechnology Depatment, Zapopan, Jalisco, Mexico
4 Ikerbasque, Basque Foundation for Science, BIlbao, Spain


Protein immobilization is key enabling technology for flow biocatalysis. With this purpose, many different immobilization protocols and characterization techniques have been developed in the last decades. However, examples where the proteins are directly immobilized on ready-to-use reactors are scarce, likely due to the lack of analytical tools to monitor the protein immobilization in flow in a non invasive manner.


In this work, we have unprecedentedly exploited Magnetic Resonance Imaging (MRI) for the characterization of in-flow protein immobilization on pre-packed bed columns. This concept was proven by immobilizing a green fluorescence protein in flow, using a Gd-DTPA solution as mobile phase. T1(RARE with variable TR) and T2 (MSME) maps where acquired using a 7T Bruker Biospec system.


Protein immobilization on packed column induces a measurable reduction on T2 relaxation times that is proportional to the amount of fixed protein. MRI image analysis reveals that both protein concentration of the flushed solution and flow rate play key roles to control the protein spatial organization across the packed-bed reactor. This analytical tool coupled to the in-flow protein immobilization has been expanded to more industrially relevant enzymes such as the lipase from Thermomyces lanuginosus, achieving a ready-to-use reactor packed with an heterogeneous biocatalyst with high expressed activity (up to 3000 U x g-1) and high stability (75% residual activity after 1h incubation at 60 ºC).


Introducing new analytical tools during the fabrication of heterogeneous biocatalysts will contribute to make the process of immobilizing proteins on solid carriers more rational than currently is.


We want to acknowledge the Spanish ministry of economy and compentence (SAF2014-53412-R), the Basque Government (PC2015-1-05 (53-80)) and COST action CM103-System biocatalysis for financial support. we would like to thank IKERBASQUE (the Basque foundation for Science) for the funding of FLG and PRC, and CONACYT for granting DG (Beca mixta 291212).

Biocatalysis reactor
Picture (A), T2 map (B) and, fluorescence microscopy at two locations (black and white dot) (C) of a bed column loaded with protein.
Keywords: MRI, biocalatysis, bioreactor, protein loading
# 220

Respiratory-Gated B0 Field Stabilisation for High Resolution Mouse Brain Imaging (#148)

P. Kinchesh1, S. Gilchrist1, N. Zarghami1, A. Khrapichev1, N. R. Sibson1, S. Smart1

1 University of Oxford, CRUK-MRC Oxford Institute for Radiation Oncology, Oxford, United Kingdom


3D multi gradient echo (MGE) enables efficient high-resolution scanning with T2* contrast for the detection of USPIO/MPIO. Such scanning causes gradient heating, inducing a B0 drift that predominantly affects image registration in the slow phase-encoding dimension of 3D MGE images. For sequential phase-encode order successive echoes appear shifted and distorted in image space. A real-time adaptive B0 correction during 3D MGE scanning has been developed to reduce this effect. The respiratory-gated B0 correction is demonstrated to outperform ungated B0 correction and improves image fidelity.


MRI was performed at 7 T using a 3D MGE scan that included a navigator echo collection, frequency offset calculation and respiration-gated update of the magnetic field strength applied once every 30 TRs, Fig. 1. The scan was applied to unilaterally inflammed mouse brain following delivery of VCAM-MPIO. Scans were acquired with and without B0 correction.


B0 correction resulted in images that were correctly positioned even though a >160 Hz temperature-dependent field shift was invoked during the scan. In the absence of B0 correction images were shifted in a TE-dependent manner, and in the absence of respiratory-gated B0 correction, images were ghosted. Even small, <30 Hz, field shifts were shown to cause image misregistration and blurring. Through the combined use of B0 correction with the conditional, respiratory gated B0 update maximum image stability was achieved and the sharpest images were produced with a scan time increase <3.5 %. Images of mouse brain featuring MPIO are shown for scans run with respiratory gating and B0 correction, with ungated B0 correction, and with no B0 correction in Fig 2. The improved image fidelity of respiratory-gated and ungated B0 correction increased the hypointense brain voxel count measured in uncorrected data by 38% and 24% respectively.


Respiratory-gated B0 corrections provide an efficient means for avoiding the temperature dependent field instability that results from the use of gradient sets that can heat the magnet bore. This allows acquisition of images that are not corrupted by these field drifts and maximum resolution can be achieved.

Fig. 1. Pulse sequence

3D MGE scan with embedded navigator acquisition that maintains the MR steady state.  MRI signal frequency evaluation is used to adjust the voltage across the B0 coil during the same TR. A respiration gated B0 update is performed only if the MRI measurement occurs during a stable inter breath period.

Fig. 2. Brain images.

Sum of squares combination image of a slice of MGE data acquired from inflammed mouse brain with VCAM-MPIO. Image fidelity is highest  with respiratory-gated B0 correction. Ghosting is present without respiratory gating, and blurring is present in the absence of B0 correction.

# 221

Fast metabolite mapping for hyperpolarized agents using a me-bSSFP sequence at 9.4 T (#102)

J. G. Skinner1, C. Müller1, J. Leupold1, J. Hennig1, D. von-Elverfeldt1, J. - B. Hövener2

1 Medical Center - University of Freiburg, Department of Radiology, Medical Physics, Freiburg, Baden-Württemberg, Germany
2 University of Kiel, Department of Radiology and Neuroradiology, Kiel, Schleswig-Holstein, Germany


Dynamic nuclear polarization has made it possible to enhance signals of 13C-labelled metabolic MR tracers by up to five orders of magnitude,1 and a number of such tracers have been shown to be markers of cancer.2,3 These tracers can be mapped at high temporal and spatial resolutions using specialized MRI sequences which take into account the transient nature of the hyperpolarized signal. To this end, we present a multi-echo balanced steady-state free precession sequence4 (me-bSSFP) implemented at a 9.4 T pre-clinical MRI and demonstrate MRSI of thermally polarized [1-13C] lactate and glycine.


MR-scanner: small-bore 9.4 T system (Bruker BioSpec 94/20) with 1H-13C volume coil (RAPID).

13C-MRI: custom me-bSSFP sequence, α = 40°, pulse bandwidth (BW) = 5 KHz (Gaussian), FOV = 112 x 112 mm, slice thickness = 40 mm, matrix = 32 × 32, 6 echoes, TE/TR = 4.25/10 ms, readout BW = 500 KHz, 200 acquisitions. Reconstructed at 507 Hz ([1-13C]lactate) and -507 Hz ([1-13C]glycine) based on IDEAL iterative least squares algorithm with bipolar gradient and T2* correction (MATLAB),5 2 × zero-filling, and filtered with the center lobe of a 2D sinc.

13C-spectroscopy: None-localised with 90° excitation pulse.

1H MRI: FLASH based localiser sequence.

Model solutions: aqueous 5 ml, 2.2 M, 99% enriched [1-13C]lactate and 4 ml, 2.0 M, 99% enriched [1-13C]glycine, both in 5 ml glass vials.


In many active cancer tumours, pyruvate (173 ppm6) is converted to lactate (185 ppm6) via lactate dehydrodgenase.2,6 The goal of the sequence presented here is to image each such resonance separately, and high-resolution images with clear separation of both metabolites were achieved with an in plane resolution of (1.75 mm)2 or (3.5 mm)2, with or without zero filling respectively (figure 2 c, d). As pyruvate is chemically unstable in solution, glycine – with a similar resonance frequency (175 ppm7) - was used instead. Bloch simulations were performed beforehand to obtain ideal values for TR and alpha and a high readout BW was used to minimise well known bSSFP banding artefacts.

While this method was used at clinical field strengths before,4,8 to our knowledge, this is the first demonstration at a considerably larger field of 9.4 T where inhomogeneities are more pronounced and spectral separations are much larger - requiring a larger bandwidth. 


Metabolite mapping of thermal [1-13C]lactate and [1-13C]glycine was demonstrated at 9.4 T using a me-bSSFP sequence and IDEAL reconstruction with an in-plane voxel size of (1.75 / 3.5 mm)2, with and without zero filling respectively.

In a hyperpolarised (HP) case where averaging is not required, the scan time for a single scan with these settings reduces to 1 s, making the sequence well suited to frequency resolved, dynamic studies of in vivo kinetics using HP metabolic tracers.


1. Ardenkjær-Larsen, Jan H., et al. "Increase in signal-to-noise ratio of> 10,000 times in liquid-state NMR." Proceedings of the National Academy of Sciences 100.18 (2003): 10158-10163.

2. Nelson, Sarah J., et al. "Metabolic imaging of patients with prostate cancer using hyperpolarized [1-13 C] pyruvate." Science translational medicine 5.198 (2013): 198ra108-198ra108.

3. Gallagher, Ferdia A., et al. "Production of hyperpolarized [1, 4-13C2] malate from [1, 4-13C2] fumarate is a marker of cell necrosis and treatment response in tumors." Proceedings of the National Academy of Sciences 106.47 (2009): 19801-19806.

4. Leupold, Jochen, et al. "Fast multiecho balanced SSFP metabolite mapping of 1H and hyperpolarized 13C compounds." Magnetic Resonance Materials in Physics, Biology and Medicine 22.4 (2009): 251-256.

5. Peterson, Pernilla, and Sven Månsson. "Fat quantification using multiecho sequences with bipolar gradients: investigation of accuracy and noise performance." Magnetic resonance in medicine 71.1 (2014): 219-229.

6. Merritt, Matthew E., et al. "Hyperpolarized 13C allows a direct measure of flux through a single enzyme-catalyzed step by NMR." Proceedings of the National Academy of Sciences104.50 (2007): 19773-19777.

7. Wishart, David S., et al. "HMDB: a knowledgebase for the human metabolome." Nucleic acids research 37.suppl_1 (2008): D603-D610.

8. Milshteyn, Eugene, et al. "Development of high resolution 3D hyperpolarized carbon-13 MR molecular imaging techniques." Magnetic resonance imaging 38 (2017): 152-162.


This project has received funding from: 1) the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 642773. 2) DKTK. 3) Emmy Noether Programme of the DFG, award No HO-4604/2-1.

Figure 1: me-bSSFP sequence diagram

1. A rf-pulse with flip angle fa = α/2 & phase ϕ = 0° excites a slice of spins.
2. Preparation pulses with fa = α and alternating ϕ [0°; 180°] can be applied to reach steady state.
3. When steady state is reached phase encoding gradients (P) and ADC is turned on.

Figure 2: a) 1H MRI, b) 13C spectroscopy and c), d), 13C MRSI of [1-13C]lactate and [1-13C]glycine
reconstructed in c) and d)  at -507 Hz and 507 Hz, respectively. High-resolution images with clear separation of both metabolites was achieved with an in plane resolution of (1.75 mm)2 or (3.5 mm)2, with or without zero filling respectively. Glycine was used here as a model for pyruvate due to its instability in-solution.
Keywords: metabolic imaging, MRSI, hyperpolarised, spectroscopic imaging
# 222

Signal to noise comparison of metabolic imaging methods on a clinical 3T MRI (#62)

C. A. Müller1, 2, R. B. Hansen3, J. G. Skinner1, A. Eldirdiri1, J. Leupold1, J. Hennig1, J. H. Ardenkjaer-Larsen3, A. E. Hansen4, J. - B. Hövener5

1 Medical Center - University of Freiburg, Faculty of Medicine, University of Freiburg, Dept.of Radiology, Medical Physics, Freiburg, Baden-Württemberg, Germany
2 German Cancer Research Center (DKFZ), German Cancer Consortium (DKTK), partner site Freiburg, Heidelberg, Baden-Württemberg, Germany
3 Technical University of Denmark, Department of Electrical Engineering, Kgs. Lyngby, Denmark
4 Reigshospitalet, University of Copenhagen, Department of Clinical Physiology, Nuclear Medicine & PET, Copenhagen, Denmark
5 University medical Center Schleswig-Holstein, University of Kiel, Section for Biomedical Imaging, Molecular Imaging North Competence Center (MOINCC), Department of Radiology and Neuroradiology, Kiel, Germany


MRI with hyperpolarized tracers has enabled new diagnostic applications, e.g. metabolic imaging in cancer research. However, the acquisition of the transient, hyperpolarized signal with spatial and frequency resolution requires dedicated imaging methods. Here, we compare three promising candidates for 2D MR spectroscopic imaging (MRSI): (i) multi-echo balanced steady-state free precession (me-bSSFP),1,2 (ii) echo planar spectroscopic imaging (EPSI) sequence and (iii) phase-encoded, pulse-acquisition chemical-shift imaging (CSI).


All sequences were implemented on a clinical 3T PET/MR system (Siemens) equipped with a 1H-13C birdcage coil (Rapid Biomedical). A multi-compartment container filled with Gd-doped (c=0.23%v/v) aqueous model solutions of 1.0 M 13C-bicarbonate (bi), 13C-urea (ur) and [1-13C]-acetate (ac) was used for imaging.

For better comparison, field of view (FOV), voxel size and scan time was identical for all methods. Faster methods were repeated until one acquisition of the CSI was completed resulting in 1, 16 or 670 averages for CSI, EPSI or me-bSSFP (Tab. 1).

Chemical shift (CS) maps were calculated with dedicated methods for each sequence, signal-to-noise (SNR) ratios were obtained by dividing the mean signal of a region of interest (ROI) by the mean signal of an ROI, where no 13C-signal is expected.


Non-localized 13C spectroscopy revealed three peaks for ac (+328 Hz), ur (-252 Hz) and bi (-328 Hz) (Figure 1). In a single 13C MRSI acquisition, CSI provided most SNR. Over the entire scan time, me-bSSFP yielded more SNR (max. 146, NA=670) than CSI (max. 105, NA=1) and EPSI (max. 66, NA=16). The direct comparison of the SNR is possibly inaccurate (due to the different number of averages, signal evolution and the reconstruction algorithms), but a similar trend is expected using HP tracers in vivo.

With respect to the max. signal, me-bSSFP revealed the smallest artifacts (which are not apparent on a linear scale, Figure 1, C.1-3).The logarithmic scale (C.4-6) reveals two kinds of artifacts for all methods: localized signal-leakage (e.g. signal at the position of ur in the bi image), as well as signal banding.

The SNR and image quality provided by me-bSSFP may allow shortening the acquisition time further. Simulation of signal evolution may also help understanding the observed results.


We presented and compared 13C MRSI of thermally polarized model solutions using the methods me‑bSSFP, CSI and EPSI. For a given time, imaging volume and resolution, me-bSSFP results in highest SNR at the reconstructed metabolite maps and least artefacts (with lowest flip angle and shortest scan time per average).

These results make me-bSSFP a suitable candidate for 3D MRSI of large FOVs. Metabolite separation and robustness appears to be sufficient in vivo, too, as is indicated by first results of a different study that is ongoing.


1. Leupold, J. et al. Fast chemical shift mapping with multiecho balanced SSFP. Magn. Reson. Mater. Phys. Biol. Med. 19, 267–273 (2006).

2. Leupold, J. et al. Fast multiecho balanced SSFP metabolite mapping of 1H and hyperpolarized 13C compounds. Magn. Reson. Mater. Phys. Biol. Med. 22, 251–256 (2009).

3. Reeder, S. B. et al. Multicoil Dixon chemical species separation with an iterative least-squares estimation method. Magn. Reson. Med. 51, 35–45 (2004).

4. Peterson, P. & Månsson, S. Fat quantification using multiecho sequences with bipolar gradients: Investigation of accuracy and noise performance. Magn. Reson. Med. 71, 219–229 (2014).


Support by the following programs is acknowledged: 1) the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 642773. 2) DKTK. 3) Emmy Noether Programme of the DFG, award No HO-4604/2-1.

"> Table 1:
Acquisition parameters for 13C MRSI. Metabolic maps (for ac: +328 Hz, ur: -252 Hz, bi: -328 Hz) were obtained by using a Dixon based IDEAL3,4, for the me-bSSFP, and peak amplitude per voxel reconstruction for CSI and EPSI. SNR was calculated for ROIs (ac, bi, ur), which were manually segmented based on the 1H MRI (Figure 1,B.).
Figure 1: 13C MR-spectrum (A.), 1H MRI (B.) and 13C MRSI (C., linear 1-3 & logarithmic 4-6 scale)
of acetate, bicarbonate and urea using me-bSSFP, CSI and EPSI. Spatial localization, metabolite separation as well as high SNR (white numbers) was achieved with each method, with me-bSSFP giving highest SNR. Note that some signal leaking is apparent in all logarithmic images. Due to small chemical shifts the reconstructed metabolite maps can have artifacts at the corresponding positions.
Keywords: Hyperpolarization, metabolic imaging, EPSI, CSI, me-bSSFP
# 223

Feasibility of preoperative staging with USPIO-MRI in patients with resectable esophageal carcinoma (PRECIES study) (#299)

D. de Gouw1, B. R. Klarenbeek1, R. S. van der Post2, M. C. Maas3, J. J. Hermans3, C. Slagt4, M. M. Rovers5, T. W. Scheenen3, C. Rosman1

1 Radboud university medical center, department of surgery, Nijmegen, Netherlands
2 Radboud university medical center, department of pathology, Nijmegen, Netherlands
3 Radboud university medical center, department of radiology and nuclear medicine, Nijmegen, Netherlands
4 Radboud university medical center, department of anesthesiology, pain and palliative medicine, Nijmegen, Netherlands
5 Radboud university medical center, department of health technology, Nijmegen, Netherlands


In operable patients suffering from esophageal cancer, the percentage of patients without metastatic lymph nodes after neoadjuvant chemoradiotherapy (nCRT) is 69%. Lymph node dissections during esophagectomy may be omitted in these patients, reducing associated morbidity. Recently, MRI with ultrasmall superparamagnetic iron oxide nanoparticles (USPIO, ferumoxtran-10[1]) has been reintroduced to detect metastatic lymph nodes in prostate cancer[2]. The aim of this study is to assess the feasibility of USPIO-MRI to detect loco-regional lymph node metastases in patients with esophageal cancer.


USPIO-nanoparticles are intravenously infused 24 to 36 hours before MRI. USPIO-enhanced MRI is performed before and after nCRT. After nCRT, but prior to surgery, the anesthetized patient is examined in the MR system of the hybrid MITeC operation room with controlled mechanical ventilation. During controlled prolonged apneu, a four-minute iron-sensitive MRI acquisition is used to visualize suspicious esophageal lymph nodes without motion artefacts. Resected specimens, still containing USPIO, are measured ex-vivo in a preclinical 7T MR system before histopathological examination. A radiological assessment of the presence of suspicious lymph nodes in-vivo is matched to the ex-vivo nodes on preclinical MRI, providing the ground truth for the presence of metastases.


Currently, three patients were included in the study of which one patient has been examined before and after nCRT. MRI under anesthesia prior to surgery with controlled mechanical ventilation was possible, resulting in a clinically relevant spatial resolution to visualize possible malignant lymph nodes. Suspicious nodes were identified and could be matched using corresponding anatomical landmarks to the ex-vivo MRI, which showed good visual agreement with esophageal specimen after resection.


A successful method was proposed to validate USPIO-enhanced MRI to detect metastatic lymph nodes in patients with esophageal cancer. Matching ex-vivo USPIO-MRI images with histopathology results provides direct information for validation of in vivo USPIO-MRI and characteristics of loco-regional lymph nodes. Final results on the feasibility of USPIO-MRI to detect metastatic lymph nodes after nCRT are still awaited. Feasibility and preliminary values of the accuracy of the technique are the starting point for a phase two study.


1. Harisinghani, Mukesh G., et al. "Noninvasive detection of clinically occult lymph-node metastases in prostate cancer." New England Journal of Medicine 348.25 (2003): 2491-2499.

2. Fortuin, Ansje S., et al. "Ultra‐small superparamagnetic iron oxides for metastatic lymph node detection: back on the block." Wiley Interdisciplinary Reviews: Nanomedicine and Nanobiotechnology (2017)

Figure 1
Mechanism of action of injected nanoparticles [1]. With water-selective, iron-sensitive MRI the signal of healthy nodes disappears, whereas metastatic lymph nodes, in which the nanoparticles do not accumulate, retain MRI signal.
Figure 2
Example of iron-sensitive MRI acquisition on the hybrid MITeC operation room with controlled mechanical ventilation. The yellow circle represents a metastatic lymph node.