REVIEW ARTICLE

Frontiers of Sodium MRI Revisited: From

Cartilage to Brain Imaging
Olgica Zaric, PhD,1 Vladimir Juras, PhD,2,3* Pavol Szomolanyi, PhD,2 Markus Schreiner, MD,4
Marcus Raudner, MD,2 Chiara Giraudo, PhD,5 and Siegfried Trattnig, MD1,2,6

Sodium magnetic resonance imaging (23Na-MRI) is a highly promising imaging modality that offers the possibility to noninva-
sively quantify sodium content in the tissue, one of the most relevant parameters for biochemical investigations. Despite its
great potential, due to the intrinsically low signal-to-noise ratio (SNR) of sodium imaging generated by low in vivo sodium con-
centrations, low gyromagnetic ratio, and substantially shorter relaxation times than for proton (1H) imaging, 23Na-MRI is
extremely challenging. In this article, we aim to provide a comprehensive overview of the literature that has been published in
the last 10–15 years and which has demonstrated different technical designs for a range of 23Na-MRI methods applicable for
disease diagnoses and treatment efficacy evaluations. Currently, a wider use of 3.0T and 7.0T systems provide imaging with
the expected increase in SNR and, consequently, an increased image resolution and a reduced scanning time. A great interest
in translational research has enlarged the field of sodium MRI applications to almost all parts of the body: articular cartilage
tendons, spine, heart, breast, muscle, kidney, and brain, etc., and several pathological conditions, such as tumors, neurological
and degenerative diseases, and others. The quantitative parameter, tissue sodium concentration, which reflects changes in
intracellular sodium concentration, extracellular sodium concentration, and intra–/extracellular volume fractions is becoming
acknowledged as a reliable biomarker. Although the great potential of this technique is evident, there must be steady techni-
cal development for 23Na-MRI to become a standard imaging tool. The future role of sodium imaging is not to be considered
as an alternative to 1H MRI, but to provide early, diagnostically valuable information about altered metabolism or tissue func-
tion associated with disease genesis and progression.
Level of Evidence: 1
Technical Efficacy Stage: 1
J. MAGN. RESON. IMAGING 2021;54:58–75.

I N VIVO SODIUM MAGNETIC RESONANCE IMAG-

ING (23Na-MRI) is an advanced imaging modality that
offers noninvasive metabolic imaging for early and accurate
healthy tissue, the large concentration gradient between the
cells and the extracellular space is maintained primarily by the
energy-dependent sodium–potassium pump (Na+/K+-ATPase).
diagnosis, characterization, and treatment efficacy evaluations If the cell membrane or energy metabolism is destroyed, it can
of several diseases. drive an impairment of the sodium–potassium pump, which
The 23Na ion is one of the most important electrolytes will further increase intracellular sodium concentration and
in the human body and it plays a crucial role in osmoregula- induce cell malfunction and, eventually, cell death.2
tion and physiology of the cell.1 A large sodium concentration An in vivo 23Na MRI can noninvasively provide valu-
gradient is required for appropriate function of the cells, and able information on cell metabolism; however, the technique
intracellular sodium concentration (ISC) is one-tenth of that in is extremely challenging. 23Na has a spin of 3/2 and belongs
the extracellular space (ESC) (10–15 vs. 100–150 mmol/L). In to the group of quadrupole nuclei. The sensitivity of nuclear

View this article online at wileyonlinelibrary.com. DOI: 10.1002/jmri.27326

Received Mar 18, 2020, Accepted for publication May 20, 2020.

*Address reprint requests to: V.J., High-Field MR Center, Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna,
Lazarettgasse 14, 1090 Vienna, Austria. E-mail: vladimir.juras@meduniwien.ac.at

From the 1Institute for Clinical Molecular MRI in the Musculoskeletal System, Karl Landsteiner Society, Vienna, Austria; 2High-Field MR Center, Department of
Biomedical Imaging and Image-guided Therapy, Medical University of Vienna, Vienna, Austria; 3Department of Imaging Methods, Institute of Measurement
Science, Slovak Academy of Sciences, Bratislava, Slovakia; 4Deartment of Orthopaedics and Trauma Surgery, Medical University of Vienna, Vienna, Austria;
5
Radiology Institute, Department of Medicine, DIMED Padova University Via Giustiniani 2, Padova, Italy; and 6Christian Doppler Laboratory for Clinical
Molecular MRI, Christian Doppler Forschungsgesellschaft, Vienna, Austria
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.

58 © 2020 The Authors. Journal of Magnetic Resonance Imaging published by Wiley Periodicals LLC.
on behalf of International Society for Magnetic Resonance in Medicine.
 Zaric et al.: Frontiers of Sodium MRI Revisited

magnetic resonance (NMR) experiments is given by signal-to- high efficacy is an important factor in achieving high SNR
noise-ratio (SNR) and, for 23Na, it is 9.2% that of proton and resolution in an acceptable scan time.
(1H) sensitivity. There are three major factors that limit the The development of non-Cartesian sequences for sodium
SNR of sodium imaging: the concentration of the 23Na ion imaging that aim to maximize the efficiency of k-space sampling
in vivo is low and is in the range from 15 mM (muscle started in early 1990 and continues today. Hilal et al in 1992
15–30 mM) to 350 mM (articular cartilage 250–350 mM); demonstrated the feasibility of a three-dimensional radial projec-
relaxation times of sodium are about two orders of magnitude tion (3DRPI) acquisition method in which the center of k-space
smaller than those of protons (T1: 15–55 msec, T2short: 0.5– is densely sampled, while the edges of k-space are under-
2.5 msec, and T2long: 10–65 msec), and the gyromagnetic sampled.13 Almost a decade later, Nagel et al developed a pulse
ratio of sodium (γNa = 11.26 MHz/T) is approximately four sequence, called the density-adapted three-dimensional radial
projection reconstruction pulse sequence (DA-3DPR) that was
times lower than that for hydrogen (γH = 42.57 MHz/T).3
designed on the basis of a conventional 3DPR sequence. The
Fortunately, most of these limitations were abolished by the
sequence was modified such that the sampling density in the
introduction of 3.0T (high-field) and ≥7.0T (ultrahigh-field)
outer part is kept constant, while an inner sphere of k-space is
scanners and an essential SNR increase was achieved
sampled with no density adaptation. This approach enabled sub-
(SNR  B01.65).4 The modern MRI systems are currently
stantially improved image quality and an increase in SNR.14
equipped with strong gradients and sophisticated electronics,
In parallel, Boada et al developed the twisted projection
multiarray coils, and fast (non-Cartesian) sequences that allow imaging (TPI) sampling scheme.15 This method replaced radial
further SNR enhancement of 23Na-MRI.5 linear gradients with time-varying gradients with a short radial
The purpose of this article is to provide readers with an component to move promptly away from the center of k-space,
overview of the current literature, including methodological followed by a time-inconstant gradient that dismissed a 3D spi-
improvements in the 23Na-MRI technique and its application ral trajectory.15,16 TPI uses spokes up to a threshold k-space
in preclinical and clinical investigations at 3.0T and 7.0T. value, at which point the trajectory transits into a 3D spiral-
Lately, several review articles that cover some important like trajectory.16 This sequence was shown to be the most
aspects of the technique have been published, including bio- favorable for in vivo quantitative sodium measurements in
medical applications for sodium imaging, evaluations of carti- brain.17 Shortly after, Boada’s group developed a sequence
lage repair techniques and osteoarthritis, sodium imaging of called the acquisition-weighted stack of spirals (AWSOS)
the heart and the brain, 23Na-MRI radiofrequency sequence, which attempts to decrease excitation and acquisition
(RF) systems for brain and musculoskeletal or body imaging, delay by introducing a variable slice-encoding, and separating
and quantitative sodium imaging, etc.3,6–12 Considering that slice thickness from in-plane resolution to lower the number of
the majority of previously published review articles were slice-encoding steps, while at the same time, using a spiral
focused on a single organ or disease, we attempt here to sum- readout to increase the efficiency of in-plane acquisition.18 In
marize a significant amount of new publications since the last addition to the above-mentioned sequences, other k-space tra-
comprehensive sodium review article appeared. In this work, jectories, ie, employing distributed spiral trajectories, such as
we aim to provide a review of recent 23Na-MRI findings SPRITE (Single-Point Ramped Imaging with T1-Enhance-
when applied in nearly all parts of the body and revisit the ment) and FLORET (Fermat Looped, Orthogonally Encoded
frontiers of sodium imaging in modern medicine. Trajectories), have been proposed and used for 23Na-MRI.19,20
The necessity to increase SNR and image resolution of
Imaging Sequences 23
Na-MRI was accompanied by the needs of researchers and
23
Na in tissue has a biexponential relaxation behavior, which clinicians for selective measurements of intracellular sodium
means that most of the 23Na signal is lost within a few milli- changes. Their goal was to develop a method that could pro-
seconds. The rapid signal loss renders quantitative imaging vide more specific information and an image biomarker of
challenging. The short T1 relaxation time facilitates a short compromised ionic homeostasis. Biexponential relaxation,
repetition time (TR) and fast averaging, which can partially typical of sodium nuclei in slow motion in the intracellular
compensate for the low intrinsic SNR. Imaging with pulse space, also observable in the extracellular space, can pass
sequences designed in a way that enables measurements with through a state of multiple quantum coherences (MQC).21
a very short echo time (TE, 1 msec), such as an ultrashort MQ filtering techniques are essentially sensitive to changes in
echo time (UTE) sequence, may overcome the challenge of intracellular sodium concentration, and therefore, these tech-
the very short T2 of sodium nuclei. On the other hand, a niques are ideally suited for the noninvasive, in vivo observa-
short TE will substantially limit the duration of the signal tions of ISC level changes. Tsang et al confirmed the
readout and cause image blurring and a decrease in SNR. possibility of double quantum-magic angle (DQ-MA) signal
However, UTE acquisition methods are generally slow com- generation from the human brain.22 Initial experiments, per-
pared to echo-sampled MRI and the sampling of k-space with formed with the highest nominal isotropic resolution of

July 2021 59
Journal of Magnetic Resonance Imaging

(8.4 mm)3 and 48 minutes scanning time, demonstrated the Image Reconstruction Methods
presence of sodium nuclei in ordered environments. Later on, MR is fundamentally a low SNR imaging method, and the
the DQ-MA method was used to visualize the sodium signal use of nuclei other than protons for imaging purposes may
that originates from anisotropic structures, such as muscle have a considerable impact on the resulting image SNR. The
fibers.23 acceleration of image acquisition allows significant improve-
However, the main drawback of the MQ techniques is ments in SNR per unit time.
their low SNR. Using a train of three RF pulses (with the Pioneering work in this field was done by Qian et al,30
corresponding phase cycling), Hancu et al demonstrated that who demonstrated the advantages of parallel imaging with a
3D triple-quantum-filtered (TQF) sodium images of the TPI trajectory. In computer simulations, these authors tested
human brain can be acquired at moderate field strengths the TPI-SENSE (sensitivity encoding) method with an
(3.0T) with examination times acceptable for clinical applica- applied acceleration factor of 5.53 and simulations were veri-
tions.24 Tsang et al, however, demonstrated the utilization of fied on the proton human head studies. The results showed
sufficiently long RF pulses and a reduced TR that might lead that parallel sodium imaging may reduce scan time substan-
to further SNR enhancement for TQF images.25 tially compared to the conventional TPI acquisition
Essentially, low SNR was not the only issue with TQF (3 minutes vs. 16 minutes) without a substantial loss in image
sequences; they entail a lower power deposition and they are quality.30 Several years later, Benkhedah et al investigated fea-
still prone to image artifacts caused by off-resonances. There- sibility of the adaptive combination reconstruction (ADC)
fore, development continued and a method called method for multichannel coil array data combining and
biexponentially weighted (BW) 23Na-MRI was developed. found that mean SNR may be increased from 8–50% com-
The initial results published were very promising, and dem- pared to standard sum-of-squares (SOS) image reconstruc-
onstrated a three times higher image SNR obtained by BW tion.31 Additionally, an alternative possibility for high-quality
23
Na compared to the six-step phase-cycling TQF MRI.26 image reconstruction from undersampled datasets is a com-
For the excitation and detection of MQC, a three-pulse prep- pressed sensing (CS) method. Gnahm et al recently showed
aration is applied during the pulse train, and two images are that an anatomically weighted second-order total variation
generated: a spin-density-weighted image (SDW), and a reconstruction of 23Na MRI using prior information from 1H
single-quantum-filtered image (SQF). The BW image is MRI (AnaWeTV) provides an increase in image quality due
derived by subtracting the SQF image from the SDW image to maintained tissue borders and reduced partial volume
and shows the signal from sodium ions with biexponential effects (PVE).32 A method that uses a 3D, dictionary-learning
relaxation. Nagel et al proposed a relaxation-based CS reconstruction algorithm (3D-DLCS) for the reconstruc-
(RW) method as a possibility to differentiate the tissue tion of undersampled 3D radial 23Na data was presented by
sodium signal based on differences in 23Na relaxation proper- Behl et al.33 Using the dictionary, it is possible to learn the
ties in different tissues.27 Two different approaches based on sparsifying transform with a K-singular-value-decomposition
the DA-3DPR sequence were provided: the first used an (K-SVD) algorithm. The same method was shown to be feasi-
inversion recovery (IR) preparation pulse to exploit T1- ble for tissue sodium concentration (TSC) quantification of
differences of 23Na ions, and the second approach was based skeletal muscle.34
on 23Na multiecho sequences to exploit differences in T2*-
relaxation times. However, the relaxation-weighted 23Na sig-
nal describes a compartment defined by 23Na relaxation prop- Quantitative Sodium MRI
erties and does not necessarily correspond to the intra- or A comprehensive review article about the evolving role of
extracellular space.28 quantitative sodium imaging in medicine was recently publi-
Although several different methods for sodium DQ shed by Thulborn.12 Different applications of 23Na-MRI for
and TQ coherence discrimination have been proposed in quantitative analysis, particularly of the musculoskeletal sys-
the literature, a clear confirmation that intra- and extracellu- tem, were reviewed by Bangerter et al.35
lar signals can be separated in vivo is still lacking..29 MQC In a study published in 2010, Lu et al presented a flow
from both the intra- and extracellular spaces are similar chart for an image reconstruction and tissue sodium quantifi-
because of the labile macromolecular interactions that result cation (TSC) process that included B0, B1 mapping, and eddy
in comparable relaxation properties (T2 values) of sodium current corrections, assuming that 1H and 23Na coils have
nuclei in different environments. Therefore, the difference the same eddy current characteristics.16 These authors investi-
between intra- and extracellular sodium signals cannot be gated the effect of B0 corrections on TSC measurements and
established based on relaxation time constant characteristics. found that a substantial improvement in image quality and
This is certainly one of the most important topics on which quantification accuracy can be achieved by introducing B0
the sodium MRI community should come to a consensus in corrections in an image postprocessing scheme. The effect of
the future. B1 inhomogeneity on absolute 23Na concentration

60 Volume 54, No. 1

 Zaric et al.: Frontiers of Sodium MRI Revisited

quantification was further studied by Lommen et al, who pro- Sodium Imaging of Cartilage
posed a method for simultaneous B1 mapping (implemented Cartilage Repair
into the 3D radial projection sequence) and 23Na imaging to Within the last two decades, a large number of cartilage repair
increase accuracy and to reduce measurement time.36 surgical techniques were developed, such as bone marrow
Niesporek et al developed an algorithm for partial volume stimulation techniques (Pridie drilling, microfracturing,
effect (PVE) corrections and demonstrated a high perfor- MFX), first-, second-, and third-generation cell-based autolo-
mance of the method to reduce the discrepancy between the gous chondrocyte implantation (ACI), autologous
measured and expected sodium concentration value by 11% osteochondral transplantation (AOT), and cell-free implant
to a mean PVE-caused inconsistency of 5.7% after techniques. A recent review article by Zbyn et al provided an
correction.37 extensive overview of different repair techniques evaluations
using 23Na-MRI.8
Proton MRI allows the morphological assessment of the
RF Systems for Sodium MRI cartilage and cartilage repair tissue and, more recently, bio-
RF coils are one of the most important elements for high- chemical assessment using the glycosaminoglycan (GAG)
quality, high-SNR sodium MRI. There are a number of chal- chemical exchange saturation transfer (CEST) technique.
lenges connected to RF coil design, first among them the Information about the fine structure or biochemical composi-
operating frequencies, which have to be low; consequently, tion and the quality of the repair tissue is, however, of partic-
this can lead to problems maintaining good coil loading, and ular interest, as it has been demonstrated that the
thus, directly limits coil sensitivity. The performances of dif- composition of the repair tissue may affect the long-term out-
ferent RF coil designs for sodium imaging of brain and mus- come. In addition to gagCEST, 23Na-MRI is able to assess
culoskeletal applications have been extensively discussed in a changes in ion content connected to GAG molecules. The
review article published by Wiggins et al.10 The overview of GAGs are negatively charged molecules and belong to the
the advantages and disadvantages of various RF coil topolo- group of the most important constituents of cartilage. GAG
gies for sodium body imaging has been reported by Bangerter is the central point of molecular investigations of the cartilage
et al.11 tissue because it has a significant influence on its function
Because of the very short T2 of sodium, short RF pulses and homeostasis. Furthermore, the GAG content significantly
are required to minimize echo time, which drives a highly correlates with the biomechanical properties of cartilage, in
efficient transmit coil to limit the maximum voltages needed. particular, compressive stiffness.46 In articular cartilage, the
The benefit of a linear increase in SNR with field strength is negatively charged GAGs are counterbalanced by positively
diminished by higher specific absorption rates (SARs), which charged sodium ions; thus, the sodium concentration can be
require a lengthening of the RF pulse durations or repetition used as an indirect indicator of the amount of GAG, which,
times. in turn, can be noninvasively assessed using sodium imag-
One of the major demands of clinicians is that the ing47,48 (Fig. 1). For quantitative measurements, phantoms
same coil provides proton and sodium imaging within one with known sodium concentrations may be placed close to
imaging session. Moreover, to maximize sodium MRI per- the organ under investigation and may provide absolute tissue
formance, it is important to maximize the uniformity of the sodium concentrations.
main B0 static magnetic field. The most efficient and robust The first study that included patients after cartilage repair
way is to calculate the correct shim currents using proton- treatment was published by Trattnig et al in 2010.49 Twelve
based methods to map the B0 field and then transfer shim patients were involved in the study and scanned using a 3D-
parameters to sodium imaging. This involves the use of GRE (gradient echo) sequence with a sodium-only, birdcage
double-tuned, multichannel phased array coils. However, knee coil. Each patient was examined 56 months after
the major issue with dual frequency designs is that they matrix-associated chondrocyte transplantation (MACT). The
degrade the performance at one or both frequencies (1H coil normalized values of the sodium normalized signal intensity
structures have to cope with screening effects as a result of (NMSI) were significantly lower in transplanted tissue
currents induced on the 23Na coil part).10 Recently, Gast (174 ± 53) than in reference cartilage (267 ± 42) (P < 0.001).
et al developed a 23Na-MRI-based method for localized B0 When the results were compared with dGEMRIC (delayed
shimming at 7.0T and showed that an external field inho- Gadolinium Enhanced MRI of cartilage), another GAG-spe-
mogeneity can be reduced up to 77% using this approach.38 cific method, the authors concluded that sodium MRI allows
Many different coil designs, optimized for improved sodium the differentiation between MACT repair tissue and native car-
sensitivity, have been developed and reported in the litera- tilage of patients without the need for contrast agent applica-
ture recently, suitable for almost all body parts, such as the tion. For validation of gagCEST imaging as a new method
knee, the breast, and the brain.10,11,39–45 sensitive to changes in GAG content, Schmitt et al compared

July 2021 61
Journal of Magnetic Resonance Imaging

FIGURE 1: A 39-year-old male patient who had a small cartilage defect less than 15% of the cartilage thickness (ICRS grade 2 lesion)
in the proximal trochlear region of the lateral femoral condyle. (a) A fat-suppressed proton density (FS-PD) image in a sagittal
orientation shows a lesion with intra-chondral signal alterations. Sodium 23Na-MRI images generated (b) 3 days and (c) 4 weeks after
the defect.

gagCEST results at 7.0T with sodium MRI in patients with Cartilage defects can occur also in the ankle joint,
femoral cartilage repair.50 Results based on five MFX and seven mostly after injury or in patients with osteochondritis dis-
MACT patients showed a strong correlation between sodium secans. Cartilage repair procedures used in the knee joint are
and gagCEST values and demonstrated the feasibility of this also performed in the ankle joint. MRI of the ankle joint is
method for the cartilage GAG content assessment. particularly challenging for a number of reasons. The cartilage
Zbyn et al further investigated the quality of newly lining is curved, highly congruent, and significantly thinner
developed repair tissue on femoral condyle cartilage after two than in the knee joint, with an average thickness of
different repair procedures: bone marrow stimulation (BMS) 1.1 ± 0.18 mm for talar cartilage and 1.16 ± 0.14 mm for the
and MACT.51 The NMSI of repaired tissue was significantly cartilage of the distal tibia.55 This leads to increasing prob-
lower in patients after BMS than in those who underwent a lems regarding partial volume effects and SNR. The first
MACT procedure ((164 ± 31) vs. (210 ± 36)) (P = 0.028). results of sodium imaging in cartilage repair technique evalua-
However, the properties of the repair tissue, evaluated by the tions were promising; therefore, Zbyn et al applied a similar
MOCART scoring system based on morphological MR approach to cartilage repair technique evaluations after MFX
evaluations,52 were not significantly different between two and MACT in the ankle joint.48 After biochemical validation
groups of patients after BMS and after MACT treatments of the sodium imaging protocol using ex vivo measurements
(P = 0.915). The sodium results suggest that a higher GAG of ankle joint cadavers, which demonstrated a strong linear
content is common for the sophisticated cell-based MACT correlation between the NMSI and the histochemically
technique compared to the more simple BMS technique. assessed GAG content (r = 0.800; P < 0.001; R2 = 0.639),
This basically means that MACT produces high-quality repair reference values were obtained for healthy volunteers. Subse-
tissue with a more hyaline-like composition compared to quently, patients after MFX and MACT of the talar dome
BMS, which mainly produces fibrous repair tissue with a very were examined. The repair tissue of both treatment groups
low GAG content. Sodium MRI can differentiate between exhibited significantly lower corrected signal intensities (cSI)
repair tissues with different amounts of GAG, and thus, serve compared to healthy cartilage (MFX, P = 0.007; MACT,
for the noninvasive evaluation of the performance of new car- P = 0.008), indicating a lower GAG content than in healthy
tilage repair techniques. reference cartilage of the same patients or the healthy con-
An article by Chang et al compared sodium MRI results trols, but no significant difference was found between both
obtained from in 11 patients with and without fluid suppres- treatment groups, indicating that cartilage composition and
sion, after several different procedures of cartilage repair the effect on repair seems to be different in the ankle joint
(MFX, MACT, osteochondral grafting, juvenile cartilage compared to the knee joint (P = 0.185). No significant differ-
implantation). Examinations were done at 7.0T employing a ence in cSI was found between healthy cartilage of subjects
radial UTE sequence, a sodium-only, birdcage knee coil, and and patients (P = 0.355) as well.
IR sequence with an adiabatic inversion pulse.53 The results Quantitative TSC imaging of articular cartilage is chal-
of the study demonstrated that fluid-suppressed sodium MRI lenging and its accuracy is limited by several factors, such as
is more robust compared to standard sequences. In addition, the partial volume effect caused by the relatively low spatial
sodium concentration in neighboring cartilage to transplanted resolution, signal loss from T2short decay due to the insuffi-
tissue was significantly lower than in healthy cartilage tissue. ciently short TE, and T2 blurring due to rapid signal decay
This is in agreement with the findings of in vitro studies, during the readout. In the future, protocols for TSC quantifi-
which demonstrated that the amount of viable chondrocytes cation that are feasible for clinical applications should be fur-
is lower close to the site of injury.54 ther developed.

62 Volume 54, No. 1

 Zaric et al.: Frontiers of Sodium MRI Revisited

FIGURE 2: A 50-year-old male patient with a lesion in the patellar cartilage in the area crista patellae. (a) A proton density sequence
with fat suppression (FS-PD) demonstrates an early stage degeneration of articular cartilage with a minor chondral signal alteration
and a minor reduction of cartilage thickness, and a surface that appears intact (ICRS grade 1 lesion). (b) Yellow arrow shows the
corresponding area on the sodium image.

23
Na-MRI in Osteoarthritis (OA) The results of all the above-mentioned publications
Proton (1H) MRI methods provide different information confirmed that sodium is a feasible and reliable method for
about the morphology of the knee joint, but the more valu- the assessment of early OA changes. Findings of the clinical
able diagnostic information is that regarding compositional studies support the concept that compositional cartilage
changes in the joint, which often occur before morphological changes can develop earlier than progressive morphological
changes appear (Fig. 2). This gives rise to the need for bio- changes are detectable. This information may be crucial for
chemical and quantitative MRI to reveal the early changes in early OA discovery and the efficacy of new treatment
the complex molecular composition of articular cartilage. options in OA.
After validation in in vitro studies and in studies on ani-
mal models,56 sodium MRI was employed for the evaluation 23
Na-MRI of the Spine
of OA patients. Wheaton et al performed a study using a sur- The intervertebral disk shows a characteristic architecture tai-
face coil at 4.0T and acquired sodium images with a UTE lored to its biomechanical purpose. The nucleus pulposus
radial sequence.57 These authors measured the knees of nine (NP) shows a high concentration of proteoglycans (PG) in
healthy volunteers without and three patients with symptoms the extracellular matrix that consists of large complexes of
of early OA. The mean sodium content measured in the bound GAGs. Those, in turn, have the essential ability to
patellae of the volunteers corresponded to a mean fixed charge passively store water due to their negatively charged side
density (FCD) measurement of –182 ± 9 mmol/L. These chains that attract free-floating, positively charged 23Na ions.
authors generated sodium maps for each OA patient, which Therefore, unlike most other tissues in the human body, car-
illustrated cartilage regions with significantly lower FCD tilage, in general, and the intervertebral disc (IVD), specifi-
(from –108 to –144 mmol/L) when compared to the FCD of cally, have the vast majority of 23Na ions present in the
healthy volunteers. extracellular volume.6
Zbyn et al very recently demonstrated that in vivo 23Na- The idea to detect early biochemical changes in the
MRI is a feasible method for the differentiation between low- intervertebral disc due to disk degeneration using 23Na-MRI
grade lesions and normal-appearing articular cartilage.58 Mor- was conceptualized and measured in vivo using a DQF, a
phological MRI at 3.0T and biochemical 23Na-MRI at 7.0T DQF-MA, as well as a TQF sequence by Ooms et al.59,60
investigations were performed at four timepoints (baseline, The group identified age-related changes in 23Na-MRI of the
1 week, 3 months, and 6 months). After every MR session, thoracic and lumbar intervertebral discs. First, the content
patients underwent the Knee Injury and Osteoarthritis Out- and the residual quadrupole interaction in the nucleus
come Score (KOOS) questionnaire for clinical evaluation. The pulposus and the annulus fibrous differed at different ages.
authors found significantly lower 23Na-cSI values in all lesions Also, the quadrupole coupling (ωQ) and the relaxation rate of
than in healthy cartilage tissue at all timepoints (all P ≤ 0.002). satellite transitions (1/T2f) increased with age.
KOOS scores improved in all subscales at the 3- and 6-month An ex vivo study published by Wang et al used bovine
visits, with a significant increase observed only in the quality of intervertebral discs and measured the PG content using
life subscale (P = 0.004). 1,9-dimethylmethylene blue (DMMB) assays of 28 nucleus

July 2021 63
Journal of Magnetic Resonance Imaging

pulposus punch samples.61 As already shown by Urban et al, as a marker for Achilles tendinopathy, as it provides informa-
a nucleus pulposus without measurable PG content still has a tion about elevated sodium concentration in the tendon.65
remaining 23Na concentration of about 111.54 mmol/L.62 They investigated eight patients with clinical findings of
Wang et al combined their measurements and the prior work chronic Achilles tendinopathy and scanned them with a 3D-
published by Urban et al62 and showed a correlation coeffi- gradient echo sequence using an in-plane resolution of
cient of 0.71 between the PG content measured by DMMB 0.89 mm and a total measurement time of about 32 minutes.
and the 23Na measured by sodium MRI. Simulating the rela- Mean bulk sodium SNR observed in this study was 4.9 ± 2.1
tionship only for a 23Na concentration from 150 mmol/L to (a.i.) in healthy control subjects and 9.3 ± 2.3 (a.i.) in
350 mmol/L, a linear correlation coefficient of 0.998 was patients with Achilles tendinopathy, and these means were
reached for sodium MRI, making it a very potent imaging statistically significant. Interestingly, these increases were not
biomarker for the indirect quantification of GAGs in the only local, suggesting that the tendon is completely affected
intervertebral disk.61 in Achilles tendinopathy. Also, the validation of the sodium
Noebauer-Huhmann et al published an in vivo study signal and sodium content in tendons was investigated in the
confirming that the concept of sodium imaging was a feasible, same study.65 Using 15 cadaver samples of Achilles tendons,
noninvasive imaging biomarker, and compared it to T2 map- which were analyzed with regard to the proteoglycan content,
ping and morphological grading.63 As degeneration is also the Pearson correlation coefficient between the sodium SI
associated with the diminishing synthesis of proteoglycans, and the proteoglycan content in dry weight was 0.71.
sodium imaging might be able to longitudinally depict tem- In another study, the increased sodium content in the
poral changes in different disc pathologies.63 However, those Achilles tendon, after fluoroquinolone antibiotic therapy, was
authors did not find a significant correlation of T2 values and investigated with 23Na-MRI.66 It has been previously
23
Na imaging. This was expected, since T2 mapping is mainly observed that, after fluoroquinolone treatment, patients can
related to water content, collagen fiber content, and organiza- develop Achilles tendinopathy, including all the typical symp-
tion, while sodium imaging correlates with GAG content. In toms, such as acute onset of tendon pain, tenderness, and
addition, it is known that natural degradation of the disc swelling that affects the function of the tendon and, in the
lowers the measured T2 relaxation times due to the accumula- worst case, may lead to tendon tear. Seven healthy male sub-
tion of lipids and adducts from carbohydrates, which does jects underwent voluntary ciprofloxacin treatment (1000 mg/
not affect the 23Na imaging measurements. This allows for day in two doses: 500 mg in the morning and 500 mg in the
the assumption that 23Na-MRI can distinguish between non- evening for 10 days) and were scanned with 23Na MRI at
degenerative, age-related pathological changes in the extracel- three timepoints: baseline; at 10 days; and at 5 months after
lular matrix of the nucleus pulposus, potentially making it the treatment. Here, the variable echo-time (vTE) sequence
more sensitive to pathophysiological changes than other adapted to x-nuclei capabilities was used with an echo of
quantitative MRI methods. 2.45 msec and a total acquisition time of 15 minutes. Despite
the fact that there were no morphologically detectable
changes in these volunteers after the treatment, the NMSI
Sodium in Tendons decreased by almost 25% (from 130 ± 8 to 98 ± 5 a.u.) and
Sodium MRI is a useful imaging modality for Achilles tendon then reached 116 ± 10 a.u. after 5 months. The results pro-
biochemical investigations. The Achilles tendon connects the vide evidence that sodium MRI is a technique robust and
calcaneus and the calf muscle and must withstand a remark- sensitive enough to detect the changes in the sodium content
able load during movement. To facilitate that, the highly in the Achilles tendon, which could be associated with altered
organized collagen matrix provides the base for static biome- proteoglycan content and may pose the tendon to risk for
chanical properties with the aid of proteoglycans (3.5% of tendinopathy and tear.
dry weight) responsible for dynamic biomechanical properties.
In tendinopathy, due to stimulated proteosynthesis, which
helps to overcome disaggregation of the microfibrillar bun- Muscle
dles, increased proteoglycan concentration has been observed Recent studies demonstrated that 23Na-MRI of a lower leg
using biochemical assays.64 As negatively charged sulfate and muscle can be a useful imaging modality for the detection of
carboxyl groups of proteoglycans attract positively charged sodium content changes during exercise or disease.67,68 The
sodium ions, there is a direct proportion of sodium and pro- sodium concentration increase or decrease in the body mani-
teoglycan concentration in the Achilles tendon. To acquire a fests as a condition known as hypernatremia or hyponatremia,
23
Na signal from the tendon, typically no dedicated coils are which may be induced by a hormonal imbalance or related
used, but sodium knee coils are used instead—they provide diseases, such as diabetes mellitus, hypertension, and acute
enough space for lower leg placement and satisfactory coil heart failure. 23Na concentrations alterations in calf muscle,
loading. Juras et al showed that the sodium signal can serve however, may be closely related to developed pathologies of

64 Volume 54, No. 1

 Zaric et al.: Frontiers of Sodium MRI Revisited

the muscle tissue, such as channelopathy and muscular dys- inflammation.73 Gerhalter et al studied 13 patients with
trophy. Several studies confirmed the reproducibility and DMD and found that TSC (26.0 ± 1.3 mM, P < 0.05) and
repeatability of quantitative sodium imaging in the lower leg ICS (0.69 ± 0.05 a.u., P < 0.05) were elevated in DMD com-
muscle, which is considered a basic condition for an accurate pared to healthy controls (16.5 ± 1.3 mM and 0.47 ± 0.04 a.
evaluation of several different muscle pathologies in u.). The ICS was frequently abnormal in DMD compared to
patients.69,70 heathy controls, and was present even in the absence of fatty
degenerative changes and water T2 increases73 (Fig. 3). One
Sodium Evaluations in Diabetic Patients explanation for the observed increased ICS and TSC values
Na+/K+-ATPase activity is decreased in patients suffering despite normal water T2 values might be that sodium MRI is
from diabetes. A sufficiently high concentration of hormone more sensitive in the detection of dystrophic changes. Since
insulin in the blood will directly enhance sodium/potassium there were no significant changes in the ICS/TSC ratio
pump activity in muscle, kidney, liver, etc.71 Chang et al between the healthy and dystrophic muscle, this observation
evaluated the signal intensity (SI) of sodium pre- and post- could be compatible with two concurrent phenomena: an
exercise. The sodium signal was assessed in patients with dia- increase in the intracellular sodium and an increase of the
betes and in healthy subjects, in all three compartments of extracellular volume fraction.
the calf muscle: tibialis anterior (TA), soleus (S), and the gas-
trocnemius (G) muscles. It was found that the sodium signal Hypertension
intensity in the S and G immediately increased significantly Quantitative 23Na-MRI in muscle may be used to assess an
after exercise for both diabetic patients and healthy subjects. increase in body sodium content due to high blood pressure
However, the signal intensity decrease to baseline was slower (HBP). Kopp et al found a 29% increase in 23Na content in
in diabetics. An explanation for this can be supported by the patients with aldosteronism associated with hypertension
fact that, in patients with diabetes mellitus, muscle function compared to healthy subjects.74,75 Although the authors
is reduced and accompanied by impaired functioning of the stated that the achieved resolution (3 × 3 × 30) mm3 was
Na+ /K+ pump. adequate, they considered it as unsatisfactory for the skin
sodium concentration evaluations. However, the authors’ aim
Muscular Channelopathy and Muscular Dystrophy was to test the basic hypothesis about Na+ storage in muscle,
Muscular channelopathies are a group of nondystrophic and particularly, in skin, without apparent accompanying
myopathies, which are caused by gene mutations that result fluid retention. Linz et al developed a dedicated, two-channel
in a malfunction of the ionic channels of the muscle. Nagel transceiver RF coil array for skin measurements and per-
et al, as a clinical model, chose patients with hypokalemic formed sodium imaging in humans with a gradient echo
periodic paralysis (hypoPP) and paramyotonia congenita sequence and 0.9 × 0.9 × 30 mm3 resolution at 7.0T.76 Skin
(PC).72 The purpose of the study was to investigate which of sodium concentration results showed a discrepancy between
three different sequences: 23Na-TSC; T1-weighted sodium the Na+ content obtained in vivo at 3.0T vs. 7.0T. The
imaging (23Na-T1); or 23Na-IR would provide the strongest inconsistency could be the result of the higher spatial resolu-
weighting toward intracellular sodium. All three sequences tion at 7.0T, which may compensate for the PVE that
demonstrated significantly higher signal intensities in hypoPP occurred at the lower field strength. Submillimeter image res-
compared to those in PC patients and healthy subjects. How- olution, achievable on ultrahigh MR systems, should provide
ever, after inducing a provocation in PC patients, a significant a better understanding of the physiological processes related
(P = 0.007) increase (>20%) in the muscular 23Na-IR signal to HBP in the future.
and a corresponding decrease of muscle strength was detected.
These results provide strong evidence that 23Na-IR substan-
tially advances weighting toward intracellular sodium and Cardiac 23Na MRI
enables an improved evaluation of pathophysiological changes Technical Considerations
of muscles in patients suffering from rare diseases. Sodium is an essential ion for myocyte function and integrity.
In contrast to skeletal muscle channelopathies, which It plays an important role in balancing osmotic pressure
are rare, inherited childhood-onset disorders, myotonic dys- through the sodium–potassium pump. The concentration of
trophy (DM) is the most common form of muscular dystro- extracellular sodium is 10 times higher than that of the
phy that begins in adulthood. A special severe type of DM, intracellular concentration to provide proper muscle excita-
investigated with 23Na-MRI, is Duchenne’s muscular dystro- tion, conduction, and contraction. The pathological change
phy (DMD) that is usually diagnosed in childhood. DMD is in TSC and the impairment of sodium flux often occurs in
caused by a mutation in the dystrophin gene, which leads to hypertrophic myopathies, ischemic cardiac diseases, and
progressive muscle weakness and destruction and is associated infarction. Hence, 23Na-MRI provides a valuable and power-
with ion homeostasis dysregulation and chronic ful imaging tool for the detection of sodium concentration in

July 2021 65
Journal of Magnetic Resonance Imaging

FIGURE 3: Illustration of (a) fat fraction maps, (b) water T2 maps, (c) DW 23Na images, and (d) IR 23Na images in the leg of 7-year-old,
9-year-old, and 11-year-old DMD patients, as well as a 10-year-old control. For 23Na MRI, four reference tubes were used (1, 40 mM
NaCl; 2, 40 mM NaCl, and 5% agarose gel; 3, 20 mM NaCl; 4, 20 mM NaCl, and 5% agarose gel). The leg muscles of DMD patients
showed generally elevated FF, water T2, and sodium signals compared with age-matched controls. While the 7-year-old DMD
patient exhibited slightly elevated FF (mean FF 0.08 ± 0.04), the FF was much higher in the muscles of the 11-year-old DMD patient
(mean FF 0.19 ± 0.1), who was not able to walk more than 10 m without human assistance. Increased total sodium and intracellular-
weighted sodium signals are also visible in the dystrophic muscle tissue with normal T2 and FF values (figure reproduced from Ref.
73 with permission from Wiley).

heart muscle in vivo. However, similar to other sodium MR signal in heart tissue usually exhibits a bicomponent decay,
applications, cardiac sodium MR faces many issues related to with an 40% fast component (T2f, ranging from 0.5–-
physical and technical obstacles. An article published by Bot- 4 msec) and a 60% slow component (T2s, ranging from
tomley provided a critical review of the properties, methods, 12–32 msec).78–80 With conventional MR techniques using
and potential clinical applications of 23Na-MRI in the human an echo time (TE) longer than 3 msec, the majority of T2f is
heart.7 lost; however, using short- and ultrashort MR sequences, the
Cardiac 23Na-MRI studies may be performed on MRI TSC can be measured quite precisely and the values match
scanners from 1.5–7.0T with a spatial resolution in the range the assay.78
of 100–1000 mm3, typically 160 mm3 at 3.0T in a For clinical applications, it would be desirable to sepa-
10-minute measurement time.7,77 The relatively short T1 rate intracellular sodium, typically 10–15 mM and extracellu-
relaxation in heart muscle allows for further scan time short- lar sodium, typically 135–150 mM. The theory that
ening by increasing the number of signal averages. The 23Na attributes intra- and extracellular sodium to T2f and T2s is

66 Volume 54, No. 1

 Zaric et al.: Frontiers of Sodium MRI Revisited

controversially discussed in the literature, as there is a lack of high-resolution sodium images of the mouse heart.93 The
solid evidence to prove this theory. There is even some evi- sodium SNR in the left ventricle, the right ventricle, the left
dence against this theory, suggesting that, from an NMR ventricular free wall, and the septum was found to be
point of view, the 23Na ion is similar inside the cell and out- 9.2 ± 2.3, 8.3 ± 2.06, 3.8 ± 1.0, and 5.6 ± 1.4, respectively. In
side the cell.81 The sodium signal is a relative value deter- another study, sodium imaging and spectroscopy was investi-
mined by a combination of many factors, such as T1, T2, B1,, gated for use as a potential marker with which to assess viabil-
and B0 field homogeneity, and is linearly correlated with ity after low-flow ischemia, using a rat heart model.92
sodium concentration in heart muscle tissue. It is also possi- Intracellular sodium image intensity increased significantly dur-
ble to measure absolute TSC in the heart using reference ing ischemia of the left side, whereas that of the right side
tubes with a known sodium concentration. These reference remained unchanged; however, total sodium image intensity
tubes are scanned either separately with the same protocol or remained unchanged in both sides of the heart.
in the field of view of an actual 23 Na-MRI measurement of 23
Na-MRI offers a tool with which to study myocardial
the heart, taking into account the coil sensitivity. There have ion homeostasis in vivo and can be used for different areas of
also been attempts to use an internal reference that does not cardiovascular disease. Most of the clinical studies are related
change between subjects, eg, myocardial and ventricular to the investigation of sodium levels in myocardial infarction
blood ratio; however, a direct relation to TSC has not, as yet, (MI). In a feasibility study by Standstede et al, the elevated
been validated.82 To acquire the best possible 23Na signal, sodium signal was observed in the myocardium affected by
dedicated transmit/receive resonators should be used. Some of acute infarction (60.6 ± 21.6) compared to healthy controls
these are commercially available,83,84 but many research (37.2 ± 12.8); however, no statistically significant increase
groups rely on custom-made, dedicated cardiac coils.7,64,85 was measured in infarcted myocardium in the subacute and
Boehmert et al built a dedicated 4/4 channel 1H/23Na coil the chronic groups.95 To acquire the sodium signal from the
with high B1+ homogeneity and penetration depth to investi- heart, an ECG-triggered, 3D, spoiled gradient-echo (fast low-
gate the cardiorenal syndrome.86 The complexity of the dedi- angle shot) sequence with a TE of 3.1 msec was used. In a
cated resonators grows with the field strength. Typically, a follow-up study, the same authors monitored 12 patients on
surface coil design is preferred, with the diameter approxi- days 4, 14, and 90 after infarction and found an increase in
mately equal to the depth of interest, which is 10–15 cm. the sodium signal of 39 ± 18, 31 ± 17, and 28 ± 13 [%],
Another option for increasing 23Na SNR is phased array sur- respectively, suggesting that 23Na-MRI may be an applicable
face coils; however, multichannel 23Na-MRI is required, method for imaging nonviable myocardium in vivo.84 Abso-
which is available on only a small number of MR scanners lute TSC was measured in 20 patients with a history of prior
installed worldwide.85, 87 MI using a 150 mmol/L Na+ concentration reference and a
To acquire the maximum sodium signal from the heart coregistration to 1H images acquired with a contrast agent.90
and allow for TSC quantification, MR sequences with short Although the TSC was elevated by 30% in patients com-
TEs are necessary.64,77,79 Typically, 3D projection recon- pared to healthy controls, TSC does not appear to be linked
struction techniques are used, with a hard excitation pulse to infarct age or size or to global ventricular function. Quanti-
followed by a constant gradient. To acquire an optimal SNR, tative sodium MRI is heavily influenced by cardiac and respi-
the projection increment is 111 , referred to as the “golden ratory motion, PVEs, and inhomogeneity of the static
ratio.”79,88 Other possible trajectories are twisted projection magnetic field, as well as of the transmit and receive field.
imaging16 and spiral imaging.89 As hard pulses are very sensi- Lott et al showed that a thorough correction for these influ-
tive to B1 inhomogeneity, adiabatic pulses are alternatively ences provides more accurate TSC results from the heart,
used to provide a B1-independent excitation. Cardiac motion albeit somewhat lower compared to uncorrected values
is also a pitfall for 23Na-MRI sequences, and electrocardiogra- (Fig. 4).96 Christa et al measured increased myocardial
phy (ECG) gating, as often used in 1H cardiac imaging, is sodium SI in Conn’s syndrome, which is manifested by
not helpful because the advantage of the short T1 of 23Na hyperaldosteronemia that results in an alteration of sodium
would be lost. Therefore, either only motion-quiet frames are and potassium levels.97 They used a 3D gradient echo
acquired84,90 or retrospective ECG gating is used.90,91 sequence with a TR of 100 msec; a TE of 2.01 msec; flip
angle 90 ; field of view (FOV) 500 × 500 × 200 mm3;
Preclinical and Clinical Studies which provided an acquisition matrix of 128 × 128 × 10,
Animal models provide an important step toward understand- with a resolution of 3.9 × 3.9 × 20 mm3; and eight signal
ing the physiology and pathology of sodium in the heart, averages; resulting in a total acquisition time of 17 minutes.
although the small structures present considerable technical Patients with Conn’s syndrome exhibited significantly higher
challenges.78,92–94 However, it is possible to compensate for relative sodium signal intensities in the myocardium com-
the lower SNR by using small-bore, ultrahigh-field MR scan- pared to healthy controls (0.31, ranging from 0.26–0.34)
ners. Neuberger et al used a 17.6T animal scanner to acquire vs. 0.24 (ranging from 0.20–0.27). The results suggest that

July 2021 67
Journal of Magnetic Resonance Imaging

FIGURE 4: An example of proton and sodium images with segmentation masks used for quantification and postprocessing. (a) 1H
MRI image (with navigator stripes) before registration. (b) 23Na MRI with corresponding masks (red = blood mask,
green = myocardial mask) based on the 1H image. (c) Simulated 23Na MRI of the heart based on 1H masks. (d) 1H MRI image after
registration. (e) Registered 1H image with cardiac 23Na MRI as an overlay. (f) Cardiac 23Na MRI (figure adapted from Ref. 96 and
reproduced with permission from Wiley).

the myocardium is, along with skeletal muscle and skin, from the renal cortex in the direction of the medullary pyr-
another possible sodium storage site, and 23Na MRI can be amid. This concentration change is called the cor-
used to monitor patients who are undergoing treatment for ticomedullary 23Na gradient and it is highly important for
an aldosterone excess. proper kidney function evaluation and for the detection of
abnormalities.
23
Na -MRI in Kidney Function Evaluation
The most important role of the kidney is to maintain the Breast Tumors
overall fluid balance in the body. Renal function is deter- Ultrasound (US) and mammography are the most-used imag-
mined by a proper regulation of extracellular sodium in the ing modalities for breast lesion detection and characterization.
kidney, established by a sodium concentration gradient from However, the diagnostic accuracy of breast sonography exami-
the cortex to the medulla. The malfunction of the sodium nations is low, and thus, US is usually recommended as a
concentration gradient may be caused by several different complementary technique to mammography or other imaging
renal diseases, such as nephropathy, acute kidney failure, or techniques. Mammography, however, is primarily advised for
irregular kidney function after transplantation. patients over 50 with less dense breast tissue. Low sensitivity
A study by Haneder et al demonstrated the feasibility and specificity, as well as accompanying ionizing radiation,
of in vivo 23Na imaging of human kidneys on a whole- are strong limitations of this imaging method. The develop-
body, ultrahigh-field MR system at 7.0T. Heathy subjects ment of noninvasive biochemical MRI techniques is, there-
were examined using a 3D Cartesian spoiled gradient echo fore, necessary, especially those that can define predictive
sequence with a variable echo time scheme with a nominal biomarkers for breast cancer diagnosis and characterization
in-plane resolution of 4 × 4 mm2 and a slice thickness of with high sensitivity and specificity. One of those techniques
5 mm.98 These authors confirmed the validity of the con- that could provide supplementary information to conven-
cept of increasing the 23Na SNR or 23Na concentration tional MRI, such as contrast-based imaging techniques, is

68 Volume 54, No. 1

 Zaric et al.: Frontiers of Sodium MRI Revisited

reconstruction could utilize a conventional total variation

(TV) denoising technique, it was shown that not only arti-
facts and noise are diminished, but also small-tissue struc-
tures. As a possible solution, the authors proposed a method
that adapts the TV using anatomical weighting factors, which
represent known tissue boundaries (AnaWeTV). After inclu-
sion of prior information from proton (1H) MRI into the CS
reconstruction, image quality was further increased due to
preserved tissue boundaries and PV effects. Since proton and
sodium images are highly correlated, these weighting factors
could be obtained from a high-resolution 1H MR image and
then transferred to 23Na images. Applied like this, CS recon-
struction simultaneously maintained known tissue boundaries
and reduced image artifacts. The method still has to be
applied and validated in patients with breast lesions.

FIGURE 5: A 60-year-old female patient with invasive ductal Brain

carcinoma (IDC) and a grade 3 proliferation (G3) in the lateral The first clinical NMR images of cerebral sodium distribution
part of the left breast. (a) DIXON water image shows a
heterogeneous lesion with irregular margins typical of malignant in normal volunteers and in patients with a variety of patho-
tumors, and (b) a corresponding color-coded bilateral 23Na logical lesions were shown by Hilal et al in 1985.101 Subse-
image corrected for coil sensitivity and obtained with a 3D radial quently, many different studies explored the potential of
projection sequence (DA-3DPR).
using sodium MRI as a noninvasive imaging modality for
biochemical investigations of brain tumors, stroke, and neuro-
23
Na-MRI, which is sensitive to sodium concentration logical disorders. The potential of a combination of sodium
changes in tissue as a reliable biomarker for cell viability and imaging with other imaging modalities, such as positron
function. Thus, 23Na-MRI may increase overall diagnostic emission tomography (PET) in metabolic change studies con-
accuracy and contribute to other established imaging methods nected with human brain pathologies, have been discussed in
for the monitoring of treatment efficacy. The technique has a review article by Shah et al.9 High reproducibility and
been shown to be feasible for the differentiation between repeatability of cerebral 23Na-MRI was recently reported by
malignant and healthy breast tissue and may prevent false- Meyer et al.102
positive and false-negative findings in patients at high risk for
malignancy (Fig. 5). Zaric et al recently showed that the Brain Tumors
increased field strength provides the increased sensitivity nec- Two main histological properties of malignant tumors are
essary to achieve an acceptable spatial resolution for the meta- increased angiogenesis and cell proliferation. Unregulated
bolic interpretation of the tissue under investigation.99 These Na+/H+ exchange kinetics and altered Na+/K+-ATPase activ-
authors showed that the low 23Na content in healthy glandu- ity may increase the number of cells, which then will lead to
lar breast tissue can be assessed using improved imaging tech- tumor genesis and growth. The amount of sodium inside the
niques and hardware at 7.0T. The quantitative sodium MRI cell will rise and may be considered a biomarker of tumor
was performed using an AWSOS sequence. The measurement malignancy.103 One of the first 23Na-MRI studies included
protocol was optimized, and sodium images with a 1.5 mm 20 patients with brain gliomas, which reported elevated TSC
in-plane resolution were acquired in 16 minutes. It was levels for both tumors and surrounding tissues.104,105 How-
shown that TSC in carcinomas is 28% and 49% higher com- ever, the changes in TSC in a tissue of interest may provide
pared to benign tumors and healthy glandular tissue, respec- low specificity information on the origin of sodium signal
tively. Lachner et al proposed an advanced CS reconstruction changes and their connection to tumor malignancy.106 Intra-
algorithm for radial 23Na multicoil data, applied to simulated cellular sodium signal that originates from Na+ ions with
and measured breast datasets.100 For data acquisition, a DA- restricted mobility may be distinguished from extracellular
3DPR pulse sequence, which enables more efficient k-space sodium signal utilizing TQF sequences. There is currently a
sampling, can be used, as well as multicoil arrays that can fur- controversial discussion within the sodium MRI community
ther enhance the SNR. For the first time, CS was used for regarding the last statement. According to Burstein et al, the
undersampled breast sodium data reconstruction. It was dem- intracellular sodium signal cannot be separated from extracel-
onstrated that CS can be successfully implemented to reduce lular sodium in human tissue.29 The authors explain that,
acquisition time and enhance image quality. Although CS based on chemical kinetics principles, sodium ions in

July 2021 69
Journal of Magnetic Resonance Imaging

biological systems are not in a “bound” or a “mobile” state Sodium MRI in Neurological Disorders
due to the fast relaxation rate constants they have. Multiple sclerosis (MS) is an inflammatory demyelinating dis-
To date, there is a limited number of publications that ease that causes the development of focal and diffuse lesions
have dealt with an application of TQF 23Na imaging in in white matter (WM) and gray matter (GM). Chronically
patients with brain tumors. Initial reports have suggested a demyelinated MS lesions are accompanied by a substantially
promising role for TQF 23Na-MRI in discriminating vital reduced axonal Na+/K+-ATPase expression.113 In vivo MRI
parts of tumor, with high cell proliferation from areas of studies using 23Na imaging showed increased brain TSC in
tumor necrosis.24,107 One method allows simultaneous acqui- patients with MS. Inglese et al performed a sodium MRI
sition of TSC-weighted, as well as TQF images (SISTINA), study at 3.0T, including patients with advanced relapsing–
utilizing a sequence that interleaves an ultrashort TE, radial remitting MS, and applying a 3D radial gradient-echo UTE
projection readout into the three-pulse, triple-quantum prepa- sequence.114 These authors quantified absolute TSC in a
ration, and was developed by Fiege et al.107 A drawback of patient’s lesions and several brain regions with normal-
the SISTINA method was limited SNR and resolution; tumor appearing white and gray matter (NAWM and NAGM). The
images could not be analyzed without a proton reference same measurement was performed in corresponding areas in
image. Significant improvements of this technical issue can be controls. In MS patients, TSCs were higher in MS lesions
achieved with improved readouts, such as DA-3DPR or TPI compared to areas of NAWM. Also, TSC measured in
sequences. Based on the DA-3DPR sequence, Nagel et al ear- NAWM were also significantly higher than those in
lier proposed a relaxation-weighted (RW) 23Na-MRI and corresponding WM regions in healthy controls. Further stud-
applied it in patients with brain carcinomas. The results ies confirmed these findings.115–118 In order to obtain more
showed that RW 23Na imaging allows excellent differentia- specific information, Petracca et al used a method that com-
tion between grade I–III and grade IV gliomas.28 Neto et al bined SQ and TQF MRI, to quantify TSC and the intracel-
recently performed a study with eight brain carcinoma lular sodium molar fraction (ISMF) and then derived ISC
patients before surgical, chemo, or radiation therapy treat- and ISVF, an indirect measure of ESC. The results generated
ment. Patients were scanned at 3.0T using a custom-made from 19 relapsing–remitting MS patients and 17 heathy con-
double-tuned (23Na/1H) head coil and a FLORET sequence, trols showed that global TSC and ISC evaluated in GM were
with and without fluid suppression by inversion recovery higher, while GM and WM ISVF (indirect measure of extra-
(IR). The authors generated maps of pseudo-intracellular cellular sodium concentration) were lower in patients com-
sodium concentration (C1), pseudo-extracellular volume frac- pared with healthy controls.119 In 11 patients with acute MS
tion (α2), apparent intracellular sodium concentration (aISC), lesions, 1H and 23Na-MRI examinations were performed at
and apparent total sodium concentration (aTSC). These 3.0T.120 Initial examinations showed that contrast-enhancing
parameters were significantly elevated in the normal- lesions had high TSC, while, 4 weeks later, MRI TSC in
appearing putamen compared to NAWM. Analysis of all solid these lesions was reduced. The authors of the study con-
tumors demonstrated a significant increase of aTSC and α2, cluded that tissue structure is relatively preserved in the early
and a significant decrease of aISC when compared with stage of lesion progression. Stobbe et al investigated potential
NAWM.108 In a study published by Biller et al, 34 patients errors in TSC quantification in patients with primary progres-
with brain tumors were examined with 23Na-MRI, performed sive MS disease, often characterized by small lesions.121 The
using a 7.0T MR system and a double-tuned (1H/23Na) quad- authors found that signal from volumes-of-interest (VOIs) in
rature birdcage head coil, and the DA-3DPR technique.109 large spheres (10 cm3) was 20% higher than expected. In
The results demonstrated that the initial sodium signal, mea- smaller VOIs (0.35 cm3), the 23Na signal was even more
sured in a brain tumor patient without any previous treatment, underestimated (40–60%). This may be one of the most criti-
is a valuable predictor of isocitrate dehydrogenase (IDH) muta- cal limitations of low-resolution quantitative methods applied
tion status and tumor progression. The study confirmed a great in evaluation of small lesions. A case–control study that
potential for 23Na-MRI for an improved and individualized included patients who suffered migraines was recently publi-
approach in neuro-oncology. shed by Meyer et al.122 The results showed a significantly
higher sodium concentration in cerebrospinal fluid (CSF) in
migraine and tension-type headache (TTH) patients compared
Ischemic Stroke with healthy controls (P = 0.007, P < 0.001, respectively).
Changes in intracellular 23Na concentration are known to Alzheimer’s disease (AD) is a chronic neurodegenerative disease
occur shortly after an ischemic insult due to impaired func- accompanied by alterations of the sodium levels in brain due
tioning of Na+/K+-ATPase, which is responsible for human to cell death or loss of functional characteristics. The patho-
homeostasis. The role of 23Na-MRI in identifying patients physiological changes in a brain affected by AD can be mea-
suitable for therapeutic intervention immediately after stroke sured with sodium MRI and may provide valuable additional
may be crucial (Fig. 6).16,110–112 information for early discovery of the disease. A small study of

70 Volume 54, No. 1

 Zaric et al.: Frontiers of Sodium MRI Revisited

FIGURE 6: Representative sodium (23Na), diffusion-weighted (DWI), T2-weighted, and fluid-attenuated inversion recovery (FLAIR)
images from patients 7 to 52 hours after ischemic stroke onset. Areas of sodium signal intensity increase correspond to the lesions
identified on DWI. The sodium signal intensity in the areas of ischemia qualitatively increased with time (figure reproduced from Ref.
112 with permission from Wiley).

AD patients (n = 5) reported a 7.5% brain TSC increase with Conclusion

an inverse correlation to the volume of the hippocampus.123 The great potential of 23Na-MRI has been extensively dem-
Another possible application for 23Na-MRI is in Huntington’s onstrated. Several hundred publications have presented meth-
disease diagnosis. In patients with this neurological condition, odological and feasibility studies showing that 23Na-MRI can
the highest TSC was found in the caudate nucleus, which cor- serve as a reliable imaging tool for the diagnosis and treat-
related with GM atrophy.124 In a group of healthy subjects, ment monitoring of many diseases. The future task of the sci-
neuroglial-vascular mechanisms were studied by dynamic entific community is to continue to work on technical
sodium imaging using a triple-echo, 3D-center-out radial improvements of the technique to enable 23Na-MRI to
sequence at 7.0T. These authors demonstrated an activation in become an approved imaging modality for biochemical clini-
the left central regions, the supplementary motor areas, and the cal investigations and to provide a better understanding of
left cerebellum, manifested as an increase of the sodium signal different pathophysiological conditions.
at an ultrashort TE and a decrease of the signal at a long
TE.125 The presumption of the existence of “restricted” and
“mobile” sodium ions and differences in the T2 relaxation Acknowledgments
times of sodium nuclei in the intra- and extracellular space is, This study was funded by the Vienna Science and Technol-
however, still controversial.29 ogy Fund (WWTF; Project LS14-096), the Austrian Science

July 2021 71
Journal of Magnetic Resonance Imaging

Fund (FWF, Project: KLIF 541-B30), and VEGA2/0003/20 22. Tsang A, Stobbe RW, Beaulieu C. In vivo double quantum filtered
sodium magnetic resonance imaging of human brain. Magn Reson
of the Scientific Grant Agency of the Ministry of Education, Med 2015;73(2):497-504.
Science, Research, and Sports of the Slovak Republic and the 23. Gast LV, Gerhalter T, Hensel B, Uder M, Nagel AM. Double quantum
Slovak Academy of Sciences. filtered (23) Na MRI with magic angle excitation of human skeletal
muscle in the presence of B0 and B1 inhomogeneities. NMR Biomed
2018;31(12):e4010.
References 24. Hancu I, Boada FE, Shen GX. Three-dimensional triple-quantum-
1. Clausen MJ, Poulsen H. Sodium/potassium homeostasis in the cell. filtered (23)Na imaging of in vivo human brain. Magn Reson Med
Met Ions Life Sci 2013;12:41-67. 1999;42(6):1146-1154.
2. McCarthy JV, Cotter TG. Cell shrinkage and apoptosis: A role for 25. Tsang A, Stobbe RW, Beaulieu C. Triple-quantum-filtered sodium
potassium and sodium ion efflux. Cell Death Differ 1997;4(8):756-770. imaging of the human brain at 4.7 T. Magn Reson Med 2012;67(6):
1633-1643.
3. Bottomley PA. Sodium MRI in man: Technique and findings. eMagRes
2012;1:353-366. 26. Benkhedah N, Bachert P, Semmler W, Nagel AM. Three-dimensional
biexponential weighted (23)Na imaging of the human brain with
4. Pohmann R, Speck O, Scheffler K. Signal-to-noise ratio and MR tissue
higher SNR and shorter acquisition time. Magn Reson Med 2013;70
parameters in human brain imaging at 3, 7, and 9.4 Tesla using cur-
(3):754-765.
rent receive coil arrays. Magn Reson Med 2016;75(2):801-809.

5. Ugurbil K. Magnetic resonance imaging at ultrahigh fields. IEEE Trans 27. Stobbe R, Beaulieu C. In vivo sodium magnetic resonance imaging of
Biomed Eng 2014;61(5):1364-1379. the human brain using soft inversion recovery fluid attenuation. Magn
Reson Med 2005;54(5):1305-1310.
6. Madelin G, Regatte RR. Biomedical applications of sodium MRI
in vivo. J Magn Reson Imaging 2013;38(3):511-529. 28. Nagel AM, Bock M, Hartmann C, et al. The potential of relaxation-
weighted sodium magnetic resonance imaging as demonstrated on
7. Bottomley PA. Sodium MRI in human heart: A review. NMR Biomed brain tumors. Invest Radiol 2011;46(9):539-547.
2016;29(2):187-196.
29. Burstein D, Springer CS Jr. Sodium MRI revisited. Magn Reson Med
8. Zbyn S, Mlynárik V, Juras V, Szomolanyi P, Trattnig S. Evaluation of 2019;82(2):521-524.
cartilage repair and osteoarthritis with sodium MRI. NMR Biomed
2016;29(2):206-215. 30. Qian Y, Stenger VA, Boada FE. Parallel imaging with 3D TPI trajectory:
SNR and acceleration benefits. Magn Reson Imaging 2009;27(5):656-663.
9. Shah NJ, Worthoff WA, Langen KJ. Imaging of sodium in the brain: A
brief review. NMR Biomed 2016;29(2):162-174. 31. Benkhedah N, Hoffmann SH, Biller A, Nagel AM. Evaluation of adap-
tive combination of 30-channel head receive coil array data in 23Na
10. Wiggins GC, Brown R, Lakshmanan K. High-performance radio- MR imaging. Magn Reson Med 2016;75(2):527-536.
frequency coils for (23)Na MRI: Brain and musculoskeletal applica-
tions. NMR Biomed 2016;29(2):96-106. 32. Gnahm C, Nagel AM. Anatomically weighted second-order total varia-
tion reconstruction of 23Na MRI using prior information from 1H MRI.
11. Bangerter NK, Kaggie JD, Taylor MD, Hadley JR. Sodium MRI radio- Neuroimage 2015;105:452-461.
frequency coils for body imaging. NMR Biomed 2016;29(2):107-118.
33. Behl NG, Gnahm C, Bachert P, Ladd ME, Nagel AM. Three-
12. Thulborn KR. Quantitative sodium MR imaging: A review of its evolv- dimensional dictionary-learning reconstruction of (23)Na MRI data.
ing role in medicine. Neuroimage 2018;168:250-268. Magn Reson Med 2016;75(4):1605-1616.
13. Hilal SK, Oh CH, Mun IK, Silver AJ. Sodium imaging. In: Stark DD, 34. Utzschneider M, Behl NGR, Lachne S, et al. [Formula: see text] Accel-
Bradley WG, editors. Magnetic resonance imaging. St Louis: Mosby; erated quantification of tissue sodium concentration in skeletal muscle
1992. p 1091-1110. tissue: Quantitative capability of dictionary learning compressed sens-
14. Nagel AM, Laun FB, Weber MA, Matthies C, Semmler W, Schad LR. ing. Magnetic Resonance Materials in Physics, Biology and Medicine
Sodium MRI using a density-adapted 3D radial acquisition technique. 2020;33:495-505. epub ahead of print.
Magn Reson Med 2009;62(6):1565-1573. 35. Bangerter NK, Tarbox GJ, Taylor MD, Kaggie JD. Quantitative sodium
15. Boada FE, Gillen JS, Shen GX, Chang SY, Thulborn KR. Fast three magnetic resonance imaging of cartilage, muscle, and tendon. Quant
dimensional sodium imaging. Magn Reson Med 1997;37(5):706-715. Imaging Med Surg 2016;6(6):699-714.

16. Lu AM, Atkinson IC, Claiborne TC, Damen FC, Thulborn KR. Quantita- 36. Lommen J, Konstandin S, Krämer P, Schad LR. Enhancing the quantifi-
tive sodium imaging with a flexible twisted projection pulse sequence. cation of tissue sodium content by MRI: Time-efficient sodium B1
Magn Reson Med 2010;63(6):1583-1593. mapping at clinical field strengths. NMR Biomed 2016;29(2):129-136.

17. Romanzetti S, Mirkes CC, Fiege DP, Celik A, Felder J, Shah NJ. Map- 37. Niesporek SC, Hoffmann SH, Berger MC, et al. Partial volume correc-
ping tissue sodium concentration in the human brain: A comparison tion for in vivo (23)Na-MRI data of the human brain. Neuroimage
of MR sequences at 9.4Tesla. Neuroimage 2014;96:44-53. 2015;112:353-363.

18. Qian Y, Boada FE. Acquisition-weighted stack of spirals for fast high- 38. Gast LV, Henning A, Hensel B, Uder M, Nagel AM. Localized B0 shim-
resolution three-dimensional ultra-short echo time MR imaging. Magn ming based on (23) Na MRI at 7 T. Magn Reson Med 2020;83(4):
Reson Med 2008;60(1):135-145. 1339-1347.

19. Romanzetti S, Halse M, Kaffanke J, Zilles K, Balcom BJ, Shah NJ. A 39. Brown R, Madelin G, Lattanzi R, et al. Design of a nested eight-
comparison of three SPRITE techniques for the quantitative 3D imag- channel sodium and four-channel proton coil for 7T knee imaging.
ing of the 23Na spin density on a 4T whole-body machine. J Magn Magn Reson Med 2013;70(1):259-268.
Reson 2006;179(1):64-72.
40. Brown R, Lakshmanan K, Madelin G, et al. A flexible nested sodium
20. Pipe JG, Zwart NR, Aboussouan EA, Robison RK, Devaraj A, and proton coil array with wideband matching for knee cartilage MRI
Johnson KO. A new design and rationale for 3D orthogonally over- at 3T. Magn Reson Med 2016;76(4):1325-1334.
sampled k-space trajectories. Magn Reson Med 2011;66(5):1303-
41. Ianniello, C., G Madelin, K Lakshmanan, B Zhang, L Moy, R Brown,
1311.
Design and performance of a dual tuned 7 T proton/sodium breast
21. Pekar J, Renshaw PF, Leigh JS. Selective detection of intracellular coil. In: 2017 International Conference on Electromagnetics in
sodium by coherence-transfer NMR. J Magn Reson 1987;72: Advanced Applications (ICEAA). Verona, Italy: Institute of Electrical
159-161. and Electronics Engineers (IEEE); 2017:1890-1893.

72 Volume 54, No. 1

 Zaric et al.: Frontiers of Sodium MRI Revisited

42. Ianniello C, Madelin G, Moy L, Brown R. A dual-tuned multichannel 61. Wang C, McArdle E, Fenty M, et al. Validation of sodium magnetic
bilateral RF coil for (1) H/(23) Na breast MRI at 7 T. Magn Reson Med resonance imaging of intervertebral disc. Spine (Phila Pa 1976) 2010;
2019;82(4):1566-1575. 35(5):505-510.

43. Lakshmanan K, Brown R, Madelin G, Qian Y, Boada F, Wiggins GC. 62. Urban JP, Winlove CP. Pathophysiology of the intervertebral disc and
An eight-channel sodium/proton coil for brain MRI at 3 T. NMR the challenges for MRI. J Magn Reson Imaging 2007;25(2):419-432.
Biomed 2018;31(2):1-9.
63. Noebauer-Huhmann IM, Juras V, Pfirrmann CWA, et al. Sodium MR
44. Wetterling F, Corteville DM, Kalayciyan R, et al. Whole body sodium imaging of the lumbar intervertebral disk at 7 T: Correlation with T2
MRI at 3T using an asymmetric birdcage resonator and short echo mapping and modified Pfirrmann score at 3 T—Preliminary results.
time sequence: First images of a male volunteer. Phys Med Biol 2012; Radiology 2012;265(2):555-564.
57(14):4555-4567.
64. Ouwerkerk R, Weiss RG, Bottomley PA. Measuring human cardiac tis-
45. Shajan G, Mirkes C, Buckenmaier K, Hoffmann J, Pohmann R, sue sodium concentrations using surface coils, adiabatic excitation,
Scheffler K. Three-layered radio frequency coil arrangement for and twisted projection imaging with minimal T-2 losses. J Magn
sodium MRI of the human brain at 9.4 Tesla. Magn Reson Med 2016; Reson Imaging 2005;21(5):546-555.
75(2):906-916.
65. Juras V, Zbýň Š, Pressl C, et al. Sodium MR imaging of Achilles ten-
46. Kempson GE, Muir H, Swanson SA, Freeman MA. Correlations dinopathy at 7 T: Preliminary results. Radiology 2012;262(1):199-205.
between stiffness and the chemical constituents of cartilage on the
human femoral head. Biochim Biophys Acta 1970;215(1):70-77. 66. Juras V, Winhofer Y, Szomolanyi P, et al. Multiparametric MR imaging
depicts glycosaminoglycan change in the Achilles tendon during cip-
47. Mankin HJ. Biochemical and metabolic aspects of osteoarthritis. rofloxacin administration in healthy men: Initial observation. Radiology
Orthop Clin North Am 1971;2(1):19-31. 2015;275(3):763-771.
48. Zbyn S, Brix MO, Juras V, et al. Sodium magnetic resonance imaging 67. Bansal N, Szczepaniak L, Ternullo D, Fleckenstein JL, Malloy CR.
of ankle joint in cadaver specimens, volunteers, and patients after dif- Effect of exercise on (23)Na MRI and relaxation characteristics of the
ferent cartilage repair techniques at 7 T: Initial results. Invest Radiol human calf muscle. J Magn Reson Imaging 2000;11(5):532-538.
2015;50(4):246-254.
68. Constantinides CD, Gillen JS, Boada FE, Pomper MG, Bottomley PA.
49. Trattnig S, Welsch GH, Juras V, et al. 23Na MR imaging at 7 T after Human skeletal muscle: Sodium MR imaging and quantification-
knee matrix-associated autologous chondrocyte transplantation pre- potential applications in exercise and disease. Radiology 2000;216(2):
liminary results. Radiology 2010;257(1):175-184. 559-568.
50. Schmitt B, Zbýň Š, Stelzeneder D, et al. Cartilage quality assessment 69. Dyke JP, Meyring-Wösten A, Zhao Y, Linz P, Thijssen S, Kotanko P.
by using glycosaminoglycan chemical exchange saturation transfer Reliability and agreement of sodium ((23)Na) MRI in calf muscle and
and (23)Na MR imaging at 7 T. Radiology 2011;260(1):257-264. skin of healthy subjects from the US. Clin Imaging 2018;52:100-105.
51. Zbyn S, Stelzeneder D, Welsc GH, et al. Evaluation of native hyaline 70. Gerhalter T, Gast LV, Marty B, Uder M, Carlier PG, Nagel AM.
cartilage and repair tissue after two cartilage repair surgery techniques Assessing the variability of (23) Na MRI in skeletal muscle tissue:
with 23Na MR imaging at 7 T: Initial experience. Osteoarthr Cartil Reproducibility and repeatability of tissue sodium concentration mea-
2012;20(8):837-845. surements in the lower leg at 3 T. NMR Biomed 2020;33(5):e4279.
52. Marlovits S, Singer P, Zeller P, Mandl I, Haller J, Trattnig S. Magnetic 71. Chang G, Wang L, Schweitzer ME, Regatte RR. 3D 23Na MRI of
resonance observation of cartilage repair tissue (MOCART) for the human skeletal muscle at 7 Tesla: Initial experience. Eur Radiol 2010;
evaluation of autologous chondrocyte transplantation: Determination 20(8):2039-2046.
of interobserver variability and correlation to clinical outcome after
2 years. Eur J Radiol 2006;57(1):16-23. 72. Nagel AM, Amarteifio E, Lehmann-Horn F, et al. 3 Tesla sodium inver-
sion recovery magnetic resonance imaging allows for improved visual-
53. Madelin G, Lee JS, Inati S, Jerschow A, Regatte RR. Sodium inversion ization of intracellular sodium content changes in muscular
recovery MRI of the knee joint in vivo at 7T. J Magn Reson 2010;207 channelopathies. Invest Radiol 2011;46(12):759-766.
(1):42-52.
73. Gerhalter T, Gast LV, Marty B, et al. (23) Na MRI depicts early changes
54. Tochigi Y, Buckwalter JA, Martin JA, et al. Distribution and progres- in ion homeostasis in skeletal muscle tissue of patients with Duchenne
sion of chondrocyte damage in a whole-organ model of human ankle muscular dystrophy. J Magn Reson Imaging 2019;50(4):1103-1113.
intra-articular fracture. J Bone Joint Surg Am 2011;93(6):533-539.
74. Kopp C, Linz P, Dahlmann A, et al. 23Na magnetic resonance
55. Millington SA, Li B, Tang J, et al. Quantitative and topographical eval- imaging-determined tissue sodium in healthy subjects and hyperten-
uation of ankle articular cartilage using high resolution MRI. J Orthop sive patients. Hypertension 2013;61(3):635-640.
Res 2007;25(2):143-151.
75. Kopp C, Linz P, Wachsmuth L, et al. (23)Na magnetic resonance imag-
56. Wheaton AJ, Borthakur A, Dodge GR, Kneeland JB, Schumacher HR, ing of tissue sodium. Hypertension 2012;59(1):167-172.
Reddy R. Sodium magnetic resonance imaging of proteoglycan
depletion in an in vivo model of osteoarthritis. Acad Radiol 2004;11 76. Linz P, Santoro D, Renz W, et al. Skin sodium measured with (2)(3)Na
(1):21-28. MRI at 7.0T. NMR Biomed 2015;28(1):54-62.

57. Wheaton AJ, Borthakur A, Shapiro EM, et al. Proteoglycan loss in 77. Gai ND, Rochitte C, Nacif MS, Bluemke DA. Optimized three-
human knee cartilage: Quantitation with sodium MR imaging– dimensional sodium imaging of the human heart on a clinical 3T scan-
feasibility study. Radiology 2004;231(3):900-905. ner. Magn Reson Med 2015;73(2):623-632.

58. Zbyn S, Schreiner M, Juras V, et al. Assessment of low-grade focal car- 78. Constantinides CD, Kraitchman DL, O’Brien KO, Boada FE, Gillen J,
tilage lesions in the knee with sodium MRI at 7 T: Reproducibility and Bottomley PA. Noninvasive quantification of total sodium concentra-
short-term, 6-month follow-up data. Investigative Radiology 2020;55 tions in acute reperfused myocardial infarction using Na-23 MRI.
(7):430-437. Magn Reson Med 2001;46(6):1144-1151.

59. Ooms KJ, Cannella M, Vega AJ, Marcolongo M, Polenova T. 23Na 79. Konstandin S, Schad LR. Two-dimensional radial sodium heart MRI
TQF NMR imaging for the study of spinal disc tissue. J Magn Reson using variable-rate selective excitation and retrospective electrocar-
2008;195(1):112-115. diogram gating with golden angle increments. Magn Reson Med
2013;70(3):791-799.
60. Ooms KJ, Cannella M, Vega AJ, Marcolongo M, Polenova T. The
application of 23Na double-quantum-filter (DQF) NMR spectroscopy 80. Pabst T, Sandstede J, Beer M, Kenn W, Neubauer S, Hahn D. Sodium
for the study of spinal disc degeneration. Magn Reson Med 2008;60 T-2* relaxation times in human heart muscle. J Magn Reson Imaging
(2):246-252. 2002;15(2):215-218.

July 2021 73
Journal of Magnetic Resonance Imaging

81. Chang DC, Woessner DE. Spin-echo study of Na-23 relaxation in 101. Hilal SK, Maudsley AA, Ra JB, et al. Invivo NMR imaging of Na-23 in
skeletal-muscle — Evidence of sodium ion binding inside a biological the human head. J Comput Assist Tomogr 1985;9(1):1-7.
cell. J Magn Reson 1978;30(2):185-191.
102. Meyer MM, Haneder S, Konstandin S, et al. Repeatability and repro-
82. Pabst T, Sandstede J, Beer M, et al. Optimization of ECG-triggered ducibility of cerebral (23)Na imaging in healthy subjects. BMC Med
3D Na-23 MRI of the human heart. Magn Reson Med 2001;45(1): Imaging 2019;19(1):26.
164-166.
103. Boada FE, LaVerde G, Jungreis C, Nemoto E, Tanase C, Hancu I. Loss
83. Nielles-Vallespin S, Weber MA, Bock M, et al. 3D radial projection tech- of cell ion homeostasis and cell viability in the brain: What sodium
nique with ultrashort echo times for sodium MRI: Clinical applications in MRI can tell us. Curr Top Dev Biol 2005;70:77-101.
human brain and skeletal muscle. Magn Reson Med 2007;57(1):74-81.
104. Ouwerkerk R, Bleich KB, Gillen JS, Pomper MG, Bottomley PA. Tissue
84. Sandstede JJW, Hillenbrand H, Beer M, et al. Time course of Na-23 sodium concentration in human brain tumors as measured with Na-23
signal intensity after myocardial infarction in humans. Magn Reson MR imaging. Radiology 2003;227(2):529-537.
Med 2004;52(3):545-551.
105. Ouwerkerk R, Pomper MG, Bottomley PA. Quantitative 3D sodium
85. Lee RF, Giaquinto R, Constantinides C, Souza S, Weiss RG, MRI of brain tumors. Radiology 2001;221:294-294.
Bottomley PA. A broadband phased-array system for direct phospho-
rus and sodium metabolic MRI on a clinical scanner. Magn Reson Med 106. Gilles A, Nagel AM, Madelin G. Multipulse sodium magnetic reso-
2000;43(2):269-277. nance imaging for multicompartment quantification: Proof-of-concept.
Sci Rep 2017;7(1):17435.
86. Boehmert L, Waiczies H, Kuehne A, et al. Cardiorenal sodium MRI in
small rodents using a quadrature birdcage volume resonator at 9.4 T. 107. Fiege DP, Romanzetti S, Mirkes CC, Brenner D, Shah NJ. Simulta-
Magnetic Resonance Materials in Physics, Biology and Medicine 2019; neous single-quantum and triple-quantum-filtered MRI of 23Na
33:121-130. (SISTINA). Magn Reson Med 2013;69(6):1691-1696.

87. Noeske R, Seifert F, Rhein KH, Rinneberg H. Human cardiac imaging 108. Neto LPN, Madelin G, Sood TP, et al. Quantitative sodium imag-
at 3 T using phased array coils. Magn Reson Med 2000;44(6):978-982. ing and gliomas: A feasibility study. Neuroradiology 2018;60(8):
795-802.
88. Winkelmann S, Schaeffter T, Koehler T, Eggers H, Doessel O. An opti-
mal radial profile order based on the golden ratio for time-resolved 109. Biller A, Badde S, Nagel A, et al. Improved brain tumor classification
MRI. IEEE Trans Med Imaging 2007;26(1):68-76. by sodium MR imaging: Prediction of IDH mutation status and tumor
progression. AJNR Am J Neuroradiol 2016;37(1):66-73.
89. Foltz WD, al-Kwifi O, Sussman MS, Stainsby JA, Wright GA. Opti-
mized spiral imaging for measurement of myocardial T-2 relaxation. 110. Tsang A, Stobbe RW, Asdaghi N, et al. Relationship between sodium
Magn Reson Med 2003;49(6):1089-1097. intensity and perfusion deficits in acute ischemic stroke. J Magn
90. Ouwerkerk R, Bottomley PA, Solaiyappan M, et al. Tissue sodium con- Reson Imaging 2011;33(1):41-47.
centration in myocardial infarction in humans: A quantitative Na-23 111. Thulborn KR, Davis D, Snyder J, Yonas H, Kassam A. Sodium MR
MR imaging study. Radiology 2008;248(1):88-96. imaging of acute and subacute stroke for assessment of tissue viabil-
91. Resetar A, Hoffmann SH, Graessl A, et al. Retrospectively-gated CINE ity. Neuroimaging Clin N Am 2005;15(3):639-653.
Na-23 imaging of the heart at 7.0 Tesla using density-adapted 3D 112. Hussain MS, Stobbe RW, Bhagat YA, et al. Sodium imaging intensity
projection reconstruction. Magn Reson Imaging 2015;33(9):1091- increases with time after human ischemic stroke. Ann Neurol 2009;66
1097. (1):55-62.
92. Jansen MA, van Emous JG, Nederhoff MGJ, van Echteld CJA. Assess-
113. Young EA, Fowler CD, Kidd GJ, et al. Imaging correlates of
ment of myocardial viability by intracellular Na-23 magnetic resonance
decreased axonal Na+/K+ ATPase in chronic multiple sclerosis
imaging. Circulation 2004;110(22):3457-3464.
lesions. Ann Neurol 2008;63(4):428-435.
93. Neuberger T, Greiser A, Nahrendorf M, Jakob PM, Faber C,
114. Inglese M, Madelin G, Herbert J. Quantitative tissue sodium concen-
Webb AG. Na-23 microscopy of the mouse heart in vivo using
tration mapping in multiple sclerosis with sodium magnetic resonance
density-weighted chemical shift imaging. Magn Reson Mater Phys Biol
imaging. Neurology 2008;70(11):A6-A6.
Med 2004;17(3–6):196-200.
115. Zaaraoui W, Konstandin S, Audoin B, et al. Distribution of brain
94. Vanechteld CJA, Kirkels JH, Eijgelshoven MHJ, van der Meer P,
sodium accumulation correlates with disability in multiple sclerosis: A
Ruigrok TJC. Intracellular sodium during ischemia and calcium-free
cross-sectional 23Na MR imaging study. Radiology 2012;264(3):
perfusion — A Na-23 NMR-study. J Mol Cell Cardiol 1991;23(3):
859-867.
297-307.

95. Sandstede JJ, Pabst T, Beer M, et al. Assessment of myocardial infarc- 116. Huhn K, Mennecke A, Linz P, et al. (23)Na MRI reveals persistent
tion in humans with (23)Na MR imaging: Comparison with cine MR sodium accumulation in tumefactive MS lesions. J Neurol Sci 2017;
imaging and delayed contrast enhancement. Radiology 2001;221(1): 379:163-166.
222-228. 117. Huhn K, Engelhorn T, Linker RA, Nagel AM. Potential of sodium MRI
96. Lott J, Platt T, Niesporek SC, et al. Corrections of myocardial tissue as a biomarker for neurodegeneration and Neuroinflammation in mul-
sodium concentration measurements in human cardiac Na-23 MRI at tiple sclerosis. Front Neurol 2019;10:84.
7 Tesla. Magn Reson Med 2019;82(1):159-173.
118. Maarouf A, Audoin B, Konstandin S, et al. Topography of brain
97. Christa M, Weng AM, Geier B, et al. Increased myocardial sodium sig- sodium accumulation in progressive multiple sclerosis. MAGMA 2014;
nal intensity in Conn’s syndrome detected by Na-23 magnetic reso- 27(1):53-62.
nance imaging. Eur Heart J-Cardiovasc Imag 2019;20(3):263-270.
119. Petracca M, Vancea RO, Fleysher L, Jonkman LE,
98. Haneder S, Juras V, Michaely HJ, et al. In vivo sodium (Na-23) imag- Oesingmann N, Inglese M. Ultra-high field MRI of intra- and
ing of the human kidneys at 7 T: Preliminary results. Eur Radiol 2014; extra-cellular sodium concentration in multiple sclerosis. Mult
24(2):494-501. Scler J 2014;20:276-276.

99. Zaric O, Pinker K, Zbyn S, et al. Quantitative sodium MR imaging at 120. Eisele P, Konstandin S, Szabo K, et al. Temporal evolution of acute
7 T: Initial results and comparison with diffusion-weighted imaging in multiple sclerosis lesions on serial sodium ((23)Na) MRI. Mult Scler
patients with breast tumors. Radiology 2016;280(1):39-48. Relat Disord 2019;29:48-54.

100. Lachner S, Zaric O, Utzschneider M, et al. Compressed sensing recon- 121. Stobbe RW, Beaulieu C. Calculating potential error in sodium MRI
struction of 7 Tesla (23)Na multichannel breast data using (1)H MRI with respect to the analysis of small objects. Magn Reson Med 2018;
constraint. Magn Reson Imaging 2019;60:145-156. 79(6):2968-2977.

74 Volume 54, No. 1

 Zaric et al.: Frontiers of Sodium MRI Revisited

122. Meyer MM, Schmidt A, Benrath J, et al. Cerebral sodium ((23)Na) 124. Reetz K, Romanzetti S, Dogan I, et al. Increased brain tissue sodium
magnetic resonance imaging in patients with migraine —A case- concentration in Huntington’s disease — A sodium imaging study at
control study. Eur Radiol 2019;29(12):7055-7062. 4 T. Neuroimage 2012;63(1):517-524.

123. Mellon EA, Pilkinton DT, Clark CM, et al. Sodium MR imaging detec- 125. Bydder M, Zaaraoui W, Ridley B, et al. Dynamic (23)Na MRI — A non-
tion of mild Alzheimer disease: Preliminary study. AJNR invasive window on neuroglial-vascular mechanisms underlying brain
Am J Neuroradiol 2009;30(5):978-984. function. Neuroimage 2019;184:771-780.

July 2021 75