Amide proton transfer-weighted (APTw) MRI at 3T provides a unique contrast for brain tumor imaging. However, APTw imaging suffers from hyperintensities in liquid compartments such as cystic or necrotic structures and provides a distorted APTw signal intensity. Recently, it has been shown that heuristically motivated fluid suppression can remove such artifacts and significantly improve the readability of APTw imaging.

Theory and Methods

In this work, we show that the fluid suppression can actually be understood by the known concept of spillover dilution, which itself can be derived from the Bloch-McConnell equations in comparison to the heuristic approach. Therefore, we derive a novel post-processing formula that efficiently removes fluid artifact, and explains previous approaches. We demonstrate the utility of this APTw assessment in silico, in vitro, and in vivo in brain tumor patients acquired at MR scanners from different vendors.

Results

Our results show a reduction of the CEST signals from fluid environments while keeping the APTw-CEST signal intensity almost unchanged for semi-solid tissue structures such as the contralateral normal appearing white matter. This further allows us to use the same color bar settings as for conventional APTw imaging.

Conclusion

Fluid suppression has considerable value in improving the readability of APTw maps in the neuro-oncological field. In this work, we derive a novel post-processing formula from the underlying Bloch-McConnell equations that efficiently removes fluid artifact, and explains previous approaches which justify the derivation of this metric from a theoretical point of view, to reassure the scientific and medical field about its use.

1 INTRODUCTION

Amide proton transfer-weighted (APTw) MRI is a CEST imaging technique with contrast correlation on the molecular level.^{1, 2} Despite its valuable clinical utility, especially in the neuro-oncology field,^3-6 the APTw signal intensity is known to be influenced by other tissue parameters.² For example, APTw images, computed with traditional magnetic transfer ratio asymmetry (MTR_asym) metrics, often appear brighter in fluid compartments, such as in necrotic tissues or cysts when compared to semi-solid tissue.² These artificial hyperintensities present a practical problem in their clinical interpretation and raise the question of the molecular origin of APTw imaging.

In 2018, Keupp and Togao came up with an efficient post-processing step that was able to suppress these artificially elevated APTw signal intensities.⁷ The major benefit of this fluid-suppression post-processing approach is that the correction can be calculated from the same CEST dataset without requiring additional information. Moreover, the fluid suppression yields nearly unchanged APTw values for semi-solid tissue compartments, which allows to directly compare fluid-suppressed MTR_asym maps to conventional MTR_asym maps using the same colormap limits. In addition, the proposed fluid suppression already showed valuable clinical results in tumor imaging.^{2, 8-10}

A major drawback of the equation proposed by the authors is that they are heuristically motivated, as described in more detail below. This leads to problems of clear interpretation of the physical mechanisms underlying this method, which is difficult to justify for the future clinical translation of this novel APTw map. In this work, we show that fluid-suppression formulas can actually be understood by the known concept of spillover dilution, which itself can be derived from the Bloch-McConnell (BMC) equations.

In APTw-CEST, a selective RF irradiation pulse is applied to the amide proton pool, which yields to saturation that is transferred to the water pool by chemical exchange. The term RF spillover refers to the concomitant saturation of other nearby protons, in this case the water proton pool is partly directly saturated.^11-13 Interestingly, this RF spillover leads not only to overlapping saturation effects, but also to a so-called spillover dilution of CEST effects. Thereby, the detectable CEST effect becomes effectively smaller due to the simultaneous occurring saturation of the water pool. This spillover dilution effect of CEST signals was first explained by Sun et al.^14-16 who already proposed related corrections. Later studies presented further understanding of spillover dilution by spin lock theory or eigenspace theory, by which could be shown that an inverse metric efficiently removes spillover effects.^17-19 These corrections were applied to the multi-Lorentzian pool fits at 3T and 7T, but have not yet found their way to standard APTw imaging based on MTR_asym analysis. A possible explanation is that the spillover correction increases the CEST signal intensity of semi-solid tissue and, thus, the windowing and color map limits of the APTw images. This made it difficult to compare the APTw maps before and after spillover correction. On the contrary, the heuristic formula⁷ suppresses the signal intensity of the liquid component, maintaining similar color map limits of the MTR_asym based APTw map.

As already mentioned, the correction of spillover dilution effects leads to the increase in the CEST signal intensities of semi-solid tissues: We show herein that the same concept can be used to decrease the CEST signal intensities of liquid environments. This also leads to nearly unchanged APTw values for semi-solid tissue and thus (i) solves the windowing and comparison problem, (ii) explains the mechanism of existing fluid-suppression formulas, and (iii) puts fluid suppression of CEST contrasts now on a solid theoretical basis.

We further demonstrate the benefits of our properly-derived fluid-suppression formula and comparison with the heuristic one and the conventional MTR_asym evaluation in simulations and phantoms experiments, as well as in nine clinical datasets (eight brain tumor and one brain vasogenic edema) acquired at 3T in different clinical centers and with different MRI scanners.

2 THEORY

In an APTw-CEST measurement, we acquire the water signal after frequency-selective RF irradiation at the amide proton resonance frequency. This signal S(

\Delta \omega

) is normalized by an unsaturated scan S₀ to generate the normalized CEST contrast—or Z-spectrum

Z\left(\Delta \omega \right)

. In steady-state, that is, for long enough saturation times, one can derive the quantitative description of the Z-spectrum from the BMC equations. For an in-vivo-like systems that include a water-, semi-solid MT (ssMT), and at least one CEST pool, the steady-state Z-spectrum at the amide proton resonance is given by^20-22

{Z}_{\mathrm{lab}}=Z\left(\Delta \omega =3.5\kern0.4em \mathrm{ppm}\right)={P}^2\frac{R_{1,\mathrm{obs}}}{R_{1\uprho, \mathrm{wmt}}+{f}_{\mathrm{s}}{k}_{\mathrm{s}}\alpha }

(1)

Here,

{R}_{1,\mathrm{obs}}

is the observed longitudinal relaxation rate, as typically determined by inversion recovery experiments, while R_1ρ,wmt

\left(\Delta \omega \right)

describes the RF-driven rotating frame water relaxation rate including the contribution of T₁, T₂, and the ssMT pool.²² The projection factor

{P}^2

can in good approximation to be equal to 1, which is done in the following. The term

{f}_s{k}_s\alpha

actually generates the CEST effect of the solute amide pool s, with the proton fraction

{f}_s

and the exchange rate

{k}_s

, as well as the so-called labeling efficiency α given by:

\alpha =\frac{\omega_1^2}{k_s\left({k}_s+{R}_{2s}\right)+{\omega}_1^2}

(2)

Equation (1) is also called the label scan at the amide proton frequency 3.5 ppm from the water frequency. The assumption of APTw imaging is that, at the opposite frequency

\Delta \omega =-3.5\ \mathrm{ppm}

, the contribution of water and MT (

{R}_{1\uprho, \mathrm{wmt}}

) is identical and the CEST contribution is zero, forming the so-called reference scan:

{Z}_{\mathrm{ref}}=\frac{R_{1,\mathrm{obs}}}{R_{1\rho, \mathrm{wmt}}}

(3)

Using label scan Equation (1) and reference scan Equation (3) allows to generate the so-called magnetization transfer ratio asymmetry

\mathrm{MT}{R}_{\mathrm{asym}}={Z}_{\mathrm{ref}}-{Z}_{\mathrm{lab}}=\frac{f_s{k}_s\alpha \cdotp {R}_{1,\mathrm{obs}}}{R_{1\rho, \mathrm{wmt}}\left({R}_{1\rho, \mathrm{wmt}}+{f}_s{k}_s\alpha\ \right)}

(4)

that quantifies the CEST effect. This evaluation looks different from the ideal steady-state CEST effect as derived previously.²³

\mathrm{MT}{R}_{\mathrm{asym}}=\frac{f_s{k}_s\alpha }{R_{1,\mathrm{obs}}+{f}_s{k}_s\alpha }

(5)

The additional

{R}_{1\rho, \mathrm{wmt}}

terms in the denominator of Equation (4) compared to the ideal CEST effect (Equation 5), explain the actual spillover dilution due to background saturation. The brilliant idea of Sun et al.^14-16 was to separate the ideal CEST effect and the labeling efficiency from the spillover term, that is a spillover dilution factor

\left(1-\sigma \right)

leading to:

\mathrm{MT}{R}_{\mathrm{asym}}=\frac{f_s{k}_s\alpha }{R_{1,\mathrm{obs}}+{f}_s{k}_s\ \alpha}\cdotp \left(1-\sigma \right)

(6)

The original description of

\left(1-\sigma \right)

was rather complex,^14-16 but when we now compare Equations (6) and (4), using again the definitions of the label and reference scans we can derive a new and simple formula for the spillover dilution factor:

\mathrm{MT}{R}_{\mathrm{asym}}=\frac{f_s{k}_s\alpha }{R_{1\mathrm{obs}}+{f}_s{k}_s\ \alpha}\cdotp \underset{\left(1-\sigma \right)={\sigma}^{\prime }}{\underbrace{\frac{R_{1,\mathrm{obs}}+{f}_s{k}_s\alpha }{P^2{R}_{1,\mathrm{obs}}}{Z}_{\mathrm{ref}}{Z}_{\mathrm{lab}}}}

(7)

This is now an exact formula containing the ideal steady-state CEST effect from Equation (5) and a factor that describes the spillover dilution of the ideal CEST effect. For brevity, we use herein σ′ = (1 − σ). Under the assumption

{f}_s{k}_s\ll {R}_{1,\mathrm{obs}}

^{24, 25} and

{Z}_{\mathrm{ref}}\approx {Z}_{\mathrm{lab}}

, the spillover factor is in good approximation given by an astonishingly simple equation, namely the square of the reference scan:

{\sigma}^{\prime }={Z}_{\mathrm{ref}}^2

(8)

This insight has important implications. We now know directly through the additional acquired

{Z}_{\mathrm{ref}}

value the amount of spillover dilution a detected CEST effect suffers from. This implies (i) that we can correct for the spillover dilution by dividing our detected CEST effect by

{\sigma}^{\prime }

, but also (ii) that we can now transfer a detectable CEST effect from the original background (e.g., tumor tissue, cyst)

\sigma^{\prime }={Z}_{\mathrm{ref}}^2

to a target background (e.g., contralateral normal appearing white matter)

{\sigma}_{\mathrm{target}}^{\prime }={Z}_{\mathrm{ref},\mathrm{target}}^2

. Therefore, according to the BMC equations, the MTR_asym is divided by the current voxel's

{\sigma}^{\prime }

and multiplied with

{\sigma_{\mathrm{target}}}^{\prime }

\mathrm{MT}{R}_{\mathrm{asym},\mathrm{target}}=\mathrm{MT}{R}_{\mathrm{asym}}\frac{{\sigma_{\mathrm{target}}}^{\prime }}{\sigma^{\prime }}=\mathrm{MT}{R}_{\mathrm{asym}}\frac{Z_{\mathrm{ref},\mathrm{target}}^2\ }{Z_{\mathrm{ref}}^2}

(9)

Most importantly, we can now translate the CEST effect of a liquid environment to a semi-solid tissue environment.

As stated in the literature⁷ the average

{Z}_{\mathrm{ref},\mathrm{target}}^2

value for brain tissue can be determined to be ˜

{0.5}^2

for APTw imaging, thus any CEST effect transferred to this tissue environment reads:

\mathrm{MT}{R}_{\mathrm{asym},\mathrm{tissue}}=\mathrm{MT}{R}_{\mathrm{asym},\mathrm{FS}}=\mathrm{MT}{R}_{\mathrm{asym}}\frac{0.5^2\ }{Z_{\mathrm{ref}}^2}

(10)

As this equation also transforms CEST effects of liquid areas to a tissue environment, it forms directly the searched- for fluid-suppression formula $\mathrm{MT}{R}_{\mathrm{asym},\mathrm{FS}}$ .

In contrast, the heuristic fluid-suppression equation of Keupp and Togao reads⁷:

\mathrm{MT}{R}_{\mathrm{asym},\mathrm{FS}}=\mathrm{MT}{R}_{\mathrm{asym}}\cdotp \varepsilon \cdotp \left(2-{Z}_{\mathrm{lab}}-{Z}_{\mathrm{ref}}\right)

(11)

Both formulas have now additional weighting factors either $\frac{0.5^2\ }{Z_{\mathrm{ref}}^2}$ or (2 $-{Z}_{\mathrm{lab}}-{Z}_{\mathrm{lab}}$ ) that share the properties (i) that they are close to 1 for semi-solid tissue compartments ( ${Z}_{\mathrm{ref}}$ = 0.5) allowing us to keep the same windowing and colormap, and (ii) must be small for liquid tissue types ( ${Z}_{\mathrm{ref}}$ = 1). Hence, our proposed Equation (10) is expected to show similar features as the heuristic Equation (11), but is based on the BMC equations and, therefore, theoretically justified. In addition, ε should be close to 1.⁷ For this reason, ε was set to 1 for the phantom experiments, whereas in vivo it was set to 0.7 to have similar APTw signal intensity values in normal appearing white matter (NAWM) between MTR_asym maps and heuristic fluid-suppressed MTR_asym maps.

As mentioned in Equation (9) the ${\sigma}_{\mathrm{target}}^{\prime }$ can be chosen according to the tissue we want to transform our signals to. In this work we have chosen that ${\sigma}_{\mathrm{target}}^{\prime }$ for in vivo data will be that of the contralateral NAWM (cNAWM), because ${Z}_{\mathrm{ref}}$ values along the various patients are similar ( ${\sigma}_{\mathrm{target}}^{\prime }={0.35}^2$ ).

3 METHODS

3.1 Standardized pre-saturation using Pulseq-CEST

For all simulations, all phantom measurements, and five out of nine in vivo scans, we used the same standardized APTw CEST protocol APTw_001 as defined in the Pulseq-CEST library https://github.com/kherz/pulseq-cest-library/tree/mrm2023-fluid-supp-APTw, which follows the recommendation of the APTw consensus paper.^{2, 26} It consists of a pulse train of 36 sinc-gauss pulses with a pulse duration t_p = 50 ms, delay time t_d = 5 ms, DC = 91%, T_prep = 1.975 s at a B_1rms of 2 μT. It was repeated at 30 off-resonance frequency offsets in the range between −6 ppm and 6 ppm, and at −300 ppm for the unsaturated reference image. A recovery time of 3.5 s was placed before each pulse train saturation. The exact definition of the CEST sequence can be found here: https://github.com/kherz/pulseq-cest-library/tree/mrm2023-fluid-supp-APTw/seq-library/APTw_3T_001_2uT_36SincGauss_DC90_2s_braintumor.

3.2 In silico: Pulseq-CEST BMC simulation

CEST data were simulated using the Matlab BMC simulation integrated Pulseq-CEST tool version, which is available at https://github.com/kherz/pulseq-cest/tree/v1.1.0. Five different pool systems were simulated with the described sequence APTw_001; their parameters mimic healthy tissue, tumor tissue, and liquid cysts, the detailed parameters are given in Table 1 or are accessible in the Pulseq-CEST library https://github.com/kherz/pulseq-cest-library/tree/mrm2023-fluid-supp-APTw/sim-library/phantoms/l-arginin.

TABLE 1. Simulation parameters mimicking the falcon tubes at pH: 4.0.

	Type	T₁ [s]	T₂ [ms]	[L-Arginine]	f_s	MT
Tube 1	Cyst-like	1.45	1000	20 mM	3.6036 × 10⁻⁰⁴	No
Tube 2	Cyst-like	1.45	1000	27 mM	4.8649 × 10⁻⁰⁴	No
Tube 3	Tissue-like	1.45	50	40 mM	7.2072 × 10⁻⁰⁴	Yes
Tube 4	Tumor-like	1.45	50	60 mM	11 × 10⁻⁰⁴	Yes
Tube 5	Tumor-like	1.45	50	70 mM	13 × 10⁻⁰⁴	Yes

Note: Fixed parameters of the amide pool s are k_s = 350 Hz, δ_s = 3.0 ppm, T_1,s = 1 s, T_2,s = 100 ms. MT pool parameters were (f_MT = 0.035, T₁,_mt = 1 s, T_2,mt = 40 μs, k_mt = 30 Hz).Abbreviation: MT, magnetic transfer.

3.3 In vitro: Phantoms

To demonstrate the effect of fluid suppression experimentally, five 50 mL-falcon-tubes were prepared using L-arginine at different concentrations to provide the exchanging amide proton pool. To ensure an exchange rate of about 350 Hz,²⁷ a pH-value of approximately 4.0 was titrated using hydrochloric acid (HCl) in a 0.9% sodium chloride (NaCl) solution.

Analog to the simulation, the five different tubes mimic healthy tissue (1x), tumor tissue (2x), and liquid cysts (2x). Since the tumor and reference tissue phantoms contain semi-solid compartments, 2% agarose was added to the heated model solution. The T1 relaxation time was adjusted to approximately 1500 ms by the addition of Gadovist in all phantom tubes. The detailed parameters are given in Table 2.

TABLE 2. Composition of the 50 mL falcon tubes at pH: 4.0.

	Type	Gadovist [1 mM]	Agarose	[L-Arginine]
Tube 1	Cyst-like	3.9 μL	0%	20 mM
Tube 2	Cyst-like	3.9 μL	0%	27 mM
Tube 3	Tissue-like	3.9 μL	2%	40 mM
Tube 4	Tumor-like	3.9 μL	2%	60 mM
Tube 5	Tumor-like	3.9 μL	2%	70 mM

Phantom measurements were performed on a 3T scanner (Magnetom Prisma, SIEMENS Healthineers) with a 64-channel head and neck coil. APTw preparation was performed via the APTw_001 protocol before image acquisition using a 3D snapshot GRE.⁹ Readout parameters were: 220 × 180 × 60 mm³ with a matrix size 128 × 104 × 12, voxel size of 1.7 × 1.7 × 5 mm³, TE = 2 ms, TR = 4 ms, bandwidth = 700 Hz/pixel and a flip angle (FA) of 7°. With the snapshot readout the acquisition time per offset was TA = 6.1 s the total scan time was ˜3 min. For simultaneous B0-B1 mapping, the WASAB1 protocol¹⁰ (2 min) was performed with the same readout parameters as for the CEST measurements. For T1 mapping a saturation recovery sequence was used (1:15 min).

3.4 In vivo: Patients

All clinical data were acquired under approval of the local ethics committee (for more information see Declarations).

Five patients diagnosed with brain tumors were scanned with the APTw and WASAB1 protocols described in the previous session along three clinical sites (Erlangen University, University College London Hospitals, La Pitié Hospital) on three 3T scanners (2x MAGNETOM Prisma, 1x MAGNETOM Skyra, Siemens Healthineers), all with a 64-channel head and neck coil.

In addition, to test the consistency of our novel metric on other 3T scanners from other vendors, two pathological brain data sets (one brain tumor and one vasogenic edema) were acquired on a 3T Elition (Philips Healthcare) with a 32-channel head coil (Dent Neurologic Institute) and 2 other pathological brain data sets (pediatric brain tumors) were acquired on a 3T Discovery MR750 (GE HealthCare) with a 16-channel head and neck coil (Necker-Enfant Malades Hospital).

Philips and General Electric APTw data were acquired respectively using turbo spin echo (TSE) (TE = 8 ms, TR = 6000 ms, FA = 90°, matrix size = 128 × 104 × 10, voxel size of 1.8 × 1.8 × 6 mm³) and fast spin echo (FSE) (ETL = 20 ms, TR = 5000 ms, FA = 90°, matrix size = 128 × 96 × 10, acquired voxel size of 1.6 × 2.1 × 8 mm³) protocols with a 2 s saturation pulse of B1rms = 2 μT, 100% Duty Cycle. Data were acquired at following six offsets: ±2.7, ±3.5, ±4.3 ppm. For Philips B0 mapping, field maps were acquired by using MS fast field echo (FFE) sequence (number of echoes = 2, TE = 7 ms, ΔTE = 3 ms, TR = 450 ms, voxel size: 1.8 × 1.8 × 6 mm³, bandwidth = 440 Hz/pixel, FA = 80°) and for GE B0 mapping, a quantitative chemical shift-encoded (CSE) MRI was used (IDEAL-IQ).²⁸

For the post-processing of the APTw and fluid-suppressed APTw maps, the software Olea Sphere 3.0 (Olea Medical, La Ciotat, France) was used. APTw raw data were manually skull-stripped, denoised with Principal-Component-Analysis-based filter²⁹ and spatial filter,³⁰ motion corrected with Elastix³¹ after intra-volumes histogram matching, normalized by M0 volume, and B0-corrected using B0 field map. For 3D snapshot GRE data, the B0 map was computed by the software using fast WASAB1 processing approach.³² For the comparability of the in vivo data from the different scanners, the same color bar was used for all APTw maps within the limits −1% to 5%.

In addition, for each patient a T1 weighted (before (T1w) and after contrast injection (Gd-T1w)), a T1- or T2-fluid-attenuated inversion recovery (FLAIR) and ADC maps were acquired.

3.5 Definition of ${\sigma}_{\mathrm{target}}$ for in vivo data

All in vivo data were evaluated using Eqs. (1), (4), (9), and (10) at 3.5 ppm. In Eqs. (9) and (10), we used ${\sigma}_{\mathrm{target}}^{\prime }={\sigma}_{\mathrm{cNAWM}}^{\prime }$ , which is determined directly from the data to be ${\sigma}_{\mathrm{target}}^{\prime }={\sigma}_{\mathrm{cNAWM}}^{\prime }$ = 0.35². As already written in the theory section, we made this choice because the cNAWM APTw values along the various patients are similar, contrary to the APTw values of the tumor (also contaminated by ssMT which is sensitive to the increase and variable content of macromolecules in the tumor).³³ Surprisingly, cNAWM APTw values were similar for the three sequences (GRE, FSE, TSE), demonstrating the importance of the Consensus Paper for standardization of the APTw protocols.²

4 RESULTS

4.1 In silico

In a first experiment, we apply the newly derived formulas to a controlled Z-spectrum simulation. In Figure 1a), the same CEST pool is once simulated in a liquid/cyst-like environment (blue line) and once in a tissue environment (red line). From the Z-spectra, but also the MTR_asym plots (Figure 1c) it is evident that the CEST effect is much larger in the liquid environment compared to the tissue environment. This is exactly the fluid hyperintensity artifact we want to resolve. When the spillover dilution factors (Figure 1b) are multiplied with the MTR_asym of the liquid system, we get the fluid-suppressed MTR_asym that is now close to the expected tissue MTR_asym, the deviation decreases from +340% down to −16.7% using Equation (7); using Equation (8) even down to 0.6%. Important to note is that the simulation is a pulsed CEST experiment and the equations generalize to this setting.

Details are in the caption following the image — **FIGURE 1**
Open in figure viewer PowerPoint

(A) The simulated Z-spectrum of a cyst-like tissue (blue line) versus that of a tissue-like reference (red line), each with the same concentration of amide protons (40 mM). (C) The corresponding MTR_asym for both tissue types is. Here, the cyst shows a fluid hyperintensity. (B) To correct for that issue, the calculated spillover dilution factor is ( ${\sigma}_1^{\prime },{\sigma}_2^{\prime }$ ) as a function over the frequency offsets Δω. In detail, ${\sigma}_1^{\prime }$ describes an approximation via ${Z}_{\mathrm{ref}}^2$ (Equation 8), while ${\sigma}_2^{\prime }$ provides the more accurate spillover dilution factor according to Equation (7). After fluid-suppression, the MTR_asym for the cyst shows a better overlap with the reference of the same amide proton concentration (cyan line). With the additional prior knowledge of exchange rate, proton fraction, saturation efficiency and ${R}_{1\mathrm{obs}}$ as given via ${\sigma_2}^{\prime }$ , an even improved superposition of the cyst with reference tissue is observed.

4.2 In vitro

In the next experiment we go one step further. We designed a simulation and a set of phantoms to generate a worst case: the cyst-like tube generates higher CEST signals compared to tissue-like tubes and tumor-like tubes. In Figure 2, the fluid suppression was evaluated in simulation and a phantom measurement. Here, the MTR_asym in the cyst-like tubes show the highest image contrast, despite having the lowest amide proton concentrations. After fluid suppression, the image contrast for the cyst-like tubes is clearly reduced, while the image contrast for the tissue-like tubes and both tumor-like tubes remains almost unchanged. Thereby, both the heuristic and the spillover-based fluid suppression show similar behavior. In comparison, the proposed approach reveals a slightly increased intensity for liquid compartments, while solid compartments such as tumor show a marginally lower intensity compared to the heuristic suppression. In both, simulation and measurement, the CEST effect reflects the order of concentration levels only after fluid suppression.

An explanation why the heuristic formula works is given by the analysis in Figure 3. Both fluid-suppression factors show a comparable course when plotted as a function of the reference scan ${Z}_{\mathrm{ref}}$ . This explains the similar effect on the data, especially in liquid regions with high Z-values.

A pixel-by-pixel analysis was carried out to examine the fluid suppressed MTR_asym values as function of the different concentrations (Figure 4). Both fluid suppression approaches (B, C) increase the correlation with the underlying concentrations when compared to the standard MTR_asym (A). Our proposed metric shows even proportionality and passes through zero and 0 mM (B). The heuristic method seems to over-suppress the CEST effect due to a lower correction factor at ${Z}_{\mathrm{ref}}$ > 0.8 (Figure 3), which leads to reduced values at low concentrations.

4.3 In vivo

Nine patients were acquired on different MRI scanners from different vendors, with APTw sequences following the guidelines of the consensus paper, as described in the Methods section. These patients were selected to have liquid components in the peritumoral and intratumoral area that can alter the APTw signal intensity.

Figure 5 a shows three Z-spectra from different region of interest (ROI) placed in cystic, tumor, and cNAWM areas of one patient with a glioma Grade IV, IDH wildtype, acquired with the 3D snapshot-GRE protocol (Siemens 3T Prisma). As can be seen, the various Z-spectra have different shapes and resemble the Z-spectra of the numerical experiments. Focusing on ${Z}_{\mathrm{ref}}$ (−3.5 ppm), values around 0.7 for cyst, 0.4 for tumor area, and 0.35 in cNAWM, can be noticed. As ${Z}_{\mathrm{ref}}$ (−3.5 ppm) for cNAWM was similar across the nine acquired datasets, we used ${\sigma}_{\mathrm{target}}^{\prime }={\sigma}_{\mathrm{cNAWM}}^{\prime }$ = 0.35² for the clinical spillover-corrected fluid suppression according to Equation (9) as well as for the heuristic approach according to Equation (11) with ε = 0.7 for matching the cNAMW of the MTR_asym maps.

As shown in the clinical images of this patient (Figure 6a), the standard APTw map reveals hyperintensities in both cystic areas. The corresponding ADC map also shows elevated values in these areas, indicating a high degree of diffusion of water molecules and thus fluidity. By using the heuristic formula, these artificial APTw values can be suppressed in both cystic regions, while the proposed approach based on the BMC equations lowers the values even further, so that only the signal intensity of the tumor remains elevated. In addition, the fluid-suppressed APTw map shows good agreement with the contrast-enhanced tumor margin represented by the Gd-T1w map. Figure 6B,C further illustrate the effect of fluid suppression based on two more patients with glioblastoma, in which the contrast enhancement is clearly visible at the tumor rim, whereas the central necrotic core appears hypointense in Gd-T1w maps after contrast agent administration. Again, the standard APTw map shows increased signals in both the tumor and necrotic regions. After fluid suppression, only hyperintense signals are found in the tumor tissue and at the tumor margin, while the necrotic regions appear hypointense. In addition, both approaches for fluid suppression show a similar image contrast.

Fluid suppression not only plays a key role in eliminating hyperintense signals of cysts and necrosis, but is also invaluable in postoperative imaging. Figure 7a shows the maps of a patient with glioblastoma who underwent multiple tumor resections, concurrent radiochemotherapy, and ‘Gamma Knife’ radiosurgery. Subsequent MRI examination revealed disease progression with a new contrast-enhanced mass. Interestingly, the cystic surgical cavity shows the same elevated standard APTw signal intensity as the recurrent tumor: this fluid-artifact is effectively corrected by fluid suppression, returning the values in the resection area back to those of cNAWM. Figure 7b shows another glioblastoma patient who has just undergone surgery. In the subcutaneous tissues, a region of edema is visible, post craniotomy, with thickening of the dura mater. While the standard APTw map shows a high signal intensity throughout the ROI, the proposed fluid-suppression formula effectively suppresses all liquid artifacts leaving only the pathological lesion highlighted. This is in good agreement with the Gd-T1w map.

Figure 7C,D show two pediatric cases after brain surgery. In the first pediatric patient, a right parietal lesion was partially removed: a postoperative remnant and a local meningeal infiltration are visible in the MRI maps. Images for the second pediatric patient, operated for left hemispheric tumor presents residue remains and liquid post-surgical cavity. Also, in these cases, the spillover-correction-based approach suppresses the liquid artifacts, which present an elevated signal in the standard APTw maps.

The last two case studies demonstrate fluid suppression in the presence of hemorrhage. Figure 8A shows MR images of a patient with glioblastoma in the right frontal region treated by excision, subsequent radiochemotherapy according to the STUPP protocol, and adjuvant chemotherapy. The patient has both cystic and hemorrhagic compartments clearly distinguishable in the right frontal area. While the standard APTw map highlights both areas, the heuristic and proposed approach effectively suppress the signal in the cystic area but does not abolish the hyper-intensity signal in the hemorrhagic region. Figure 8B shows MR images of a patient with history of cavernous malformation, currently suffering from drug resistant seizures. The clinical images show a cavernous malformation in the left parietal lobe, with both chronic and recent hemorrhage, as well as vasogenic edema. The standard APTw map shows an elevated signal in both hemorrhagic zones. Notably, fluid suppression effectively suppresses the acute hemorrhagic area, which is in the top half of the lesion, while only reducing the signal intensity of the chronic hemorrhage region, which remains hyper-intense. This is important in the context of stroke, where hemorrhagic regions are APTw-hyperintense and ischemic areas are APTw-hypointense. Hemorrhage may have increased APTw signal due to increased protein content.³⁴ The corrected hemorrhagic APTw signal intensity is reduced by the proposed approach because it is diluted but not completely canceled in the liquid component.

5 DISCUSSION

In this work, we have shown that fluid suppression in APTw-CEST imaging can be derived from the BMC equations. It turns out that fluid-suppression can be understood as implicitly using the concept of spillover dilution correction, which was previously discussed in the literature.^{24, 25} While we could not find a theoretical derivation of the heuristic formula, the similarity of the correction factors also explains why the heuristic approach of Keupp and Togao⁷ provides an efficient fluid suppression. In contrast to the heuristic approach, our theoretically founded approach also generalizes to different frequency offset (Figure 1).

When analyzing the fluid suppression (Equation 9), a similarity to the spillover dilution correction using the inverse metric approach^{17, 19, 35} becomes obvious. The spillover-correction metric reads:

\mathrm{MT}{R}_{\mathrm{Rex}}=\frac{1}{Z_{\mathrm{lab}}}-\frac{1}{Z_{\mathrm{ref}}}=\frac{Z_{\mathrm{ref}}-{Z}_{\mathrm{lab}}}{Z_{\mathrm{lab}}{Z}_{\mathrm{ref}}}\approx \frac{{\mathrm{MTR}}_{\mathrm{asym}}}{\sigma^{\prime }}

(12)

Thus, our proposed fluid-suppression metric is just a rescaling of

{R}_{\mathrm{Rex}}

, namely:

\mathrm{MT}{R}_{\mathrm{asym},\mathrm{FS}}=\mathrm{MT}{R}_{\mathrm{Rex}}\cdotp {\sigma}_{\mathrm{target}}^{\prime }

(13)

Here

{\sigma}_{\mathrm{target}}^{\prime }

is just a constant factor reflecting the target tissue spillover dilution factor, in this case

{\sigma}_{\mathrm{target}}^{\prime }={Z}_{\mathrm{ref}}^2

. The reason we feel this rescaling is novel is because it actually describes according to the BMC equations the translation of a specific CEST effect from any tissue environment to a semi-solid reference environment, such as those of cNAWM (Figure 5). And the rescaling factor is not just an arbitrary factor, but it is given by the square of the target reference value

{Z}_{\mathrm{ref}}^2

that actually performs a spillover adjustment. In this case, we translate a CEST effect from any tissue type to a semi-solid target tissue.

Conventional spillover correction can be understood as translation of a CEST effect of any tissue type to a complete liquid (or ideal) environment. An additional effect of using semi-solid tissue as a target tissue is that this allows the usage of the same color map windowing and makes MTR_asym based metrics and spillover-adjusted/fluid-suppressed metrics for APTw imaging directly and visibly comparable.

It should be noted that the presented approach of fluid-suppression does not include any T1 correction as proposed, for example, by the apparent exchange-dependent relaxation rate (AREX) approach.¹⁷ We focused here only to the spillover dilution term, and both outcomes could be still contaminated by T1 of water. As T1 is generally higher in liquid compartments, a T1 correction would further lower the CEST effects in liquid compartments with regard to semi-solid tissue CEST effects.

Concerning the comparison between the heuristic and our proposed approach, a different signal intensity was found in simulations and real measurements. This can directly be understood by Figure 3 where the deviation of the two factors for different Z-values is obvious.

In simulation, it could be shown that the approximation of the spillover dilution factor ${\sigma}^{\prime }={Z}_{\mathrm{ref}}^2$ (Equation 8) is suitable for the correction of the APTw images. Since ${Z}_{\mathrm{ref}}$ is already available over the acquired data, the correction of MTR_asym can be performed very easily via the proposed method. Our theory provides an even more accurate correction term that includes ${f}_s{k}_s\alpha$ and ${R}_{1,\mathrm{obs}}$ (Equation 7). However, this correction factor is not practical for retrospective analysis as we do not know ${f}_s,{k}_s\ or\ \alpha$ . The deviation between both Eqs. (7) and (8) is small as long as ${f}_s{k}_s\alpha$ ≪ ${R}_{1,\mathrm{obs}}$ .

In tumor cases with clearly identified liquid regions, provided by the cystic and necrotic regions, after postoperative surgery and through hemorrhage, the proposed fluid-suppressed approach shows similar fluid suppression when compared to the heuristic method. The suppression clearly removed residual highlighted areas of liquid compartments such as cysts and necrosis (Figure 6), microfluid environments after surgery (Figure 7) and especially hemorrhage (Figure 8) identified by the acquired ADC and T1w maps. The stronger suppression in some tumor areas of the proposed method as visible in the Figures 6B, 7B or 8 is most probably due to a higher water content in these areas. This is supported by the increased ADC values, which indicate a higher water content. On the other hand, it is likely that the protein concentrations are not homogeneous at the tumor rim. As there is no additional histopathological information available nor a longitudinal study, we cannot clarify these findings herein in more detail.

As already described above, the stronger suppression is caused by the larger correction factor at higher Z-values (Figure 3). Nevertheless, both methods show their significant relevance in correcting the APTw maps in order not to derive incorrect conclusions from the data. In addition, we also tested an extension of the heuristic formula, which was previously presented on a conference.³⁶ The results can be found in the supporting information.

It is also important to note that APTw signal intensity in tumors may decrease after fluid suppression (if the chosen target tissue is the one of cNAWM). This phenomenon is in agreement with what has been previously shown in Sun and Sorensen¹⁶: the inverse formula (Equation 13) was presented to be both a metric for spillover and MT correction. In Figure 5, ${Z}_{\mathrm{ref}}$ tumor values (≈0.4) were seen slightly higher than ${Z}_{\mathrm{ref}}$ cNAWM values (≈0.35) due to the presence of the ssMT signal. Therefore, by applying the spillover-corrected/fluid-suppressed formulas, the tumor APTw values will be corrected by the ssMT contamination, resulting in a APTw tumor signal intensity more related to the mobile protein signal and less to macromolecule signal.³³ If, on the other hand, the tumor signal intensity is wanted to be left unchanged, the tumor tissue can be chosen as the target value, as was done heuristically in Keupp and Togao,⁷ which however will lead to a slightly increase in the NAWM signal intensity instead.

We further demonstrated that the proposed metric works for similar acquisition protocols realized on systems from different vendors and in different clinical centers following the guidelines proposed in the APTw consensus paper.² Once the fluid or spillover artifact is removed, the generated image contrast is also theoretically expected to correlate stronger with the actual exchange effects (see Equations 5 vs. 6), and shows first signs of a higher diagnostic value, as shown by initial studies.^10-12 In this small multicentric clinical study, we have also shown the importance of this metric for both adult and pediatric pathology assessment, both before and after brain surgery.

To conclude, with this article we want to further strengthen APTw-CEST imaging based on MTR_asym evaluation, which has been found to be a very important biomarker for tumor grading and other neurological applications. We showed that by the use of both APTw maps (before and after fluid suppression) an improved interpretation of the APTw signal is possible, improving the value of APTw imaging as an indicator for cancer. Our proposed method is a simple post-processing step that can be readily applied to new data, as well as to all existing APTw data.

6 CONCLUSIONS

In this work, we justify fluid-suppression techniques for APTw-CEST imaging from a theoretical point of view and apply a novel formula to simulations, phantoms and in vivo in brain tumor patients with consistent outcomes. We show that fluid suppression clearly improves the readability of APTw maps in the neuro-oncological field. Therefore, its integration in APTw imaging is highly desirable.

7 DECLARATIONS

Ethics approval: University College London Hospital: This prospective, on-going, investigator-initiated, single-centre, neuro-oncology imaging study received ethical and institutional review board approval (IRAS project ID: 210819; Research Ethics Committee reference: 16/LO/2081). All patients, prior to participation, provided written informed consent to participate and for publication. Dent Institute: Dent Neurologic Insitute Written consent was obtained from patients. This prospective study protocol was approved by WCG IRB, formerly (Western Institutional Review Board). Protocol name: Optimization of Neurologic Imaging Dent Neurosciences Research Center. Protocol number: 20170119. Necker Enfant Malades Hospital: Local Institutional Review Board authorization was granted (EDRACT 2014-A00541-46). Informed consent for research was prospectively obtained for the study. La Pitie Hospital (APHP): MR images of this adult patient were acquired at 3 T Skyra MR scanner (Siemens Healthineers, Erlangen, Germany) in the Neuroradiology Department of the Pitié Salpêtrière Hospital. The data were made available by the Clinical Data Warehouse of the Greater Paris University Hospitals (Assistance Publique-Hôpitaux de Paris [AP-HP], reference number: 20211216174611), in accordance with the Ethical and Scientific Board of the AP-H.

CONFLICT OF INTEREST STATEMENT

Patrick Liebig is an employee of Siemens Healthcare. Stefano Casagranda and Christos Papageorgakis are employees of Olea Medical.

Open Research

DATA AVAILABILITY STATEMENT

Sequences for simulation are accessible at our Github Library at https://github.com/kherz/pulseq-cest-library/tree/mrm2023-fluid-supp-APTw, while the in vitro *.yaml files are provided under examples at https://github.com/kherz/pulseq-cest-library/tree/mrm2023-fluid-supp-APTw/sim-library/phantoms/l-arginin. In vivo data can be requested with reasonable justification. The exact version of the Pulseq-CEST repository can be found at https://github.com/kherz/pulseq-cest/tree/v1.1.0.

Supporting Information

REFERENCES

1Zhou J, Payen JF, Wilson DA, Traystman RJ, van Zijl PCM. Using the amide proton signals of intracellular proteins and peptides to detect pH effects in MRI. Nat Med. 2003; 9: 1085-1090. doi:10.1038/nm907
10.1038/nm907
CAS PubMed Web of Science® Google Scholar
2Zhou J, Zaiss M, Knutsson L, et al. Review and consensus recommendations on clinical APT-weighted imaging approaches at 3T: application to brain tumors. Magn Reson Med. 2022; 88: 546-574. doi:10.1002/mrm.29241
10.1002/mrm.29241
PubMed Web of Science® Google Scholar
3Sotirios B, Demetriou E, Topriceanu CC, Zakrzewska Z. The role of APT imaging in gliomas grading: a systematic review and meta-analysis. Eur J Radiol. 2020; 133:109353. doi:10.1016/j.ejrad.2020.109353
10.1016/j.ejrad.2020.109353
PubMed Web of Science® Google Scholar
4Nichelli L, Casagranda S. Current emerging MRI tools for radionecrosis and pseudoprogression diagnosis. Curr Opin Oncol. 2021; 33: 597-607. doi:10.1097/CCO.0000000000000793
10.1097/CCO.0000000000000793
CAS PubMed Web of Science® Google Scholar
5Sartoretti E, Sartoretti T, Wyss M, et al. Amide proton transfer weighted (APTw) imaging based radiomics allows for the differentiation of gliomas from metastases. Sci Rep. 2021; 11: 5506. doi:10.1038/s41598-021-85168-8
10.1038/s41598-021-85168-8
CAS PubMed Web of Science® Google Scholar
6Lingl JP, Wunderlich A, Goerke S, et al. The value of APTw CEST MRI in routine clinical assessment of human brain tumor patients at 3T. Diagnostics. 2022; 12: 490. doi:10.3390/diagnostics12020490
10.3390/diagnostics12020490
CAS PubMed Google Scholar
7Keupp J TO. Magnetization Transfer Ratio based Metric for APTw or CESTw MRI Suppressing Signal from Fluid Compartments-Initial Application to Glioblastoma Assessment. https://archive.ismrm.org/2018/3156.html. Published June 16, 2018. Accessed February 28, 2023. https://cds.ismrm.org/protected/18MProceedings/PDFfiles/3156.html.
Google Scholar
8Mancini L, Casagranda S, Gautier G, et al. CEST MRI provides amide/amine surrogate biomarkers for treatment-naïve glioma sub-typing. Eur J Nucl Med Mol Imaging. 2022; 49: 2377-2391. doi:10.1007/s00259-022-05676-1
10.1007/s00259-022-05676-1
CAS PubMed Web of Science® Google Scholar
9Casagranda S, Mancini L, Gautier G, et al. Fluid suppression in CEST imaging affects predominantly IDH-mutant 1p/19q retained gliomas with T2-FLAIR mismatch. 2020. Accessed February 28, 2023. https://hal.science/hal-03241436.
Google Scholar
10Nichelli L. (ISMRM 2022) Predicting radionecrosis and recurrent disease through Amide Proton Transfer weighted imaging in brain metastases ISMRM 2022. Accessed May 24, 2023. https://archive.ismrm.org/2022/0126.html.
Google Scholar
11Spencer RGS, Horska A, Ferretti JA, Weiss GH. Spillover and incomplete saturation in kinetic measurements. J Magn Reson B. 1993; 101: 294-296. doi:10.1006/jmrb.1993.1045
10.1006/jmrb.1993.1045
CAS Web of Science® Google Scholar
12Horská A, Spencer GS. Correctly accounting for radiofrequency spillover in saturation transfer experiments: application to measurement of the creatine kinase reaction rate in human forearm muscle. Magma. 1997; 5: 159-163. doi:10.1007/BF02592247
10.1007/BF02592247
CAS PubMed Web of Science® Google Scholar
13Baguet E, Roby C. Off-resonance irradiation effect in steady-state NMR saturation transfer. J Magn Reson. 1997; 128: 149-160. doi:10.1006/jmre.1997.1230
10.1006/jmre.1997.1230
CAS PubMed Web of Science® Google Scholar
14Sun PZ, van Zijl PCM, Zhou J. Optimization of the irradiation power in chemical exchange dependent saturation transfer experiments. J Magn Reson. 2005; 175: 193-200. doi:10.1016/j.jmr.2005.04.005
10.1016/j.jmr.2005.04.005
CAS PubMed Web of Science® Google Scholar
15Sun PZ, Zhou J, Huang J, van Zijl P. Simplified quantitative description of amide proton transfer (APT) imaging during acute ischemia. Magn Reson Med. 2007; 57: 405-410. doi:10.1002/mrm.21151
10.1002/mrm.21151
CAS PubMed Web of Science® Google Scholar
16Sun PZ, Sorensen AG. Imaging pH using the chemical exchange saturation transfer (CEST) MRI: correction of concomitant RF irradiation effects to quantify CEST MRI for chemical exchange rate and pH. Magn Reson Med. 2008; 60: 390-397. doi:10.1002/mrm.21653
10.1002/mrm.21653
CAS PubMed Web of Science® Google Scholar
17Zaiss M, Xu J, Goerke S, et al. Inverse Z-spectrum analysis for spillover-, MT-, and T1-corrected steady-state pulsed CEST-MRI-application to pH-weighted MRI of acute stroke. NMR Biomed. 2014; 27: 240-252. doi:10.1002/nbm.3054
10.1002/nbm.3054
CAS PubMed Web of Science® Google Scholar
18Zaiss M, Windschuh J, Goerke S, et al. Downfield-NOE-suppressed amide-CEST-MRI at 7 tesla provides a unique contrast in human glioblastoma. Magn Reson Med. 2017; 77: 196-208. doi:10.1002/mrm.26100
10.1002/mrm.26100
CAS PubMed Web of Science® Google Scholar
19Goerke S, Soehngen Y, Deshmane A, et al. Relaxation-compensated APT and rNOE CEST-MRI of human brain tumors at 3 T. Magn Reson Med. 2019; 82: 622-632. doi:10.1002/mrm.27751
10.1002/mrm.27751
CAS PubMed Web of Science® Google Scholar
20McConnell HM. Reaction rates by nuclear magnetic resonance. J Chem Phys. 1958; 28: 430-431. doi:10.1063/1.1744152
10.1063/1.1744152
CAS Web of Science® Google Scholar
21Jin T, Autio J, Obata T, Kim SG. Spin-locking versus chemical exchange saturation transfer MRI for investigating chemical exchange process between water and labile metaboliteprotons. Magn Reson Med. 2011; 65: 1448-1460. doi:10.1002/mrm.22721
10.1002/mrm.22721
CAS PubMed Web of Science® Google Scholar
22Zaiss M, Zu Z, Xu J, et al. A combined analytical solution for chemical exchange saturation transfer and semi-solid magnetization transfer. NMR Biomed. 2015; 28: 217-230. doi:10.1002/nbm.3237
10.1002/nbm.3237
CAS PubMed Web of Science® Google Scholar
23Zaiss M, Angelovski G, Demetriou E, McMahon MT, Golay X, Scheffler K. QUESP and QUEST revisited-fast and accurate quantitative CEST experiments. Magn Reson Med. 2018; 79: 1708-1721. doi:10.1002/mrm.26813
10.1002/mrm.26813
CAS PubMed Web of Science® Google Scholar
24Sun PZ, Wang Y, Dai Z, Xiao G, Wu R. Quantitative chemical exchange saturation transfer (qCEST) MRI-RF spillover effect-corrected omega plot for simultaneous determination of labile proton fraction ratio and exchange rate. Contrast Media Mol Imaging. 2014; 9: 268-275. doi:10.1002/cmmi.1569
10.1002/cmmi.1569
CAS PubMed Web of Science® Google Scholar
25Zaiss M, Jin T, Kim SG, Gochberg DF. Theory of chemical exchange saturation transfer MRI in the context of different magnetic fields. NMR Biomed. 2022; 35:e4789. doi:10.1002/nbm.4789
10.1002/nbm.4789
CAS PubMed Web of Science® Google Scholar
26Herz K, Mueller S, Perlman O, et al. Pulseq-CEST: towards multi-site multi-vendor compatibility and reproducibility of CEST experiments using an open-source sequence standard. Magn Reson Med. 2021; 86: 1845-1858. doi:10.1002/mrm.28825
10.1002/mrm.28825
CAS PubMed Web of Science® Google Scholar
27Sedykh M. Exchange Rate and Concentration Quantification by CEST MR Fingerprinting. Master's Thesis. Friedrich Alexander Universität; 2022.
Google Scholar
28Roberts NT, Hinshaw LA, Colgan TJ, Ii T, Hernando D, Reeder SB. B0 and B1 inhomogeneities in the liver at 1.5 T and 3.0 T. Magn Reson Med. 2021; 85: 2212-2220. doi:10.1002/mrm.28549
10.1002/mrm.28549
PubMed Web of Science® Google Scholar
29Stefano C, Papageorgakis S, Romdhane F, et al. Principal Component selections and filtering by spatial information criteria for multi-acquisition CEST MRI denoising. Proceedings of Joint Annual Meeting ISMRM-ESMRMB & ISMRT 31st Annual Meeting; London. 2022. Abstract #2080.
Google Scholar
30Buckheit J, Chen S, Donoho D, Johnstone I, Scargle J. AboutWaveLab. 2005 https://statweb.stanford.edu/∼wavelab/Wavelab_850/AboutWaveLab.pdf
Google Scholar
31Klein S, Staring M, Murphy K, Viergever MA, Pluim JPW. Elastix: a toolbox for intensity-based medical image registration. IEEE Trans Med Imaging. 2010; 29: 196-205. doi:10.1109/TMI.2009.2035616
10.1109/TMI.2009.2035616
PubMed Web of Science® Google Scholar
32Papageorgakis C, Firippi E, Gy B, et al. Fast WASABI post-processing: Access to rapid B0 and B1 correction in clinical routine for CEST MRI. Magn Reson Imaging. 2023. doi:10.1016/j.mri.2023.06.001
Google Scholar
33Henkelman RM, Stanisz GJ, Graham SJ. Magnetization transfer in MRI: a review. NMR Biomed. 2001; 14: 57-64. doi:10.1002/nbm.683
10.1002/nbm.683
CAS PubMed Web of Science® Google Scholar
34Wang M, Hong X, Chang CF, et al. Simultaneous detection and separation of hyperacute intracerebral hemorrhage and cerebral ischemia using amide proton transfer MRI. Magn Reson Med. 2015; 74: 42-50. doi:10.1002/mrm.25690
10.1002/mrm.25690
CAS PubMed Web of Science® Google Scholar
35Zaiss M, Bachert P. Chemical exchange saturation transfer (CEST) and MR Z-spectroscopy in vivo: a review of theoretical approaches and methods. Phys Med Biol. 2013; 58: R221-R269. doi:10.1088/0031-9155/58/22/R221
10.1088/0031-9155/58/22/R221
PubMed Web of Science® Google Scholar
36Togao O, Keupp J, Yamashita K, Kikuchi K, Wada T, Ishigami K. Amide proton transfer-weighted MRI of brain tumors with fluid & solid compartment corrections using background magnetization transfer effects. https://submissions.mirasmart.com/ISMRM2023/ViewSubmissionFile.aspx?mode=video&validate=false&sbmID=1803. Published August 17, 2023. Accessed August 17, 2023. https://submissions.mirasmart.com/ISMRM2023/ViewSubmissionFile.aspx?mode=video&validate=false&sbmID=1803.
Google Scholar

Citing Literature

Volume91, Issue4

April 2024

Pages 1354-1367

Fluid suppression in amide proton transfer-weighted (APTw) CEST imaging: New theoretical insights and clinical benefits