Sage Journals: Discover world-class research

Abstract

French

Rationale and Objectives: Fat quantification accuracy using a commercial single-voxel high speed T₂-corrected multi-echo (HISTO) technique and its robustness to R₂* variations at 3.0 T, such as those introduced by iron in liver, has not been fully established. This study evaluated HISTO at 3.0 T and sought to reproduce results at 1.5 T. Methods: Phantoms were prepared with a range of fat content and R₂*. Data were acquired at 1.5 T and 3.0 T, using HISTO and a Dixon technique. Fat quantification accuracy was evaluated as a function of R₂*. The patient study included 239 consecutive patients. Data were acquired at 1.5 T or 3.0 T, using HISTO and Dixon techniques. The techniques were compared using Bland-Altman plots. Bias significance was evaluated using a one-sample t-test. Results: In phantoms, HISTO was accurate within 10% up to a R₂* of 100 s⁻¹ at both field strengths, while Dixon was accurate within 10% where R₂* was accurately quantified (up to 350 s⁻¹ at 1.5 T, and 550 s⁻¹ at 3.0 T). In patients, where R₂* was <100 s⁻¹, fat quantification from both techniques agreed at 1.5 T (P = .71), but not at 3.0 T (P = .007), with a bias <1%. Conclusion: Results suggest that HISTO is reliable when R₂* is <100 s⁻¹, corresponding to patients with at most mild liver iron overload, and that it should be used with caution when R₂* is >100 s⁻¹. Dixon should be preferred for hepatic fat quantification due to its robustness to R₂* variations.

Visual Abstract

This is a visual representation of the abstract.

Keywords

liver fat quantification HISTO iron

Introduction

Steatotic liver disease (SLD)¹ is a liver condition characterized by abnormal fat accumulation in more than 5% of hepatocytes.^2,3 Histopathological analysis of tissue samples obtained by percutaneous biopsy is the historical gold standard to quantify hepatic fat. However, it is subject to sampling error in heterogeneous hepatic fat.

Multiple non-invasive methods have been developed to evaluate SLD such as transient elastography and ultrasound.⁴ The current non-invasive gold-standard is magnetic resonance (MR)-based multi-echo Dixon. This technique measures the proton density fat fraction (PDFF), corresponding to the ratio of the fat protons signal to the signal of water and fat protons.⁵ PDFF has shown excellent correlation with fat content from biopsies.2,3 However, Dixon-based techniques are limited by fat-water swaps, where the fat signal is incorrectly assigned to the water signal and vice versa, leading to inaccurate quantification.⁶ A previous study investigated the prevalence of fat-water swaps and showed that 8% of cases suffered from fat-water swaps at 3.0 T.⁷

MR spectroscopy (MRS) measures the chemical composition of fat and is an alternative method to multi-echo Dixon PDFF.² However, MRS techniques offer limited spatial coverage and are sensitive to T₂-signal decay, leading to inaccurate quantification.⁸ Nonetheless, MRS is useful for fat quantification as it does not suffer from fat-water swaps.

The high speed T₂-corrected multi-echo (HISTO) technique has been proposed to address some of the limitations of MRS approaches.⁹ HISTO is an MRS technique that enables fat quantification from a single breath-hold and includes a correction for T₂-signal decay.⁸ HISTO is commercially available, and studies have demonstrated its accuracy for hepatic fat quantification at 1.5 T.^8-10

HISTO fat quantification accuracy has also been demonstrated at 3.0 T in fat-water phantoms.¹¹ Its accuracy at 3.0 T has also been investigated in a few clinical indications. Fat fractions estimated with HISTO and Dixon were notably consistent with a strong positive correlation in thigh muscles.¹² A study performed in liver also demonstrated consistent results between HISTO and Dixon for hepatic fat quantification.¹³

Hepatic fat quantification from HISTO, however, requires further investigation since iron accumulation can decrease its accuracy. Different conditions can lead to iron accumulation in liver, such as hereditary haemochromatosis and repeated blood transfusions.¹⁴ Fat and iron can also simultaneously accumulate in the liver. Notably, previous studies have shown that up to a third of patients with non-alcoholic fatty liver disease also had elevated iron content, which has been related to adverse outcomes.^15,16 Iron accumulation increases both R₂ and R₂* relaxation rates, which can bias the fat quantification.⁸ R₂ and R₂* increase with field strength, thus leading to a higher potential for bias at 3.0 T. While a previous study performed at 1.5 T has shown accurate fat quantification with HISTO in the presence of an iron-based agent in phantoms,⁸ this has not been demonstrated at 3.0 T. Therefore, the objectives of this work were to validate the commercial HISTO fat quantification accuracy at 3.0 T using multi-echo Dixon as the reference method, to establish its robustness to R₂* variations, and to compare the results to measurements performed at 1.5 T.

Methods

Phantom

Fat-water phantoms were designed to mimic a range of liver fat and iron content, using a diluted fat-water emulsion with an R₂* modulator.^17-19 The phantoms were based on a commercial safflower oil emulsion (Microlipid, Nestle Inc.) with 50% fat mass by volume. This emulsion is stable in its liquid form at room temperature, has a controlled fat content, and its fat resonances are consistent with those measured in vivo in the liver.²⁰ Additionally, it has shown good correlation between fat mass fractions and PDFF estimates in a previous study.²¹ The phantom design is described in Supplemental Materials and is illustrated in Figure 1.

Figure 1.

Example images and maps obtained with qDixon at 3.0 T for the phantom with one of the 4 fat fractions (fat mass fraction = 12.67%) used in this work. The schematic of the phantom is shown on the lower right panel, where each vial corresponds to a different [MnCl₂]. Note the variations in fat signal and PDFF with increasing [MnCl₂] (increasing R₂*). The red circle in the top left panel shows an example of the ROI that was used in all phantom vials for qDixon PDFF and R₂* measurements. The green square in the same panel illustrates typical placement of the HISTO voxel for data acquisition in phantom vials. The signal distortions visible around the vials are caused by the ink used to identify the graduations on the centrifuge tubes and were avoided in ROI placement.

Phantoms were scanned using 2 MRI scanners, 1.5 T (Siemens Aera, software VE11C) and 3.0 T (Siemens Skyra, software VE11C). The standard 20-channel head coil provided by the vendor was used. Phantoms were stored in the scanner room for 24 hours before imaging and temperature was measured in the large water compartment before and after imaging.

Phantoms were scanned with 3 different commercial sequences: 2 HISTO sequences performed for each vial, and 1 multi-echo Dixon sequence (qDixon). The HISTO sequences used different echo times (TEs). The first sequence was performed with the default TEs,⁸ while the second used shorter TEs, specifically designed to measure fat content in the presence of iron. This technique, known as HISTO Iron Overload, is provided by the vendor as an alternative in instances where iron overload is suspected. Table 1 details the parameters of these sequences.

Table 1.

MRI and MRS Pulse Sequence Parameters at 1.5 T and 3.0 T.

Parameters		HISTO (1.5 T, 3.0 T)	HISTO Iron Overload (1.5 T, 3.0 T)	qDixon (1.5 T)	qDixon (3.0 T)
Voxel size (mm³)	Phantoms	15 × 15 × 15	15 × 15 × 15	1.25 × 1.59 × 7	1.25 × 1.59 × 7
Voxel size (mm³)	Patients	30 × 30 × 30	—	2.80 × 3.13 × 7	2.80 × 3.13 × 7
Matrix size		Single voxel	Single voxel	160 × 126 × 64	160 × 126 × 64
TE (ms)		12, 24, 36, 48, 72	12, 15, 18, 21, 24	2.38, 4.76, 7.14, 9.52, 11.90, 14.28	1.41, 2.72, 4.03, 5.34, 6.65, 7.96
Bandwidth (Hz/px)		1200	1200	1080	1080
TR (ms)		3000	3000	15.8	9.76
Parallel imaging (CAIPIRINHA)		—	—	2 × 2	2 × 2
Flip angle (°)		90	90	4	4
Mixing time (ms)		10	10	—	—
Scan time (s)	Phantoms	15	15	24	14
Scan time (s)	Patients	15	15	15-20	12-18

Quantitative measurements were obtained using the vendor’s automatic post-processing tools (Liver Lab, Siemens Healthineers). PDFF and R₂* maps were automatically generated for qDixon using a signal model that includes a single R₂* and a multi-resonances fat spectrum. PDFF estimates were obtained automatically for HISTO, along with water and fat R₂.

The PDFF accuracy for HISTO and qDixon was evaluated at both field strengths. Circular regions of interest (ROI) with an area of 2.5 cm² were drawn in each vial using RadiAnt (Medixant, Poznan, Poland, https://www.radiantviewer.com) on the middle slice of qDixon datasets. The mean and standard deviation of the PDFF and R₂* were measured within all ROIs. R₂* variations were examined as a function of [MnCl₂] adjusted for the water volume ([MnCl₂]_water), which assumes that MnCl₂ is only dissolved in the water phase of the emulsion.²² A robust linear regression using bi-square weight²³ (Matlab function “robustfit” with the option “bisquare”) was performed between R₂* and [MnCl₂]_water using software implemented in MATLAB (R2019a, The MathWorks, Natick, MA). The PDFF accuracy was investigated as a function of R₂*. Variations of HISTO water R₂ were also evaluated as a function of [MnCl₂]_water, as detailed in Supplemental Materials.

Patient Study

A retrospective review of consecutive abdominal MRI examinations was performed under the approval of the local Research Ethics Board (REB). Written informed consent for participants was waived. Inclusion criteria were defined as any abdominal MRI exams performed in adults (age ≥18 years old), for which HISTO and qDixon were acquired. Exams were performed between July and September 2020 at 1.5 T, and between May and August 2020 at 3.0 T, using the same 2 scanners as for the phantoms. All exams with qDixon fat-water swaps were excluded.

A combination of a vendor-provided 18-channel phased array coil and a 32-channel spine coil was used. HISTO and qDixon were each acquired during a single inspiration breath hold. Sequences’ parameters are described in Table 1. The HISTO single voxel was positioned in the liver tissue, avoiding blood vessels.

Quantitative measurements were obtained using the vendor-provided automatic post-processing, as for the phantom. HISTO PDFF was compared to the mean qDixon PDFF estimated over the whole liver,¹³ obtained using the vendor-provided automated liver segmentation tool. The whole liver mean R₂* from qDixon and the fat and water R₂ from HISTO were also recorded.

Statistical analysis was performed using Matlab. Bland-Altman plots were generated to compare PDFF measurements from both techniques at both field strengths. Bias was tested using a one sample, two-tailed t-test. P-values <.05 were considered statistically significant.

Results

Phantom

R₂* was linear with [MnCl₂]_water for vials with low fat fractions and/or low [MnCl₂]. In vials with large fat fraction and large [MnCl₂], the mean R₂* deviated from the expected linear behaviour as a function of [MnCl₂]_water at both field strengths (Figure 2). This suggested a failure of qDixon for very large R₂*. These vials were excluded from further data analysis. Six vials were excluded at both field strengths, corresponding to the 3 largest [MnCl₂] for a fat fraction of 38.81%, the 2 largest [MnCl₂] for a fat fraction of 25.60%, and the largest [MnCl₂] for a fat fraction of 12.67%. After exclusion, a range of R₂* from 5 to 350 s⁻¹ was measured at 1.5 T, and from 5 to 550 s⁻¹ at 3.0 T. An R₂* relaxivity for MnCl₂ of 66 s⁻¹/mM was measured at 1.5 T, and of 103 s⁻¹/mM at 3.0 T (R₂ = 0.87 for both regressions).

Figure 2.

R₂* in phantoms from qDixon as a function of the [MnCl₂]_water for all fat mass fractions, at 1.5 T (left) and 3.0 T (right), showing the mean ± standard deviation measured in circular regions of interest. The variation is linear in vials with lower fat content and/or lower [MnCl₂]_water, but deviates from this trend for high fat content and [MnCl₂]_water because of the inaccurate estimate of R₂*. Vials shown with a star marker (*) were excluded from the data analysis. The same 6 vials were excluded at both field strengths.

The difference between qDixon PDFF and fat mass fraction increased as a function of the fat content and of the [MnCl₂]. At 1.5 T, qDixon showed a PDFF error larger than 10% only for the highest fat fraction (38.81%) for 3 of the [MnCl₂] (5, 3.5, and 0.5 mM) (Figure 3). At 3.0 T, qDixon showed PDFF errors smaller than 10% for all vials except for the highest [MnCl₂] combined with a fat fraction of 12.67%, 25.60%, and 38.81%, and for the highest fat fraction combined with a [MnCl₂] of 3.5 mM (Figure 3).

Figure 3.

Relative difference (%) between the proton density fat fraction (PDFF) estimated with qDixon (left), HISTO (middle), and HISTO Iron Overload (right) in phantoms and the fat mass fraction, at 1.5 T (top) and 3.0 T (bottom), shown as a function of the concentration of manganese chloride in total solution ([MnCl₂]) and of the fat mass fraction. For HISTO and HISTO Iron overload, the PDFF inaccuracy increases significantly as the [MnCl₂] increases, with an error larger than 10% for [MnCl₂] larger or equal to 2 mM. Both HISTO sequences were also inaccurate in the absence of MnCl₂. qDixon was much more accurate over the range of [MnCl₂] studied.

HISTO sequences demonstrated large PDFF inaccuracies (Figure 3). At 1.5 T, HISTO PDFF was accurate within 10% for all fat fractions only for [MnCl₂] of 0.2 and 0.5 mM, while at 3.0 T, this was observed for a [MnCl₂] of 0.2 mM. With HISTO Iron Overload, the PDFF accuracy was slightly improved at 1.5 T, where the PDFF error was smaller than 10% for a [MnCl₂] between 0.2 and 0.8 mM for all fat fractions. This slight improvement in the PDFF accuracy with HISTO Iron Overload was also observed at 3.0 T, but only for [MnCl₂] of 0.2 and 0.5 mM. Both HISTO sequences showed PDFF errors larger than 10% at both field strengths in the absence of MnCl₂ ([MnCl₂] = 0 mM). Vials without MnCl₂ were excluded from further data analysis.

HISTO PDFF accuracy was strongly dependant on R₂* (Figure 4). HISTO was accurate to 10% up to a maximum R₂* of 100 s⁻¹ (water R₂ ~75 s⁻¹). HISTO Iron Overload was slightly more accurate as a function of R₂*, where a maximum R₂* of 170 s⁻¹ (water R₂ ~120 s⁻¹) was associated with a PDFF error smaller than 10%. qDixon was more robust to R₂* variations, resulting in less than 10% PDFF error over the entire range of R₂* that was properly quantified.

Figure 4.

Proton density fat fraction (PDFF) error with HISTO and qDixon in phantoms, at 1.5 T (left) and 3.0 T (right), as a function of the mean R₂* measured based on qDixon measurements. The PDFF error is defined as the difference between the PDFF estimate and the fat mass fraction. Very large PDFF errors were obtained with both HISTO sequences for R₂* larger than 100 s⁻¹. qDixon accuracy was much more stable across the range of R₂* studied, with a PDFF error always under 10%.

Patient Study

Two hundred thirty nine abdominal MRI exams were retrieved, including 110 exams performed at 1.5 T and 129 at 3.0 T. Twelve datasets (10.9%) imaged at 1.5 T were excluded due to qDixon fat-water swaps, resulting in 98 patients (66 women and 32 men, mean age of 61 years old ± 14, median 64 years old). At 3.0 T, 7 datasets (5.4%) were excluded due qDixon fat-water swaps, resulting in 122 exams (59 women and 63 men, mean age of 62 ± 15 years old, median 66 years old).

PDFF, R₂, and R₂* distributions were skewed toward low values. Supporting Figure S2 illustrates the qDixon PDFF and R₂* distributions that were measured at both field strengths, while Supporting Figure S3 shows the HISTO PDFF and R₂ distributions. At 1.5 T, the qDixon PDFF range was between 1.5% and 31.0% (median = 6.0%). At 3.0 T, it was from 1.9% to 37.9% (median = 5.8%). The range of the mean R₂* measured over the whole liver with qDixon was from 20 to 155 s⁻¹ (median = 35 s⁻¹) at 1.5 T, while it was between 33 and 160 s⁻¹ (median = 54 s⁻¹) at 3.0 T. The vast majority of R₂* measurements were below 100 s⁻¹ (99% at 1.5 T and 94% at 3.0 T).

HISTO PDFF was comparable to qDixon PDFF (Figure 5). Both techniques demonstrated an absolute bias smaller than 1% at both field strengths. The bias was not significant at 1.5 T (P-value = .71), whereas at 3.0 T, HISTO significantly underestimated the PDFF on average by 0.89% (P-value = .007).

Figure 5.

Comparison between the hepatic fat quantification in patients with HISTO and qDixon (mean over whole liver) at 1.5 T (left) and 3.0 T (right). The bias between the 2 techniques is very small (under 1%) at both field strengths. The bias was significant (P-value <.05) at 3.0 T only.

Discussion

HISTO PDFF results obtained in this work were accurate up to 10% for R₂* smaller than 100 s⁻¹ in phantoms. HISTO results at 1.5 T disagree with a previous study⁸ performed in phantoms doped with an iron-based contrast agent, where a maximum PDFF error of 4.4% was measured for a water R₂ up to 140 s⁻¹. Phantom results also showed a much larger PDFF error than previous studies that were performed in vivo outside of the liver,¹² or in phantoms without R₂* modulations.¹¹ This discrepancy may be explained by the larger range of R₂* studied in this work. PDFF inaccuracies might also be due to a bias in water R₁ introduced by MnCl₂. The r₁/r₂ is larger for MnCl₂ compared to ferritin,^24,25 resulting in a larger R₁ for a given concentration. Based on previous work,⁸ it appears that HISTO may neglect R₁ contributions, which could lead to PDFF inaccuracies.

HISTO and qDixon provided comparable PDFF with a bias smaller than 1% in patients at 1.5 and 3.0 T, where low R₂* were measured (generally <100 s⁻¹), consistent with previous studies.^13,14 Observations in patients were also consistent with phantom experiments for R₂* smaller than 100 s⁻¹. The bias measured in patients was statistically significant at 3.0 T. However, such a small bias is clinically not significant. The agreement observed between HISTO and qDixon is also consistent with a previous phantom study performed with no R₂* modulator at 1.5 and 3.0 T¹¹.

R₂* in phantoms was representative of values that can be measured in liver. R₂* between 24 and 74 s⁻¹ have been reported for healthy liver at 1.5 T.^26-28 At 1.5 T, a threshold between 67 and 88 s⁻¹ has been reported to distinguish between healthy liver and mild iron overload.²⁸ Severe iron overload can produce R₂* larger than 900 s⁻¹ at 1.5 T.^27,29 Fewer studies have investigated R₂* in liver at 3.0 T, but one study reported a threshold of 117 s⁻¹ for the presence of mild iron overload.³⁰

qDixon provided unreliable R₂* measurements in phantoms with large [MnCl₂] and/or fat content at both 1.5 and 3.0 T. These inaccuracies were likely due to a TE selection that was too long, as the signal at longer TEs fell below the noise floor. Acquisitions with shorter TEs or custom data post-processing with compensation for longer TEs might address this issue. This was not investigated here in order to rely on commercial packages only.

qDixon PDFF quantification was robust to R₂* modulation. The PDFF error over the range of R₂* that could be studied was consistent with a previous phantom study performed with no R₂* modulations.¹¹ This suggests that qDixon PDFF accuracy might be independent of the presence of iron in the liver for an R₂* up to approximately 350 s⁻¹ at 1.5 T and 550 s⁻¹ at 3.0 T. The qDixon PDFF accuracy observed in this work is also consistent with a previous study performed in liver.³ The qDixon post-processing algorithm assumes a liver fat spectrum at body temperature, which differs from the phantom fat spectrum at 20°C. However, given the PDFF accuracy observed in this work, the difference in the water chemical shift caused by the temperature difference,³¹ as well as the differences between the phantom material and liver,²⁰ were considered negligible. The phantom’s temperature was stable and consistent within a range of 0.3°C across scan sessions.

HISTO provided unreliable results in vials without MnCl₂. This was likely due to the large water T₁ (>3 s).³² The TR used was not long enough to enable the complete longitudinal relaxation of water in the absence of MnCl₂. The HISTO TR was not increased to keep the scan time within a manageable breathhold duration. As previously suggested,³ a correction based on the measured water T₁ would be required to remove T₁-weighting. Such correction was not implemented since no T₁ measurement was performed. This would not be a problem in vivo since human tissues have much shorter T₁.

The phantom experiments have certain limitations. The emulsion was liquid, while this is not the case for liver. Liquid solutions have longer T₁ and T₂ values. A gel-based phantom could be used in future work to address this limitation. In addition, the R₂* modulator used was not iron-based, while iron is the agent responsible for R₂* variations in liver. MnCl₂ has been used to mimic liver R₂* modulation in multiple studies,^17-19 but it is an imperfect surrogate.³³ However, iron is stored in liver in proteins called ferritin,³⁴ and MnCl₂ has been shown to have similar relaxation properties as aqueous ferritin solutions,¹⁷ thus supporting its use in the context of this work. Future work should be performed in the presence of iron to confirm these findings.

The in vivo study also has some limitations. First, the HISTO PDFF was compared to the qDixon whole liver mean PDFF. Liver has an inhomogeneous fat spatial distribution.² This can lead to a significant difference between a single voxel measurement and the whole liver measurement. Second, only the HISTO sequence with the default TEs was used for data acquisition. The phantom experiments demonstrated that HISTO Iron Overload should be preferred because of its slightly higher PDFF accuracy. The use of the HISTO sequence with default TEs might explain the significant difference that was observed at 3.0 T between HISTO and qDixon. Third, low R₂* values were measured, a regime in which HISTO showed consistent results with qDixon in phantoms. Larger R₂* values remain to be tested. Most patients also had low fat content, with a median under 6%, which also contributed to low R₂*. Future in vivo validation should be performed through a prospective study that includes patients with moderate to severe liver iron overload. Fourth, no biopsy samples were available for the patient cohort. As a result, no independent ground truth was available. However, previous studies have demonstrated high correlation between Dixon PDFF and biopsy,^2,3 thus supporting the use of Dixon as a ground truth.

Conclusion

In conclusion, this study demonstrated that Dixon is more accurate than HISTO for PDFF quantification as R₂* increases. Dixon should therefore be preferred for hepatic fat quantification. The phantoms results suggest that HISTO should only be used when R₂* is below 100 s⁻¹, which correspond to patients without or with mild liver iron overload. The iron content cannot be known prior to the examination, which suggests a limited applicability of HISTO clinically, although it can be an alternative in the case of a fat-water swap with Dixon. The retrospective patient study confirmed that HISTO and Dixon provide consistent results in vivo in the absence of liver iron overload.

Supplemental Material

sj-pdf-1-caj-10.1177_08465371241255896 – Supplemental material for R2* Impact on Hepatic Fat Quantification With a Commercial Single Voxel Technique at 1.5 and 3.0 T

Supplemental material, sj-pdf-1-caj-10.1177_08465371241255896 for R2* Impact on Hepatic Fat Quantification With a Commercial Single Voxel Technique at 1.5 and 3.0 T by Véronique Fortier, Ahmed Mohamed, Evan McNabb, Jérémy Dana, Rita Zakarian, Ives R. Levesque and Caroline Reinhold in Canadian Association of Radiologists Journal

Footnotes

Acknowledgements

The authors acknowledge the Medical Physics Unit (McGill University) for access to lab space, Norma Ybarra (McGill University) for support with the phantom preparation, and the MRI Methods Research Group (McGill University) for useful discussion.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by the Research Institute of the McGill University Health Center General Research Funds.

ORCID iDs

Véronique Fortier

Jérémy Dana

Caroline Reinhold

Supplemental Material

Supplemental material for this article is available online.

References

Rinella

Lazarus

Ratziu

, et al. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. J Hepatol. 2023;79(6):1542-1556. doi:10.1016/j.jhep.2023.06.003

Starekova

Reeder

. Liver fat quantification: where do we stand? Abdom Radiol. 2020;45(11):1-14. doi:10.1007/s00261-020-02783-1

Kukuk

Hittatiya

Sprinkart

, et al. Comparison between modified Dixon MRI techniques, MR spectroscopic relaxometry, and different histologic quantification methods in the assessment of hepatic steatosis. Eur Radiol. 2015;25(10):2869-2879. doi:10.1007/s00330-015-3703-6

Dana

Venkatasamy

Saviano

, et al. Conventional and artificial intelligence-based imaging for biomarker discovery in chronic liver disease. Hepatol Int. 2022;16(3):509-522. doi:10.1007/s12072-022-10303-0

Caussy

Reeder

Sirlin

Loomba

. Non-invasive, quantitative assessment of liver fat by MRI-PDFF as an endpoint in NASH trials. Hepatology. 2018;68(2):763-772. doi:10.1002/hep.29797

Henninger

Zoller

Kannengiesser

Zhong

Jaschke

Kremser

. 3D multiecho Dixon for the evaluation of hepatic iron and fat in a clinical setting. J Magn Reson Imaging. 2017;46(3):793-800. doi:10.1002/jmri.25630

Ladefoged

Hansen

Keller

, et al. Impact of incorrect tissue classification in Dixon-based MR-AC: fat-water tissue inversion. EJNMMI Phys. 2014;1(1):101. doi:10.1186/s40658-014-0101-0

Sharma

Martin

Pineda

, et al. Quantitative analysis of T2-correction in single-voxel magnetic resonance spectroscopy of hepatic lipid fraction. J Magn Reson Imaging. 2009;29(3):629-635. doi:10.1002/jmri.21682

Pineda

Sharma

Vos

Martin

. Measurement of hepatic lipid: high-speed T2-corrected multiecho acquisition at ¹H MR spectroscopy - a rapid and accurate technique. Radiology. 2009;252(2):568-576. doi:10.1148/radiol.2523082084

10.

Sharma

Altbach

Galons

Kalb

Martin

. Measurement of liver fat fraction and iron with MRI and MR spectroscopy techniques. Diagn Interv Radiol. 2014;20(1):17-26. doi:10.5152/dir.2013.13124

11.

Jang

Lee

Kim

, et al. Agreement and reproducibility of proton density fat fraction measurements using commercial MR sequences across different platforms: a multivendor, multi-institutional phantom experiment. Invest Radiol. 2019;54(8):517-523. doi:10.1097/RLI.0000000000000561

12.

Zeng

Han

, et al. Effectiveness of high-speed T2-corrected multiecho MR spectroscopic method for quantifying thigh muscle fat content in boys with Duchenne muscular dystrophy. AJR Am J Roentgenol. 2019;212(6):1354-1360. doi:10.2214/AJR.18.20354

13.

Zhao

Zhou

Liu

, et al. Accuracy of multi-echo Dixon sequence in quantification of hepatic steatosis in Chinese children and adolescents. World J Gastroenterol. 2019;25(12):1513-1523. doi:10.3748/wjg.v25.i12.1513

14.

Horng

Hernando

Reeder

. Quantification of liver fat in the presence of iron overload. J Magn Reson Imaging. 2017;45(2):428-439. doi:10.1002/jmri.25382

15.

Aigner

Weiss

Datz

. Dysregulation of iron and copper homeostasis in nonalcoholic fatty liver. World J Hepatol. 2015;7(2):177-188. doi:10.4254/wjh.v7.i2.177

16.

Dongiovanni

Fracanzani

Fargion

Valenti

. Iron in fatty liver and in the metabolic syndrome: a promising therapeutic target. J Hepatol. 2011;55(4):920-932.

17.

Tao

Zhang

Dai

, et al. An in vitro and in vivo analysis of the correlation between susceptibility-weighted imaging phase values and R2* in cirrhotic livers. PLoS One. 2012;7(9):1-7. doi:10.1371/journal.pone.0045477

18.

St. Pierre

Clark

Chua-Anusorn

, et al. Noninvasive measurement and imaging of liver iron concentrations using proton magnetic resonance. Blood. 2005;105(2):855-861. doi:10.1182/blood-2004-01-0177

19.

Yokoo

Yuan

Sénégas

Wiethoff

Pedrosa

. Quantitative R₂* MRI of the liver with rician noise models for evaluation of hepatic iron overload: simulation, phantom, and early clinical experience. J Magn Reson Imaging. 2015;42(6):1544-1559. doi:10.1002/jmri.24948

20.

Hamilton

Yokoo

Bydder

, et al. In vivo characterization of the liver fat ¹H MR spectrum. NMR Biomed. 2011;24(7):784-790. doi:10.1002/nbm.1622.In

21.

Fortier

Levesque

. MR-oximetry with fat DESPOT. Magn Reson Imaging. 2023;97:112-121. doi:10.1016/j.mri.2022.12.023

22.

Mulkern

Hung

Ababneh

, et al. On the strong field dependence and nonlinear response to gadolinium contrast agent of proton transverse relaxation rates in dairy cream. Magn Reson Imaging. 2005;23(6):757-764. doi:10.1016/j.mri.2005.07.001

23.

Gross

. Confidence intervals for bisquare regression estimates. J Am Stat Assoc. 1977;72(358):341-354. doi:10.1080/01621459.1977.10481001

24.

Vymazal

Zak

Bulte

JWM

Aisen

Brooks

. T₁ and T₂ of ferritin solutions: effect of loading factor. Magn Reson Med. 1996;36(1):61-65.

25.

Thangavel

Saritaş

EÜ

. Aqueous paramagnetic solutions for MRI phantoms at 3 T: a detailed study on relaxivities. Turk J Electr Eng Comput Sci. 2017;25(3):2108-2121. doi:10.3906/elk-1602-123

26.

Obrzut

Atamaniuk

Glaser

, et al. Value of liver iron concentration in healthy volunteers assessed by MRI. Sci Rep. 2020;10(1):1-8. doi:10.1038/s41598-020-74968-z

27.

Wood

Enriquez

Ghugre

, et al. MRI R2 and R2* mapping accurately estimates hepatic iron concentration in transfusion-dependent thalassemia and sickle cell disease patients. Blood. 2005;106(4):1460-1465. doi:10.1182/blood-2004-10-3982

28.

Henninger

Alustiza

Garbowski

Gandon

. Practical guide to quantification of hepatic iron with MRI. Eur Radiol. 2020;30(1):383-393. doi:10.1007/s00330-019-06380-9

29.

Yokoo

Serai

Pirasteh

, et al. Linearity, bias, and precision of hepatic proton density fat fraction measurements by using MR imaging: a meta-analysis. Radiology. 2018;286(2):486-498. doi:10.1148/radiol.2017170550

30.

Dehnad

Nederveen

van der Heide

van Moorselaar

RJA

Hofman

Lagendijk

JJW

. Clinical feasibility study for the use of implanted gold seeds in the prostate as reliable positioning markers during megavoltage irradiation. Radiother Oncol. 2003;67(3):295-302. doi:10.1016/S0167-8140(03)00078-1

31.

Hernando

Sharma

Kramer

Reeder

. On the confounding effect of temperature on chemical shift-encoded fat quantification. Magn Reson Med. 2014;72(2):464-470. doi:10.1002/mrm.24951

32.

Chiarotti

Cristiani

Giulotto

. Proton relaxation in pure liquids and in liquids containing paramagnetic gases in solution. Il Nuovo Cimento. 1955;1(5):863-873. doi:10.1007/BF02731333

33.

Zhao

Hamilton

Brittain

Reeder

Hernando

. Design and evaluation of quantitative MRI phantoms to mimic the simultaneous presence of fat, iron, and fibrosis in the liver. Magn Reson Med. 2020;85(2):734-747. doi:10.1002/mrm.28452

34.

Jensen

Tang

Tosti

, et al. Separate MRI quantification of dispersed (ferritin-like) and aggregated (hemosiderin-like) storage iron. Magn Reson Med. 2010;63(5):1201-1209. doi:10.1002/mrm.22273

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

1.05 MB

R 2 * Impact on Hepatic Fat Quantification With a Commercial Single Voxel Technique at 1.5 and 3.0 T

Abstract

Keywords

Introduction

Methods

Phantom

Patient Study

Results

Phantom

Patient Study

Discussion

Conclusion

Supplemental Material

sj-pdf-1-caj-10.1177_08465371241255896 – Supplemental material for R2* Impact on Hepatic Fat Quantification With a Commercial Single Voxel Technique at 1.5 and 3.0 T

Footnotes

Acknowledgements

Declaration of Conflicting Interests

Funding

ORCID iDs

Supplemental Material

References

Supplementary Material