Method and system of MRI imaging to estimate liver water r1 and r2* variability

Abstract

The present invention relates to the field of liver imaging using MRI. In particular, the present invention provides means to estimate the variability in liver water R1 and R2* with reference to liver water, iron and fat content. Disclosed is a method of liver imaging comprising the following stages: 1. input stage; 2. T2* fat-correction stage; 3. generation of simulated MOLLI signals stage including: (a) LIC-LWC relaxation rate i) estimating the influence of LWC on R1 and R2* ii) estimating the influence of LIC on R1 and R2* iii) healthy-liver reference water R1, LIC, and R2* values; and (b) estimation of water T2 and (c) generating a dictionary of simulated MOLLI signals; 4. signal matching stage; 5. water T1 iron correction stage; and 6. water T1 standardisation stage including : (a) water T1 field strength correction; b) water T1 vendor correction.

Description

[0001]Method and System of MRI Imaging to Estimate Liver Water R1 and R2* Variability Field of the Invention The present invention relates to the field of liver imaging using MRI. In particular, the present invention provides means to estimate the variability in liver water R1 and R2*with reference to liver water, iron, and fat content.Background T1 and T2* relaxation times, and proton density fat fraction (PDFF), are MRI-derived parameters that have been shown to provide useful information for characterizing liver health. The influence of liver iron concentration (LIC) on liver R1 (1 / T1) and R2* (1 / T2*) in subjects with elevated iron is known. Liver TI is also sensitive to liver waterconcentration (LWC) and has been shown to correlate with biopsy-derivedmeasurements of fibrosis and inflammation, due to the elevation in tissue water content associated with these processes. R2* is composed of the sum of the following terms: R2 (1 / T2), the spin-spin relaxation rate, and R2’, the relaxation rate introduced by magnetic field heterogeneities. R2 is sensitive to both LWC and LIC, and as such changes in either parameter will influence liver R2* values. It is important to understand and correct for these confounding influences on liver R2*, which is addressed by the present invention. Various publications address the influence of iron, water, and fat concentration on liver T1 and T2*. For example, in relation to liver iron concentration, Wood, J. C. et al. (‘Cardiac Iron Determines Cardiac T2*, T2, and T1 in the Gerbil Model of Iron Cardiomyopathy.’Circulation 112, 535–543 (2005)) demonstrated that iron shortens T1, T2, and T2*relaxation times in the heart and liver, using a gerbil iron-overload model. They present calibration curves relating liver R1 to biopsy-determined liver iron, which show potential for estimating the influence of iron on liver R1 values measured in patients with elevated liver iron.Ghugre, et. al., (‘Mechanisms of tissue–iron relaxivity: Nuclear magnetic resonancestudies of human liver biopsy specimens.’ Magn. Reson. Med.54, 1185–1193 (2005)) found that nuclear magnetic resonance (NMR) relaxometry of human liver biopsy specimens yielded unique insights into the mechanisms of tissue-iron relaxivity. Their study aimed to address the underlying proton relaxation mechanisms within the human tissue environment on non-invasive iron quantification, and inter-site and inter- sequence variability, as well as methodological inaccuracies that limit its widespread clinical use. They performed NMR relaxometry on fresh human liver biopsy specimenstaken from patients with transfusion-dependent anaemia and compared NMRparameters to quantitative iron levels and tissue histology. They noted that R1 and R2 both increased linearly with hepatic iron content, with R2 being more sensitive to iron. In a further study, noting that relaxation rates R2 and R2* measured by MRI have different advantages for hepatic iron concentration (HIC) estimation, Wood, et al. (‘MRI R2 and R2* mapping accurately estimates hepatic iron concentration in transfusion-dependent thalassemia and sickle cell disease patients.’ Blood 106, 1460–1465(2005)) compared non-invasive iron estimates using optimized R2 and R2* methods in patients with iron overload and controls. They found that the R2 calibration curve was similar to the curve generated by other researchers, despite differences in technique and instrumentation; that combined R2 and R2* measurements did not yield more accurate results than either alone; and that both R2 and R2* can accurately measure hepatic iron concentration throughout the clinically-relevant range of HIC with appropriate MRI acquisition techniques. St Pierre et al. (‘Non-invasive measurement and imaging of liver iron concentrationsusing proton magnetic resonance.’ Blood 105.2 (2005): 855-861) have shown how theaccumulation of iron in the liver leads to the shortening of T2. They presented a calibration curve relating liver R2 to biopsy-determined LIC. Hernando et al (‘Multicenter Reproducibility of Liver Iron Quantification with 1.5-T and3.0-T MRI,’ Radiology 306, e213256 (2023)) quantified cross-vendor reproducibility ofR2*-based LIC estimation, by comparing liver R2* measurements against estimated LIC values. They evaluated the reproducibility of R2*-based LIC estimation in a multi- centre, multi-vendor study. Participants with known or suspected liver iron overload underwent single breath-hold, multi-echo gradient-echo MRI for R2* mapping at 1.5 T and 3 T (2.89 T or 3.0 T) on the same day. R2* maps from the multi-echo images were reconstructed and analysed. They confirmed a linear relationship between R2* and the LIC measurements, and all calibrations within the same field strength were highly reproducible, showing no evidence of statistically significant centre-specific differences. An MR technique which is commonly used for mapping T1 in the liver is MOdified Look-Locker Inversion-recovery (MOLLI), which widely known for being a robust and fast(single breath-hold) method for acquiring T1 data. However, it is known and accepted that MOLLI-T1 has a dependence on tissue T2, which results in an underestimation of the MOLLI-T1 as T2 is shortened (Piechnik et al. ‘Shortened Modified Look-Locker Inversion recovery (ShMOLLI) for clinical myocardial T1-mapping at 1.5 and 3 T withina 9 heartbeat breathhold.’ Journal of cardiovascular magnetic resonance 12.1 (2010):1-11). As such, in the presence of elevated liver iron, MOLLI-T1 will be reduced both directly and indirectly (the latter being due to the shortening of T2). This acts as a confounding effect to the increase in T1 resulting from increased tissue water content associated with liver inflammation or fibrosis. A correction for the confounding effect of LIC on MOLLI-T1 is described in Tunnicliffe et al. (‘A model for hepatic fibrosis: the competing effects of cell loss and iron on shortened modified Look-Locker inversion recovery T1 (shMOLLI-T1) in theliver’, Journal of Magnetic Resonance Imaging 45.2 (2017): 450-462) andPCT / GB2014 / 050815. This approach may be implemented in commercial or opensource solutions and utilizes Bloch simulations of the acquired MOLLI sequence toestimate the MOLLI T1 for different levels of LIC and extra-cellular fluid volume fraction(ECF). For a given subject, LIC is estimated using measured T2* values, and ECF isestimated by selecting the simulated MOLLI signal (for a given ECF and LIC) which best matches the measured signal. If LIC is elevated, look-up tables are used to estimate the equivalent MOLLI T1 for the same ECF value and normal LIC. This provides a corrected T1 (cT1), in which the confounding effect of LIC is minimized. T2* is sensitive to microscopic inhomogeneities in magnetic susceptibility within the liver tissue. Similar to iron, hepatic fat droplets induce local magnetic susceptibility inhomogeneities, leading to a decrease in T2* (or an increase in R2*). As such, increases in PDFF will lead to an increase in liver R2*, and in patients with normal iron but elevated PDFF, fat may be the dominant influence on the measured R2*, rather than iron. Various publications discuss this. For example, Wang et al. used Monte Carlo simulations to model the relationship between liver R2* and PDFF (‘Monte CarloModelling of Liver MR Signal in the Presence of Fat’, Proc. Intl. Soc. Mag. Reson. Med.26 (2018)). Their results showed a linear correlation between PDFF and R2* (slope of R2* (Hz) relative to PDFF (%) = 0.395 at 1.5T, and 0.772 at 3T).Wang et. al.’s results are consistent with previous in vivo studies, in which the slope ofR2* relative to PDFF (in Hz / %) ranges from 0.77 to 0.87 at 3.0T (Hernando et al. ‘Relationship between liver proton density fat fraction and R2* in the absence of iron overload’, Proc. Intl. Soc. Mag. Reson. Med. 23 (2015), Mamidipalli et al. ‘Cross- sectional correlation between hepatic R2* and proton density fat fraction (PDFF) in children with hepatic steatosis’, J Magn Reson Imaging 2017). Fat also influences T1 measurements obtained in the liver. For instance, MOLLI T1 mapping sequences generally employ a balanced steady-state free precession readout, which is influenced by off-resonance effects between the fat and water signal. This can lead to complex behaviour of the signal during T1 recovery if the imaged voxels contain a mixture of fat and water. With typical MOLLI imaging parameters, and clinically relevant fat fraction, this leads to an over-estimation of the MOLLI T1 value, which is at odds with the decreased T1 expected in a simple fat / water partial volume model (Mozes et al. “Mapping tissue water T1 in the liver using the MOLLI T1 method in the presence of fat, iron and B0 inhomogeneity”, NMR in Biomedicine 2018; e4030). The study by Mozes et al. shows how Bloch simulations of the acquired MOLLI sequence can be used to model the water and fat signals independently, which allowsa ‘water T1’ value to be estimated, which is independent of the confounding effect offat. The authors also show how, by applying a version of the liver model introduced by Tunnicliffe et al., the confounding influence of LIC on water T1 can also be minimised. The studies summarized above have investigated changes in T1 and T2* arising predominantly from changes in liver iron or fat. However, these parameters are also sensitive to changes in tissue water content, which may result from inflammation orfibrosis in the liver. For example, Luetkens et al ( ‘Quantification of liver fibrosis at T1and T2 mapping with extracellular volume fraction MRI: preclinicalresults.’ Radiology 288.3 (2018): 748-754) have described how fibrosis and portalhypertension in the liver might increase the water content of the extracellular matrix, and lead to an increase in the tissue T1 and T2. Hamilton et al. (‘Relation between T1 and T2 of liver water and fat and Proton Density Fat Fraction estimated by a flip angle corrected multi-TR, multi-TE single breath-hold1H MRS STEAM sequence.’ Proc. Intl. Soc. Mag. Reson. Med. (2019)) examined therelationship between the T1 and T2 relaxation times of liver water and fat, in adult subjects with known or suspected non-alcoholic fatty liver disease. The use of a multi-TR, multi-TE 1H MRS sequence allowed the T1 and T2 of the liver water and fat signalsto be measured independently. They found a significant linear correlation between liver water T1 and water T2, which is likely to reflect the combined response of these parameters to variability in the underlying liver water concentration within this cohort. Thompson et al. have presented a novel imaging method known as PROFIT (PDFF imaging with T1), which provides simultaneous quantification of PDFF, T2*, and water T1 in the liver (Thompson et al., ‘Simultaneous proton density fat-fraction and R2* imaging with water-specific T1 mapping (PROFIT1): application in liver’, Magn Reson Med 2021; 75:223-239). Their method addresses fat-related liver T1 biases by separation of water and fat signals, and also provides T2* and PDFF maps in a single scan. Water T1 values obtained with this method remain sensitive to both LIC and LWC. In a subsequent paper, the authors suggest using a T2*-compensation approach to account for normal physiologic variations which jointly affect water T1 and T2* and lead to increased variability in water T1 (Thompson et al., ‘Reference Values for Water- Specific T1 of the Liver at 3 T: T2*-Compensation and the Confounding Effects of Fat’, J. Magn. Reson. Imaging 2024). This resulted in reduced variability in liver water T1 in a cohort of healthy subjects. However, this approach does not differentiate between the competing influences ofLWC and LIC on water T1 and T2* and could potentially eliminate clinically-relevantpathologic signals that affect both T1 and T2*, such as increased tissue water content as a result of fibrosis or inflammation. Furthermore, the reduction in T2* due to liver fat is not accounted for, and as such, even with water-specific T1 mapping, a reduced T2* due to a high fat content would give rise to an increase in T2*-compensated water T1. In summary, such studies confirm that liver T1 and T2* are sensitive to the fat, water, and iron concentration in the tissue, and these can have opposing effects on measured T1 and T2* values. While some steps have been made to mitigate these confounding effects, such as measuring water-specific T1 to remove the influence of fat, or using T2* to apply iron- correction to the liver T1, there is currently no means to model the simultaneous, combined effects of iron, water and fat content on both water R1 and R2*. The present invention presents such a means. Summary of the Invention According to a first aspect of the invention there is a method of liver imaging using MRIas disclosed in claim 1. Further features of the invention are disclosed in dependentclaims. The present invention provides a means to determine the simultaneous and combined effects that iron, fat, and water concentration have on liver water T1 and T2* values. According to an aspect of the invention there is a method of liver imaging using MRI including: receiving inputs derived from MRI scans of a liver including a map of atransverse relaxation rate R2*, a map of PDFF, and raw MOLLI T1 mapping sequencesignal data Sraw ; for a main magnetic field value B0 of the MRI scans, presuming a firstrelation wherein a longitudinal relaxation rate R1 is linearly dependent on a liver waterconcentration LWC and separately linearly dependent on a liver iron concentration LIC of the first liver; for the main magnetic field value, presuming a second relation whereina fat corrected transverse relaxation rate R2*fat_corr that accounts for an influence of fatin the first liver is linearly dependent on the liver water concentration LWC and separately linearly dependent on the liver iron concentration LIC of the first liver; using the inputs, the first and second linear relations, and known correlations of MRI parameters derived from other MRI scans of other livers or internal body organs determined prior to the MRI scans of the first liver to determine quantitative values of:a first ratio ^R2* / PDFF of change in the transverse relaxation rate R2* per change in thePDFF; a difference ^PDFF that is a difference in a measured PDFF minus zero PDFF; a change in relaxation rate ^R2*fat that is a difference R2* due to fat relative to zero PDFF; a longitudinal water relaxation rate wR1; a relaxation rate wR1ref that is a reference water relaxation rate R1 of a healthy liver; a relaxation rate correction ^R2* is a compensator for the LWC; a relaxation rate R2*fat_corr is a transverse relaxation ratethat is compensated for influence of fat in the liver on the transverse relaxation rateR2*; a relaxation time R2*wc that is corrected for the LWC; a liver iron concentrationLICwc that is water corrected to be unconfounded by the LWC; a relaxation rate R2WC is a transverse relaxation rate that is associated with a value of the LICwc unconfounded by the LWC; a relaxation rate correction ^R2 is a compensator for the LWC; and summing the relaxation rate R2WC plus the relaxation rate correction ^R2 to determinea transverse water relaxation rate R2WCIC that accounts for influence of the LWC andthe LIC. Here R2WCIC is the reciprocal of the relaxation time T2 in equation 12below. The detailed description below provides a detailed explanation of stage 1 inputs, and method of calculation in equations 3, 4 and 12. The method enables the obtaining of an estimate for liver R2 without direct measurement of this parameter. The estimate is free of the confounding effects of LWC and LIC.According to another aspect of the invention there is a method of liver imaging usingMRI including:receiving inputs derived from MRI scans of a first liver including a map of a second starrelaxation rate R2*, a map of PDFF, and raw MOLLI T1 mapping sequence signal data Sraw ; for a main magnetic field value Bo of the MRI scans, presuming a first relation whereina first relaxation rate R1 is linearly dependent on a liver water concentration LWC andseparately linearly dependent on a liver iron concentration LIC of the first liver;for the main magnetic field value Bo of the MRI scans, presuming a second relationwherein a fat corrected second relaxation rate R2*fat_corr that accounts for influence fatin the first liver is linearly dependent on the liver water concentration LWC andseparately linearly dependent on the liver iron concentration LIC of the first liver;using: known correlations of MRI parameters derived from other MRI scans of otherlivers or internal body organs determined prior to the MRI scans of the first liver, andthe inputs, and the first and second linear relations to determine quantitative values ofat least: a relaxation rate R2WC and relaxation rate correction deltaR2, and summing the relaxation rate R2WC plus the relaxation rate correction deltaR2 to determine a second water relaxation rate R2WCIC that accounts for influence of the LWC and the LIC.Further relationships and quantitative values from the field of MRI imaging and whichmay be used in determining the second water relaxation rate R2WCIC may bedetermined by reference to measurements and procedures given in the detaileddescription. Such relationships or quantitative values may include but are notnecessarily limited to or by: a first ratio ^R2* / PDFF of change in the transverse relaxationrate R2* per change in the PDFF; a difference ^PDFF that is a difference in a measuredPDFF minus zero PDFF; a change in relaxation rate ^R2*fat that is a difference R2* dueto fat relative to zero PDFF; a longitudinal water relaxation rate wR1; a relaxation ratewR1ref that is a reference water relaxation rate R1 of a healthy liver; a relaxation ratecorrection ^R2* is a compensator for the LWC; a relaxation rate R2*fat_corr is a transverserelaxation rate that is compensated for influence of fat of the liver on the transverserelaxation rate R2*; a relaxation time R2*wc that is corrected for the LWC; a liver ironconcentration LICwc that is water corrected to be unconfounded by the LWC; arelaxation rate R2WC is a transverse relaxation rate that is associated with a value ofthe LICwc unconfounded by the LWC; a relaxation rate correction ^R2 is a compensatorfor the LWC. The method provides a means to estimate liver T2, for a given value of water T1 and T2*, without the need of a separate MRI acquisition to measure T2 directly. This is advantageous, as R2-based relaxometry methods generally require long acquisition times (10–20 minutes), provide limited spatial coverage of the liver, and are prone to motion artifacts. Conversely, R2*-based measurements generally provide good spatial coverage of the liver, and acquisition within a single short breath-hold is feasible (Hernando et al (‘Multicenter Reproducibility of Liver Iron Quantification with 1.5-T and 3.0-T MRI’, Radiology 306, e213256 (2023)).Preferably the method of liver imaging includes finding a fit between the raw MOLLI T1mapping sequence signal data Sraw and simulated MOLLI signals Ssim as a function ofthe longitudinal water relaxation rate wR1 and a transverse water relaxation rateR2WCIC that accounts for influence of the LWC and the LIC to determine a longitudinalrelaxation rate wR1that is independent of fat. The detailed description below providesa detailed explanation of stage 4 signal matching. The method helps by obtaining anestimate for liver R2 without direct measurement of this parameter, the latter being used to accurately simulate the MOLLI raw signal, in order to estimate liver water T1.The liver water T1 is independent of fat. Preferably the method of liver imaging includes using: a ratio ^R1 / LIC from the firstrelation as presumed that relates a change in the longitudinal relaxation rate R1 to achange in the LIC; a difference ^LIC of the LIC minus a healthy liver iron concentrationreference LICref, and determining a product of the longitudinal relaxation rate wR1noFatthat is independent of fat times an amount that is a difference of one minus a relaxationtime wT1noFat corresponding to the longitudinal relaxation rate wR1noFat that isindependent of fat times the ratio ^R1 / LIC times the difference ^LIC to determine a fatindependent and iron corrected water relaxation rate wR1Fe-corr. The detaileddescription below provides a detailed explanation of calculation by equations 14 and 15. The method helps to estimate a corrected liver water T1 value, in which the confounding influences of iron, fat, magnetic field strength and inter-vendor variability on liver T1 are minimised.Preferably the method of liver imaging is conducted wherein a constant offset plus aproduct of the fat independent and iron corrected water relaxation time wT1Fe-corr is multiplied times a constant coefficient, and the constant offset and coefficient are determined from the MRI scans of the first liver at the main magnetic field value B0 and second MRI scans at a second main magnetic field value B0(2) to quantify an effect of magnetic field on the fat independent and iron corrected water relaxation time wT1Fe-corr. The detailed description below provides a detailed explanation of calculation byequation 16. The method helps because the influence of LIC and LWC on liver R1,R2* and R2 is able to be estimated.Preferably the method of liver imaging is conducted wherein the relaxation ratecompensators and ^R2* and ^R2 are both directly proportional to a difference of thelongitudinal relaxation rate of water wR1 minus the relaxation rate wR1ref, and thelongitudinal relaxation rate wR1 is a reciprocal of a longitudinal relaxation time waterT1 generated by a Bloch simulation of the raw MOLLI T1 mapping sequence signaldata Sraw , and the relaxation rate wR1ref is a reciprocal of a healthy liver referencewater relaxation time T1. The detailed description below provides a detailedexplanation of calculation by equations 9 and 11.Preferably the method of liver imaging is conducted to determine the relaxation rateR2WC as a function of the main magnetic field level of the MRI scan and the liver iron concentration LICwc that is water corrected to be unconfounded by the waterconcentration LWC. The detailed description below provides a detailed explanation in stage 3(b)(C) LIC to T2 Estimation.Preferably the method of liver imaging is conducted wherein the reciprocal of therelaxation time R2*wc that is corrected for the LWC is a relaxation time T2*wc, thatdetermines the liver iron concentration LICwc using a Hernando model which provideslinear regression models relating the transverse relaxation rate R2* to the LIC by usingvalues of the relaxation time T2*wc. The detailed description below provides adetailed explanation in stage 3(b)(B) LIC estimation.Each of the water corrected values of the LIC may be converted to correspondingrelaxation time T2 which is then adjusted for the influence of the LWC on the relaxationtime T2, as shown below in equation 11, to provide an estimate of a relaxation timeT2WCIC for a corresponding value of the relaxation time T1; thereby preparing for a best- fit liver water T1 relaxation time to be estimated via a Bloch simulation of the raw MOLLI T1 mapping sequence signal data. The detailed description below provides a detailed explanation in stage 3(b)(C) LIC to T2 estimation and in stage 3(b)(D) liver-water-adjusted T2 estimation. Preferably the method of liver imaging is conducted whereineach of the values of relaxation time T2*wc that is water corrected is associated with a relaxation time water T1noFat generated by a Bloch simulation of the raw MOLLI T1 mapping sequence signal data. The detailed description below provides a detailed explanation in stage 4 signal mapping.Preferably the method of liver imaging is conducted to determine the relaxation timeT2*wc as a reciprocal of a difference of the relaxation rate R2*fat_corr minus a relaxationrate correction ^R2*. The detailed description below provides a detailed explanationof calculation by equation 10. Use of the method helps to avoid erroneously attributingchanges in R2* arising from either an elevated fat or water concentration to a change in liver iron concentration.Preferably the method of liver imaging is conducted to determine the change inrelaxation rate ^R2*fat as directly proportional to both the first ratio ^R2* / PDFF and the ^PDFF. The detailed description below provides a detailed explanation of calculation by equation 1.Preferably the method of liver imaging is conducted to determine the transverserelaxation rate R2* as the reciprocal of a transverse relaxation time T2* within a regionof interest ROI in the liver, wherein the MRI scan is performed to collect data to formulate a relaxation time T2* map made visible in a display on which the ROI is placed, wherein the relaxation time T2* is influenced by fat in the liver, the LWC, andthe LIC. The detailed description below provides a detailed explanation in stage 1inputs.Preferably the method of liver imaging is conducted wherein the first relation, thelongitudinal relaxation rate R1 linear dependency on the LWC is scaled by a first ‘a1’constant coefficient that is equal to a change in the longitudinal relaxation rate R1 perchange in the LWC for a constant LIC; and the second relation, the fat correctedtransverse relaxation rate R2*fat_corr linear dependency on the LWC is scaled by asecond ‘a2’ constant coefficient that is equal to a change in the fat corrected transverserelaxation rate R2*fat_corr per change in the LWC for a constant LIC; whereby acombined variability in the fat corrected transverse relaxation rate R2*fat_corr per changein the longitudinal relaxation rate R1 due to variability in the LWC is determined as aratio of the second ‘a2’ constant coefficient to the first ‘a1’ constant coefficient, and theratio is determined from a linear correlation between a longitudinal water relaxationrate and a transverse water relaxation rate determined from MRI scans of subjects inwhom the LIC is normal with known or suspected non-alcoholic fatty liver disease.The detailed description below provides a detailed explanation of calculation byequations 3, 4, 5c, and 7.Preferably the method of liver imaging is conducted wherein the longitudinal relaxationrate R1 linear dependency on the LIC is scaled by a first ‘b1’ constant coefficient thatis equal to a change in the longitudinal relaxation rate R1 per change in the LIC whenthe LWC is constant; and the fat corrected transverse relaxation rate R2*fat_corr lineardependency on the LIC is scaled by a second ‘b2’ constant coefficient that is equal toa change in the transverse relaxation rate R2*fat_corr per change in the LIC when theLWC is constant; whereby a combined variability in the fat corrected transverserelaxation rate R2*fat_corr per change in the longitudinal relaxation rate R1 due tovariability in the LIC is determined as a ratio of the second ‘b2’ constant coefficient to the first ‘b1’ constant coefficient, and the first ‘b1’ constant coefficient is determined from MRI scan of subjects in whom iron concentration is elevated above normal and second ‘b2’ constant coefficient is determined from a Hernando model that relates therelaxation rate R2*fat_corr to LIC. The detailed description below provides a detailedexplanation of calculation by equations 3, 4, 6a, 6b, 6c, and 8.Preferably the method of liver imaging is conducted by adjusting a combined variabilityin the fat corrected transverse relaxation rate R2*fat_corr per change in the longitudinalrelaxation rate R1 to correlate with a relationship between the main magnetic fieldvalue and the water relaxation rates R2*fat_corr , R1 acquired from MRI scans at aplurality of magnetic field levels. The detailed description below provides a detailedexplanation of calculation by equations 7 and 8 and illustration in Figure 2.Preferably the method of liver imaging includes determining from an MRI scan of atleast one healthy subject, a water longitudinal relaxation rate (R1) reference valuewR1ref at the particular main magnetic field value.According to another aspect of the invention there is a system of liver imaging toexecute the method of liver imaging, the system comprising a computer, an inputreceiving device by which the computer receives the inputs derived from the MRI scans of the first liver, a data receiving device by which the computer receives the known correlations of MRI parameters derived from other MRI scans, a display by which thecomputer shows the maps of transverse relaxation rate R2*, PDFF, or transverse waterrelaxation rate R2WCIC that accounts for influence of the LWC and the LIC, and a computer program directly loadable into a memory of the computer for controlling the method when the program is run on the computer. According to another aspect of the invention there is an MRI scanner configured to acquire MRI scans of a liver wherein the MRI scanner is in electronic communication with a system of liver imaging to execute the method of liver imaging, the systemcomprising a computer, an input receiving device by which the computer receives fromthe MRI scanner the inputs derived from the MRI scans of the first liver, a data receivingdevice by which the computer receives the known correlations of MRI parameters derived from other MRI scans, a display by which the computer shows the maps oftransverse relaxation rate R2*, PDFF, or transverse water relaxation rate R2WCIC thataccounts for influence of the LWC and the LIC, and a computer program directly loadable into a memory of the computer for controlling the method when the program is run on the computer.The invention will now be described, by way of example only, with reference to theaccompanying figures in which: Brief Description of the Figures Figure 1 shows an overview of the processing steps employed in the present invention; Figure 2 shows plots of the modelled change in liver water R1 and R2*, as a result ofchanges in the liver water concentration (LWC, dashed line), and liver ironconcentration (LIC, dashed line), at (A) 1.5T, and (B) 3T; and the reference, healthy-liver water R1 and R2* values for 1.5T and 3T are represented by black data points; Figure 3 shows a schematic of the processing steps used to estimate an associated water T2 value, for each value in the range of water T1 values; and these water T1 and T2 values are used to build the dictionary of synthetic MOLLI raw signals; and Figure 4 shows the values for the vendor-specific water T1 correction terms (^wT1vendor). Detailed Description of the Invention In an embodiment, the method and system of MRI Imaging to estimate liver water R1and R2* variability comprises the following stages:1. Inputs 2. T2* Fat-Correction 3. Generation of Simulated MOLLI Signals: (a) LIC-LWC relaxation rate i) estimating the influence of LWC on R1 and R2* ii) estimating the influence of LIC on R1 and R2* iii) healthy-liver reference water R1, LIC, and R2* values, (b) estimation of water T2 and(c) generating a dictionary of simulated MOLLI signals, 4. Signal Matching 5. Water T1 Iron Correction, and6. Water T1 Standardisation : (a) water T1 field strength correction; b) waterT1 vendor correction.An explanation for each stage is provided below.1. InputsProcessed T2* and PDFF maps and raw MOLLI signal data are inputs. Regions ofinterest (ROIs) are placed on the T2* and PDFF maps to calculate a representative T2* and PDFF value for the liver (defined as the median value across all voxels within the ROIs). Using these as inputs, the MOLLI raw images are processed on a voxel-by- voxel basis. 2. T2* Fat-Correction The measured T2* value is corrected for the known influence of fat on T2*. The variation in R2* due to liver fat content can be expressed as a linear relationship between ^R2*fat (change in R2* due to fat, relative to 0% PDFF) and ^PDFF (the difference in the measured PDFF relative to 0% PDFF), as follows: where ^R2* / PDFF relates the change in R2* (Hz) to a change in PDFF (%). ^R2* / PDFF isestimated, for example, using data disclosed by Wang et al (‘Monte Carlo Modelling ofLiver MR Signal in the Presence of Fat’, Proc. Intl. Soc. Mag. Reson. Med.26 (2018), whereby ^R2* / PDFF is 0.772 Hz / % at 3T and 0.395 Hz / % at 1.5T. A fat-corrected T2* is thus calculated (T2*fat-corr), which represents the measured liver T2* without the influence of fat: 3. Generation of Simulated MOLLI Signals Bloch simulations of the acquired MOLLI sequence are performed, for example, using the Bloch simulation methodology described in Mozes et al. (Mozes et al. “Mapping tissue water T1 in the liver using the MOLLI T1 method in the presence of fat, iron and B0 inhomogeneity”, NMR in Biomedicine 2018; e4030) and Tunnicliffe et al. (Tunnicliffe et al., "A model for hepatic fibrosis: the competing effects of cell loss and iron on shortened modified Look-Locker inversion recovery T1 (shMOLLI-T1) in the liver." Journal of Magnetic Resonance Imaging 45.2 (2017): 450-462). A dictionary of MOLLI signals is generated, by first simulating the raw MOLLI signal from the water and fat compartments of the liver independently. For the fat compartment, a multi-peak spectral model is used, for example, the six- peak spectral model described by Hamilton et al. (‘In vivo characterization of the liverfat ¹H MR spectrum,’ NMR Biomed. 2011 Aug;24(7):784-90). Literature values for theT1 and T2 of fat are used: at 3T, fat T1 = 312ms, and fat T2 = 62ms (Hamilton et al.): fat T1 at 1.5T = 253ms, which is estimated by applying a field-strength conversion to the3T value, derived using the data in Rakow-Penner et. al., (‘Relaxation times of breasttissue at 1.5T and 3T measured using IDEAL’, JMRI 23:87–91 (2006)); as the T2 of fat is equivalent at 1.5T and 3T (Rakow-Penner at al.), fat T2 = 62ms at 1.5T, to match the 3T value. For the water compartment, a vector of water T1 values (wT1range) is used, ranging from 400ms to 3000ms. A dictionary is generated by simulating the signal from the water compartment for each of the water T1 values in wT1range. To perform these Bloch simulations, a value of water T2 must be assigned to each value in wT1range, which accounts for the combined influence of LIC and LWC on both water T1 and water T2. These steps comprise: LIC-LWC relaxation rate estimation and estimating the influence of LWC on R1 and R2* (a) LIC-LWC Relaxation Rate Estimation The LIC-LWC relaxation rate estimation stage describes how water R1 and R2* values in the liver jointly-respond to a change in either the iron or water content in the tissue. The R2* values used here correspond to the inverse of the T2*fat-corr value, therefore the additional influence of fat on R2* has already been accounted for. The estimation assumes that the combined effect of LWC and LIC on water R1 and R2* can be expressed follows: ^^ = ^^ ∙ ^^^ + ^^ ∙ ^^^ + ^^(3) ^∗^ = ^^ ∙ ^^^ + ^^ ∙ ^^^ + ^^(4) where the influence of LWC on R1 and R2* is represented by a1 and a2, and the influence of LIC on R1 and R2* is represented by b1 and b2, respectively. c1 and c2 are constants. For a constant LIC, changes in R1 (dR1) and R2* (dR2*) as a result of changes in LWC (^LWC) can be expressed as: and the combined variability in R2* and R1, due to a variability in LWC, can be expressed as: Similarly, for a constant LWC, changes in R1 and R2* as a result of changes in LIC can be expressed as: ^^ ^^=^(6a) ^^^^ and the combined variability in R2* and R1, due to a variability in LIC, can be expressed as: i) Estimating the Influence of LWC on R1 and R2* To estimate the (^R2* / dR1)dLWC term in equation 5c, we can use, for example, the data presented in Hamilton et al. (‘Relation between T1 and T2 of liver water and fat and Proton Density Fat Fraction estimated by a flip angle corrected multi-TR, multi-TE single breath-hold 1H MRS STEAM sequence.’ Proc. Intl. Soc. Mag. Reson. Med. (2019)). Here, the authors show a linear correlation between water R1 and water R2 in a cohort of adults with known or suspected non-alcoholic fatty liver disease. Water R2 contributes to R2* (specifically, R2* = R2 + R2’), and if R2’ is constant (for instance, when the influence of iron is small), then ^R2 = ^R2*. We assume LWC is the primary influence on the variability in liver R1 and R2 values in the data presented in Hamilton et al. (the mean water T2 of 24.7ms at 3T (standard deviation = 3.6ms) suggests LIC is not elevated in these subjects). As such, using these data, (dR2* / dR1)dLWC in equation 5c can be estimated as 37.1 at 3T. By applying a field strength conversion to water R1 (for example via the data in de Bazelaire et al, ‘MR imaging relaxation times of abdominal and pelvic tissuesmeasured in vivo at 3.0 T: preliminary results’, Radiology 230, 652–659 (2004)) andwater R2 (for example, via the method of Ghugre et al, ‘Multi-field behaviour ofRelaxivity in an Iron-rich environment’, in Proc. Intl. Soc. Mag. Reson. Med. (2008)),the (^R2* / dR1)dLWC term in equation 5c can be estimated as 18.2 at 1.5T. Thus: 18.2 (1.5^)(7) 37.1 (3^)ii) Estimating the Influence of LIC on R1 and R2* When LIC is the primary influence on water R1 and R2*, such as in subjects with elevated iron, the b1 term in equation 6a can be estimated as, for example, 0.073 s-1 / mgFe g-1, using the data presented in Wood et al. (‘Cardiac Iron Determines Cardiac T2*, T2, and T1 in the Gerbil Model of Iron Cardiomyopathy’, Circulation 112, 535–543 (2005)). The b2 value in equation 6b can be estimated using, for example, the model in Hernando et al. (‘Multicenter Reproducibility of Liver Iron Quantification with 1.5-T and 3.0-T MRI’. Radiology 306, e213256 (2023)), which gives b2 = 0.026-1(1.5T) and 0.013-1(3T) s-1 / mg Fe g-1. Using equation 6c, we can combine our estimates of b1 and b2, to obtain (^R2* / dR1)dLIC. This gives: ^527 (1.5^)(8) 1054 (3^)iii) Healthy-Liver Reference Water R1, LIC, and R2* Values A plot of the modelled change in liver water R1 and R2*, resulting from changes In LWC and LIC, is shown in Figure 2. The changes in R1 and R2* in this plot are relative tobaseline, healthy-liver water R1 and R2* reference values (represented by the blackdata points where the dotted line and dashed lines cross). Determination of healthy-liver reference water R1, LIC, and R2* values is an illustration of the effects mentionedand not necessarily an essential element of the method of liver imaging.By way of illustrative example, for water R1, a healthy-liver reference value was derived using in-house data, comprising 83 imaging studies in healthy subjects, acquired at 1.5T (N=38) and 3T (N=45). The median, un-corrected water T1 values across these healthy subjects, at 1.5T and 3T, were used as healthy-liver reference values. These were wT1ref = 650ms (wR1ref = 1.538 s-1) at 1.5T and wT1ref = 910ms (wR1ref = 1.099 s-1) at 3T. The 3T wT1ref value is similar to the mean healthy-liver water T1 value presented in Thompson et al. (952ms), which was obtained in 124 healthy adult subjects imaged at 3T (‘Reference Values for Water-Specific T1 of the Liver at 3 T: T2*- Compensation and the Confounding Effects of Fat’, J. Magn. Reson. Imaging 2024).To obtain a healthy-liver reference T2* value, T2* measurements acquired in cohorts ofhealthy subjects, as presented in Obrzut et al. (and references therein) were used (‘Value of liver iron concentration in healthy volunteers assessed by MRI’, Scientific Reports volume 10, Article number: 17887 (2020)). This provided healthy-liver T2* values of 29.2ms (R2*ref = 34.19 s-1) at 1.5T and 17.7ms (R2*ref = 56.33 s-1) at 3T.To obtain a healthy-liver reference LIC value, the R2*ref values described above wereconverted to an equivalent LIC value, using the model presented by Hernando et al.(Multicenter Reproducibility of Liver Iron Quantification with 1.5-T and 3.0-T MRI. Radiology 306, e213256 (2023), referred to henceforth as the ‘Hernando model’). This resulted in a reference healthy-liver LIC value (LICref) of 0.73 mg / g iron dry weight. As shown in Figure 2, using wR1ref and R2*ref as a starting point, when LWC changes (but LIC is constant), the method predicts a small change in R2* relative to a given change water R1 ((dR2* / dR1)dLWC = 18.2 at 1.5T, 37.1 at 3T). However, when LIC changes (but LWC is constant), a much larger change in R2* relative to a given changein water R1 is expected ((dR2* / dR1) dLIC = 527 at 1.5T, 1054 at 3T).(b) Estimation of Water T2 The LIC-LWC relaxation rate estimation method is applied to estimate an associated water T2 value for each water T1 value in wT1range. With reference to the schematic shown in Figure 3, water T2 is estimated via the following steps:(A) T2* water correctionIn this step, the influence of LWC on T2* is minimised. The T2*fat-corr value and the range of water T1 values (wT1range) are the inputs. Differences between each water T1 value in wT1range and the healthy-liver reference water T1 values (wT1ref, see above) areassumed to result from differences in LWC. As such, using the LIC-LWC relaxationrate method (equation 7), an R2* correction term (^R2*) is calculated for each value in wT1range, to account for predicted influence of LWC on R2*: where wR1 represents a given value from the set of values in wT1range. This provides aset of ‘water-corrected’ T2* values (T2*wc), which are given by: An upper / lower limit of 60ms / 5ms (1.5T) and 35ms / 2.7ms (3T) is applied to T2*wc, to maintain water-corrected T2* values within a plausible range. This provides an estimate of T2* in which the expected variability due to changes in LWC are minimised.(B) LIC EstimationEach water T1 value is now associated with a value of T2*wc. The next step is to convert each T2*wc value into an estimate of LIC. This is performed using the Hernando model, which provides linear regression models relating R2* to LIC for magnetic field strengths of 1.5T, 2.89T, and 3T. By using water-corrected T2* values, a value of LIC is estimated which is not confounded by the competing influence of LWC on R2*. These LIC values are referred to as ‘water-corrected LIC’ (LICwc).(C) LIC to T2 EstimationA liver T2 value for each value of LICWC is estimated, for example, using the model described by St. Pierre et al. (‘Noninvasive measurement and imaging of liver ironconcentrations using proton magnetic resonance.’ Blood 105.2 (2005): 855-861). Thisprovides an estimate of liver T2 at 1.5T only, and as such, for data acquired at 3T, a field-strength conversion of the T2 value (to 3T) is performed using the modelpresented in Ghugre et al. (Multi-field behaviour of Relaxivity in an Iron-richenvironment. in Proc. Intl. Soc. Mag. Reson. Med. (2008)). These T2 values do notinclude the additional influence of LWC on T2, and as such are referred to as ‘water- corrected T2’ (T2wc).(D) Liver-Water-Adjusted T2 EstimationThe estimated T2WC values are associated with each value of LICwc, and correspond solely to the influence of LIC on T2, as the additional influence of LWC has been taken out (as described in steps (A) to (C)). The T2WC values are adjusted, based on the expected influence of LWC on T2, so that the final T2 estimation reflects both LWC and LIC. This effectively reverses the ‘water correction’ applied in step (A), however, we are now adjusting T2, rather than T2* values. This correction is applied as follows: The time T2 in equation 12 may also be referred to as time T2WCIC to indicate that theeffects of the LWC and the LIC have been accounted for.An upper / lower limit of 100ms / 5ms is applied to the liver-water-adjusted T2 valuesestimated using equation 12 to maintain water-corrected T2* values within a plausiblerange. The processing steps described in (A) to (D) allow a liver water T2 value to be estimated for each value of water T1, in which the influences of LWC and LIC on T2 are independently accounted for. (c) Generating the Dictionary of Simulated MOLLI Signals For each pair of water T1 and water T2 values in wT1range, a Bloch simulation of the acquired MOLLI sequence is performed, to generate a simulated MOLLI signal for the water compartment. This water signal is then combined with the simulated fat signal as follows: ^= (1 − ^^^^) ∙ ^^^^^^ + ^^^^ ∙ ^^^^ (13)where Swater is the simulated signal from the water compartment, Sfat is the simulated signal from the fat compartment, and PDFF is the proton density fat fraction, expressed as a fraction between 0 (0% fat) to 1.0 (100% fat). The dictionary comprises a set of simulated MOLLI signals (one for each pair of waterT1 and T2 values), representing the raw signal intensity over the series of inversion times used in the MOLLI acquisition. 4. Signal Matching The measured MOLLI signal is then compared to the simulated signals in the dictionary, by computing the vector dot-product between the measured signal and each simulated signal. The entry in the dictionary which provides the maximum dot-product is selected as the best fit to the measured data, and this entry is used to obtain the best-fit water T1 value for the measured MOLLI signal. This provides a measure of liverT1 which relates to the water compartment and is therefore independent of fat.5. Water T1 Iron Correction The best-fit water T1 value is then corrected for the known influence of iron on liver T1. The variation in R1 due to LIC (^R1,LIC) can be expressed as a linear relationship between ^R1,LIC and ^LIC, in which ^LIC is the difference between the measured LIC (LICmeasured) and the healthy-liver reference LIC value (LICref = 0.73 mg / g iron dry weight), as follows: where ^LIC = LICmeasured – LICref, and bR1 / LIC relates a change in R1 (Hz) to a change inLIC (mg / g dry weight). For this iron-correction step, a whole-liver LICmeasured value is used, by converting the fat-corrected T2* value to an equivalent LIC value, using the Hernando model. Iron correction is only applied if LICmeasured > LICref (i.e. the measured liver iron is above the healthy-liver reference value). As such, we do not apply water-correction to the T2* value used here, as LICmeasured is above the threshold for which we consider iron to be the dominant influence on T2*. bR1 / LIC in equation 14 is equal to the b1 term used in the LIC-LWC relaxation rate model (equation 6a), which is estimated, for example, using data disclosed by Wood et al. (‘Cardiac Iron Determines Cardiac T2*, T2, and T1 in the Gerbil Model of IronCardiomyopathy.’ Circulation 112, 535–543 (2005)), which gives bR1 / LIC = 0.073 Hz / (mgFe / g). Iron-corrected water T1(wT1FE-corr) is then calculated as follows: where wT1 is the best-fit (uncorrected) water T1 value. 6. Water T1 Standardisation (a) Water T1 Field Strength Correction Iron-corrected water T1 values are then corrected for the known effect of magnetic field strength on T1 relaxation times. By way of illustrative example, a field-strength correction model is used, derived using data acquired in volunteers, who were scanned consecutively at both 1.5T and 3T. Studies were performed across different scannermanufacturers (Siemens, Philips, and GE).Values of wT1FE-corr, obtained using the steps described above were compared at 1.5T and 3T for these subjects. Using these data, a linear model was derived which allows T1 values measured at 1.5T to be ‘field strength corrected’ to an equivalent 3T value: ^^^^^^^^^^ = 0.0072 where wT1FS-corr is the field-strength corrected water T1 value, and wT1,1.5T is the water T1 value measured at 1.5T (both expressed in seconds). (b) Water T1 Vendor Correction Variability in water T1 due to minor differences in the MOLLI sequence used acrossdifferent scanner manufacturers (e.g. Siemens, Philips, and GE) are then corrected. Inan illustrative example, to determine this correction, the same cohort used for the field- strength correction were used. Subjects were included if they had been scanned on a given test scanner (Siemens 1.5T, Philips 1.5T / 3T and GE 1.5T / 3T) as well as thereference scanner (Siemens 3T). Iron- and field-strength corrected waterT1 valueswere obtained for each subject, on both the test and reference scanner, and thedifference between these was recorded. The median difference in corrected waterT1(comparing each test scanner against the reference scanner) was used to determine the scanner-specific bias terms. These bias terms are applied to the wT1FS-corr values, to provide a corrected water T1 value (cwT1) which is standardised to the reference scanner (Siemens 3T), as follows: ^^^^ = ^^^^^^^^^^ + ∆^^^^^^^^^ (17)The vendor-specific correction terms (DwT1vendor) are given in Figure 4. This provides the final, corrected water T1 value. The invention has been described by way of examples only. Therefore, the foregoing is considered as illustrative only of the principles of the invention. Further, since numerous modifications and changes will readily occur to those skilled in the art, it is not desired to limit the invention to the exact construction and operation shown and described, and accordingly, all suitable modifications and equivalents may be resorted to, falling within the scope of the claims.

Claims

Claims:

1. A method of liver imaging using MRI including receiving inputs derived from MRI scans of a first liver including a map of a transverserelaxation rate R2*, a map of PDFF, and raw MOLLI T1 mapping sequence signal data Sraw,for a main magnetic field value B0 of the MRI scans, presuming a first relation wherein alongitudinal relaxation rate R1 is linearly dependent on a liver water concentrationLWC and separately linearly dependent on a liver iron concentration LIC of the firstliver, for the main magnetic field value, presuming a second relation wherein a fat corrected transverse relaxation rate R2*fat_corr that accounts for influence fat in thefirst liver is linearly dependent on the liver water concentration LWC and separatelylinearly dependent on the liver iron concentration LIC of the first liver,using the inputs, the first and second linear relations, and known correlations of MRI parameters derived from other MRI scans of other livers or internal body organsdetermined prior to the MRI scans of the first liver to determine quantitative values of: afirst ratio ^R2* / PDFF of change in the transverse relaxation rate R2* per change in thePDFF; a difference ^PDFF that is a difference in a measured PDFF minus zeroPDFF; a change in relaxation rate ^R2*fat that is a difference R2* due to fat relative tozero PDFF; a longitudinal water relaxation rate wR1; a relaxation rate wR1ref that is areference water relaxation rate R1 of a healthy liver; a relaxation rate correction ^R2*is a compensator for the LWC; a relaxation rate R2*fat_corr is a transverse relaxationrate that is compensated for influence of fat of the liver on the transverse relaxationrate R2*; a relaxation time R2*wc that is corrected for the LWC; a liver ironconcentration LICwc that is water corrected to be unconfounded by the LWC; a relaxation rate R2WC is a transverse relaxation rate that is associated with a value ofthe LICwc unconfounded by the LWC; a relaxation rate correction ^R2 is acompensator for the LWC; and summing the relaxation rate R2WC plus the relaxation rate correction ^R2 to determinea transverse water relaxation rate R2WCIC that accounts for influence of the LWC andthe LIC.

2. A method of liver imaging according to claim 1 finding a fit between the raw MOLLI T1mapping sequence signal data Sraw and simulated MOLLI signals Ssim as a functionof the longitudinal water relaxation rate wR1 and a transverse water relaxation rateR2WCIC that accounts for influence of the LWC and the LIC to determine a longitudinalrelaxation rate wR1noFat that is independent of fat.

3. A method of liver imaging according to claim 2 using: a ratio ^R1 / LIC from the first relationas presumed that relates a change in the longitudinal relaxation rate R1 to a changein the LIC; a difference ^LIC of the LIC minus a healthy liver iron concentrationreference LICref, and determining a product of the longitudinal relaxation rate wR1noFatthat is independent of fat times an amount that is a difference of one minus arelaxation time wT1noFat corresponding to the longitudinal relaxation rate wR1noFat thatis independent of fat times the ratio ^R1 / LIC times the difference ^LIC to determine afat independent and iron corrected water relaxation rate wR1Fe-corr.

4. A method of liver imaging according to claim 3 wherein a constant offset plus a product of the fat independent and iron corrected water relaxation time wT1Fe-corr is multipliedtimes a constant coefficient, and the constant offset and coefficient are determinedfrom the from the MRI scans of the first liver at the main magnetic field value B0 andsecond MRI scans at a second main magnetic field value B0_2 to quantify an effectof level main magnetic field on the fat independent and iron corrected waterrelaxation time wT1Fe-corr.

5. A method of liver imaging according to any preceding claim wherein the relaxationrate compensators and ^R2* and ^R2 are both directly proportional to a differenceof the longitudinal relaxation rate of water wR1 minus the relaxation rate wR1ref, andthe longitudinal relaxation rate wR1 is a reciprocal of a longitudinal relaxation timewater T1 generated by a Bloch simulation of the raw MOLLI T1 mapping sequencesignal data Sraw , and the relaxation rate wR1ref is a reciprocal of a healthy liverreference water relaxation time T1.

6. A method of liver imaging according to any preceding claim that determines therelaxation rate R2WC as a function of the main magnetic field level of the MRI scan and the liver iron concentration LICwc that is water corrected to be unconfounded bythe water concentration LWC.

7. A method of liver imaging according to any preceding claim wherein the reciprocal ofthe relaxation time R2*wc that is corrected for the LWC is a relaxation time T2*wc,that determines the liver iron concentration LICwc using a Hernando model which provides linear regression models relating the transverse relaxation rate R2* to theLIC by using values of the relaxation time T2*wc.

8. A method of liver imaging according to claim 7 wherein each of the water correctedvalues of the LICwc are converted to corresponding relaxation time T2 which is thenadjusted for the influence of the LWC on the relaxation time T2 to provide an estimateof a relaxation time T2WCIC for a corresponding value of the relaxation time T1;thereby preparing for a best-fit liver water T1 relaxation time to be estimated via aBloch simulation of the raw MOLLI T1 mapping sequence signal data.

9. A method of liver imaging according to claim 7 wherein each of the values of relaxation time T2*wc that is water corrected is associated with a relaxation time water T1noFatgenerated by a Bloch simulation of the raw MOLLI T1 mapping sequence signal data.

10. A method of live imaging according to claim 7, 8 or 9 that determines the relaxationtime T2*wc as a reciprocal of a difference of the relaxation rate R2*fat_corr minus arelaxation rate correction ^R2*.

11. A method of liver imaging according to any preceding claim that determines thechange in relaxation rate ^R2*fat as directly proportional to both the first ratio ^R2* / PDFFand ^PDFF.

12. A method of liver imaging according to any preceding claim that determines thetransverse relaxation rate R2* as the reciprocal of a transverse relaxation time T2*within a region of interest ROI in the liver, wherein the MRI scan is performed to collect data to formulate a relaxation time T2* map made visible in a display on which the ROI is placed, wherein the relaxation time T2* is influenced by fat in the liver, theLWC, and the LIC.

13. A method of liver imaging according to any preceding claim whereinthe first relation, the longitudinal relaxation rate R1 linear dependency on the LWC isscaled by a first ‘a1’ constant coefficient that is equal to a change in the longitudinalrelaxation rate R1 per change in the LWC for a constant LIC; and the second relation, the fat corrected transverse relaxation rate R2*fat_corr lineardependency on the LWC is scaled by a second ‘a2’ constant coefficient that is equalto a change in the fat corrected transverse relaxation rate R2*fat_corr per change in theLWC for a constant LIC; whereby a combined variability in the fat corrected transverse relaxation rateR2*fat_corr per change in the longitudinal relaxation rate R1 due to variability in theLWC is determined as a ratio of the second ‘a2’ constant coefficient to the first ‘a1’constant coefficient, and the ratio is determined from a linear correlation between a longitudinal water relaxation rate and a transverse water relaxation rate determinedfrom MRI scans of subjects in whom the LIC is normal with known or suspected non- alcoholic fatty liver disease.

14. A method of liver imaging according to any preceding claim whereinthe longitudinal relaxation rate R1 linear dependency on the LIC is scaled by a first‘b1’ constant coefficient that is equal to a change in the longitudinal relaxation rateR1 per change in the LIC as the LWC is constant; and the fat corrected transverse relaxation rate R2*fat_corr linear dependency on the LIC isscaled by a second ‘b2’ constant coefficient that is equal to a change in the transverse relaxation rate R2*fat_corr per change in the LIC as the LWC is constant;whereby a combined variability in the fat corrected transverse relaxation rateR2*fat_corr per change in the longitudinal relaxation rate R1 due to variability in the LICis determined as a ratio of the second ‘b2’ constant coefficient to the first ‘b1’ constant coefficient, and the first ‘b1’ constant coefficient is determined from MRI scans ofsubjects in whom iron concentration is elevated above normal and second ‘b2’ constant coefficient is determined from a Hernando model that relates the relaxation rate R2*fat_corr to LIC.

15. A method of liver imaging according to any preceding claim adjusting a combinedvariability in the fat corrected transverse relaxation rate R2*fat_corr per change in thelongitudinal relaxation rate R1 to correlate with a relationship between the mainmagnetic field value and the water relaxation rates R2*fat_corr , R1 acquired from MRIscans at a plurality of magnetic field levels.

16. A method of liver imaging according to any preceding claim determining from an MRIscan of at least one healthy subject, a water longitudinal relaxation rate (R1)reference value wR1ref at the main magnetic field value.

17. A system of liver imaging to execute the method according to any of claims 1 to 15comprising a computer, an input receiving device by which the computer receivesthe inputs derived from the MRI scans of the first liver, a data receiving device by which the computer receives the known correlations of MRI parameters derived from other MRI scans, a display by which the computer shows the maps of transverse relaxation rate R2*, PDFF, or transverse water relaxation rate R2WCIC that accountsfor influence of the LWC and the LIC, and a computer program directly loadable into a memory of the computer for controlling the method when the program is run on the computer.

17. An MRI scanner configured to acquire MRI scans of a liver wherein the MRI scanner is in electronic communication with a system of liver imaging to execute the method according to any of claims 1 to 15, the system comprising a computer, an inputreceiving device by which the computer receives from the MRI scanner the inputsderived from the MRI scans of the first liver, a data receiving device by which thecomputer receives the known correlations of MRI parameters derived from other MRI scans, a display by which the computer shows the maps of transverse relaxation rateR2*, PDFF, or transverse water relaxation rate R2WCIC that accounts for influence ofthe LWC and the LIC, and a computer program directly loadable into a memory of the computer for controlling the method when the program is run on the computer.

Images