Nonalcoholic fatty liver disease (NAFLD) is the most prevalent hepatic disease, affecting 20–30% of the population in the developed world, and it is considered a major public health concern 1,2. It is associated with insulin resistance, type 2 diabetes mellitus, hyperlipidemia 3,4, and cardiovascular disease 5–9. Until recently, simple hepatic steatosis was usually associated with a benign prognosis, but new evidence has shown its potential progression to fibrosis and cirrhosis 10,11. Only 50% of NAFLD patients show elevated levels of serum transaminases 12, and most of them are asymptomatic 13; therefore, a considerable percentage of these patients remain undiagnosed until the disease becomes severe. For all of these reasons, the identification of steatosis is essential for the successful management of NAFLD patients 2,3.
Current gold standard methods for the detection and quantification of liver fat are histopathological liver biopsy analysis and proton magnetic resonance spectroscopy (1H-MRS) 14,15. However, liver biopsy study is associated with high financial costs, sampling bias 16, interobserver variability 17, a morbidity rate of 3%, and a mortality rate of 0.03% 18. In contrast, 1H-MRS is a noninvasive method that directly measures the chemical composition of a tissue, on the basis of the assumption that each tissue shows a specific precessional frequency that can be measured in a voxel 19–24. Nevertheless, it is also expensive and is not in widespread clinical use. Hence, there is a need for a less invasive and more cost-effective technique to measure liver fat content accurately 10.
The ultrasound (US) diagnosis of hepatic steatosis is based on increased liver parenchymal echogenicity. The two most common conditions associated with fatty liver are alcoholic liver disease and NAFLD. In addition to alcohol abuse, various etiologic factors have been associated with fatty liver, including diabetes, hepatitis, and drug toxicity, among others. The US quantification of liver fat content is a simple, inexpensive, and noninvasive approach; its main disadvantages include the high intraobserver and interobserver variability 25, the low sensitivity to mild steatosis 26, and the qualitative and semiquantitative nature of this method. Reported sensitivity and specificity values for US are 60–100% and 77–95%, respectively 27.
However, various authors have reported that the liver fat content in individuals with known hepatic steatosis can be estimated from the ratio between the echogenicity of the liver and that of the right renal parenchyma, achieving an accuracy similar to that of histopathological analysis 28 and 1H-MRS 29,30. Other authors found that US analyses based solely on a qualitative and/or a semiquantitative approach did not predict the presence or severity of hepatic steatosis in severely obese adolescents 31.
The objective of this study was to determine whether a Sonographic Hepato-Renal Index (SHRI) computed using a standard workstation is an adequate alternative to 1H-MRS 3T, the current gold standard, for the quantification of liver fat content in healthy volunteers.
Participants were consecutively recruited between November 2010 and May 2011 from three medical centers in Granada province (Southern Spain): Virgen de la Nieves University Hospital of Granada, San Cecilio University Hospital of Granada, and the Occupational Risk Prevention Unit of the Andalusian Regional Government. Potential participants were selected from among men and women aged above 18 years visiting the centers for a routine general checkup.
Out of the total of 834 adults who came for a routine checkup during the recruitment period, 584 were excluded after application of the following exclusion criteria: the presence of renal disease, daily alcohol intake ≥20 g (men) or ≥10 g (women), positive hepatitis serologic markers, the consumption of drugs that may cause steatosis (e.g. tamoxifen, amiodarone), BMI <17 or ≥40 kg/m2, the presence of a pacemaker or other device not compatible with 1H-MRS, and/or claustrophobia. Of the 250 patients fulfilling the eligibility criteria, 130 (52%) agreed to participate in the study and signed informed consent, but 19 of these were excluded for claustrophobia or for missing appointments for the tests. Therefore, the final sample comprised 121 individuals (mean age=46 years, range=21–77 years). The study was approved by the ethics committees of the participating centers.
Anthropometric and biochemical variables
Participants’ height and weight were measured, and their BMI was calculated as weight/height squared (kg/m2). Waist circumference was measured with a soft tape on standing participants midway between the lowest rib and the iliac crest. Biochemical tests included liver enzymes (aspartate aminotransferase, alanine aminotransferase, γ-glutamyltransferase, and alkaline phosphatase), serum lipid profile (total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, and triglycerides), and plasma glucose. All analyses were carried out at the San Cecilio University Hospital of Granada (Granada, Spain).
US studies were performed by an experienced radiologist blinded to the patients’ clinical details and laboratory findings. US examinations were performed using an Acuson Antares system (Siemens, Munich, Germany) equipped with a curved phased-array transducer (CH4: 1.55–2.22 MHz). All instrument settings, including ‘gain’, ‘depth’, and ‘time-gain compensation’, were established for each measurement. For estimation of the SHRI, US images of the liver and right kidney were obtained in the sagittal liver/right kidney view in the lateral position (Fig. 1). All images were transmitted to a Picture Archiving and Communication System (PACS; Fujifilm Synapse, Tokyo, Japan).
From the sagittal liver/right kidney view, a circular region of interest (ROI) of 100 mm2 was selected for the renal parenchyma (uniform, excluding blood vessels, bile ducts, and other focal hypoechogenicity/hyperechogenicity) and an ROI of 33 mm2 was selected for the right renal cortex (excluding large vessels, sinus, and medulla). The boundary between the liver and the right kidney area was situated near the center of the image to avoid the interference of depth-dependent echo intensity attenuation and borderline echo-distorting effects. The liver and right kidney ROIs were selected at the same depth, and the mean gray-scale value of the pixels in each ROI was used to determine its echogenicity (Fig. 1). SHRI was calculated as the ratio of the echogenicity of the liver to the echogenicity of the right kidney parenchyma. This procedure was repeated with two new ROIs on the same scan. If the difference between the values obtained was less than 0.20, the arithmetic mean of the two values was chosen; if it was at least 0.20, a third measurement was performed and the mean of the two closest values was used.
The reproducibility of the test was evaluated previously in a sample of 15 volunteers who underwent two repeated measurements of SHRI (with a 1-week interval) by two independent operators, each blinded to the results of the other. The results were compared with Bland–Altman statistics and the intraclass correlation coefficient and showed a high correlation (r=0.87, P<0.001).
1H-MRS 3T analyses
Before the spectroscopy was performed, magnetic resonance imaging and in-vivo spectra were acquired at 3T using a Philips Achieva system (Royal Philips, Amsterdam, the Netherlands). Three plane localizers were used to plan the 1H-MRS, and the spectra were obtained using the body coil of the scanner. Breath holding was monitored using a respiratory belt.
A single voxel of 27 cm3 (30×30×30 mm) was selected within normal liver tissue in segment VI (where the SHRI was calculated), avoiding the edge of the liver, the diaphragm, and major blood vessels. All spectra were obtained with a stimulated echo acquisition mode sequence (STEAM), setting the following parameters: repetition time=8000; echo time=20, 40, and 60 ms; number of signal averages=4 (without water suppression); and bandwidth=2000. Data acquisition was performed within a breath hold. A T2 correction was applied. Field homogeneity was adjusted automatically for each voxel.
MRS was reconstructed using Extended MR WorkSpace software (Philips). Raw data were zero-filled once with no filter. Data were phase corrected, Fourier transformed, baseline corrected, and averaged. Marquardt curve fitting was performed using a combined Lorentzian–Gaussian model to calculate the area under the curve of fat and water peaks. Spectra were referenced to residual water and the dominant methylene lipid (–CH2) peak at δ=4.47 and δ=1.43 ppm, respectively. Fat fraction percentage (FF) was defined as FA/(FA+WA)×100, where FA is the area under the fat peak and WA is the area under the water peak. 1H-MRS data were interpreted by an experienced radiologist blinded to the US results.
Steatosis was defined as liver fat content greater than 5%, as proposed in previous studies 15,32, and was further classified as mild steatosis (>5 to 25% liver fat content), moderate steatosis (>25 to 50% liver fat content), or severe steatosis (>50% liver fat content) 26.
All anthropometric, biochemical, US, and 1H-MRS measurements were performed within a 24-h period.
Means with their SDs and percentages were calculated for the descriptive analyses. The distribution of continuous variables was assessed using the Shapiro–Wilk and Kolmogorov–Smirnov tests and normal probability plots, which showed that the SHRI and 1H-MRS 3T variables had a non-normal distribution, and nonparametric tests were therefore applied for the bivariate analyses. The Mann–Whitney U-test was applied for the comparison of levels of continuous variables in different categories and the Spearman correlation test was used for the assessment of a linear correlation between continuous variables.
Post-hoc comparisons with the Bonferroni test, Bland–Altman statistics, and intraclass correlation coefficient were performed to evaluate the reproducibility of the methodology.
Receiver operating characteristic (ROC) curves were created, and optimal cutoff points were estimated for the diagnosis of the different levels of steatosis. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and likelihood ratios were calculated.
P-values less than 0.05 were considered statistically significant, and all tests were two tailed. Data were stored in a database and managed using SPSS v.15.0 (IBM, Chicago, Illinois, USA).
Table 1 summarizes the main characteristics of the 71 (58.7%) men and 50 (41.3%) women in the study sample; the age range was 21–77 years.
The mean (SD) liver fat content was quantified as 23.47% (21.66) (range=0.05–73.40%) by 1H-MRS 3T and as 1.81 (0.72) (range=0.96–3.34) by SHRI. Of the 75 participants (62%) diagnosed with liver steatosis, 22 had mild steatosis, 32 had moderate steatosis, and 21 had severe steatosis (Table 1).
The SHRI was higher in participants with steatosis than in those without [mean values (SD)=2.23 (0.61) and 1.13 (0.09), respectively, P<0.001]. In addition, multiple comparison tests showed a significant difference (P<0.001) in the SHRI among the steatosis categories (Fig. 2). As shown in Fig. 3, a significant correlation was found between SHRI and 1H-MRS 3T values (Spearman’s coefficient=0.89, P<0.001). The presence of steatosis was around 20-fold more likely in participants with SHRI greater than 1.28 than in those with SHRI less than 1.28 (likelihood ratio=21.77) (Table 2).
Three ROC curves were created to assess the validity of SHRI for the different levels of steatosis (>5, >25, and >50% liver fat content) (Table 2 and Fig. 4). The optimal SHRI cutoff points were 1.28 for the estimation of steatosis greater than 5% [sensitivity=94.67%, specificity=95.65%, PPV=97.26%, NPV=91,67%, and area under the ROC curve (AUC)=99.1%]; 1.75 for steatosis greater than 25% (sensitivity=90.6%; specificity=91.2%, PPV=91.18%, NPV=92.54%, and AUC=96.6%), and 2.29 for steatosis greater than 50% (sensitivity=95.2%, specificity=84.0%, PPV=58.56%, NPV=98.82%, and AUC=95.9%).
The quantitative SHRI proposed in this study enabled the identification of changes in hepatic echogenicity and the detection of liver fat content as low as 5%. The results obtained were strongly correlated with the quantification of liver fat content by 1H-MRS 3T. SHRI therefore appears to be an excellent alternative to 1H-MRS for grading hepatic steatosis.
Other methods that use magnetic resonance for assessment of liver steatosis include gradient echo (GRE) chemical shift, turbo spin echo (TSE), and multiecho gradient echo (me-GRE) water–fat separation, a promising new technique 20. 1H-MRS was selected for the present study because it is currently considered the most accurate noninvasive procedure for the quantification of liver fat content 14,15,19. 1H-MRS is a reference technique rather than histological liver biopsy analysis, the other gold standard 32, because it has been widely validated in population-based studies and clinical trials 24,32–35 and is a noninvasive method that poses no health risk to the patient, unlike a liver biopsy 18. 1H-MRS 3T analyses were carried out with a STEAM sequence. Although point-resolved spectroscopy sequence (PRESS) shows a better signal-to-noise ratio, several authors have proposed utilization of the STEAM sequence because it is less sensitive to the effect of J coupling 36.
The study of the diagnostic accuracy of SHRI to predict liver fat content greater than 5% (upper limit of normal range) showed a sensitivity of 94.67% and a specificity of 95.65% for an SHRI value of 1.28. SHRI also showed an excellent capacity to distinguish between the different degrees of steatosis. Thus, an SHRI value of 1.75 yielded a sensitivity of 90.6% and a specificity of 91.25% for the diagnosis of moderate steatosis (liver fat content >25%), and an SHRI value of 2.29 yielded a sensitivity of 95.2% and a specificity of 84% for the diagnosis of severe steatosis (liver fat content >50%).
Furthermore, a reduction in the SHRI cutoff value from 1.28 to 1.21 yielded a sensitivity of 100% for the detection of cases with liver fat deposit greater than 5%, with a specificity of 80.4%. Similarly, a reduction in the SHRI cutoff from 1.75 to 1.28 yielded a sensitivity of 100% and a specificity of 70.6% for the detection of moderate steatosis (liver fat deposit >25%), and a reduction from 2.29 to 2.15 yielded a sensitivity of 100% and a specificity of 80% for the diagnosis of severe steatosis (>50% liver fat deposit) (Table 2). These findings suggest that SHRI may be useful as a screening test to identify individuals at risk of hepatic steatosis, allowing the appropriate clinical recommendations to be made, and to select candidates for further testing by other methods.
However, the PPV and NPV depend strongly on the prevalence of the disease, and caution should be exercised in comparing with results in different populations. The prevalence of hepatic steatosis in our population was 62%, considerably higher than the estimate of 30% for the general population 1,3. This discrepancy might be explained by the elevated proportion of our series with obesity (76%) and with moderately high levels of transaminases (32%). Nevertheless, the excellent sensitivity and specificity results suggest that SHRI might be a valid instrument for predicting the degree of liver steatosis and that these findings should be corroborated in population-based studies.
The mean SHRI obtained in the present series was 1.81, ranging from 0.96 through 3.34. Comparable results, ranging from 1.65 through 2.25, have been reported with the use of other indexes on the basis of the relative echogenicity values of the liver and right kidney parenchyma. Nevertheless, comparisons with each previous study show small differences that may be explained by the distinct characteristics of the study populations. Thus, Webb et al. 28 only studied individuals with known liver disease other than NAFLD, whereas the majority of the participants in the study by Xia et al. 30 had mild NAFLD, and all of the participants investigated by Mancini et al. 29 were obese (BMI≥30 kg/m2) type 2 diabetics. We also highlight that the cutoff value for the diagnosis of steatosis (liver fat content ≥5%) was 1.28 in the present study but was found to be 1.49 by Webb et al. 28 and 2.20 by Mancini et al. 29, again likely attributable to differences among the study populations or in the US methods used. We also highlight that the cutoff value for the diagnosis of steatosis (liver fat content ≥5%) was 1.28 in the present study but was found to be 1.49 by Webb et al. 28 and 2.20 by Mancini et al. 29, again likely attributable to differences among the study populations or the differences in the US methods used. Xia et al. 30 performed a different treatment of the data, preventing a direct comparison, but the results were similar to those of the present investigation, as was the study population (82% with steatosis). These findings confirm that this ratio may provide a useful diagnostic tool and warrants further investigation in wider samples, given that studies to date have been carried out in small populations.
Three levels of steatosis are typically defined: mild steatosis (more hyperechogenic liver tissue in comparison with the adjacent kidney cortex), moderate steatosis (moderate and diffuse increase in liver echo intensity with decreased beam penetration associated with a decrease in visualization of silhouetting of the portal vein borders), and severe steatosis (marked increase in echo intensity with no visualization of portal vein borders, obscured diaphragm, and posterior portion of the right lobe) 25,27. Very few authors 37 have preferred the experience of the observer to computerized quantitative US for the assessment of hepatic steatosis. The former has been criticized for its high intraobserver and interobserver variability 25,26,28–30 and is only considered useful for diagnosing steatosis with greater than 30% liver fat content 38, although it is still widely used in clinical practice. In fact, a recently published paper reported a lack of accuracy in the prediction of hepatic steatosis using only qualitative/semiquantitative US analyses in severely obese adolescents 31.
A potential limitation of our study is that measurements were obtained using one ROI for 1H-MRS and the average of two ROIs for the SHRI. The distribution of liver fat deposits is of major importance in the diagnosis of steatosis. Focal or patchy steatosis can lead to an erroneous diagnosis when only one ROI is measured 30. Hence, it is recommended to average results from different areas at standardized locations, and the SHRI represents a cost-effective instrument for performing these calculations.
In addition, excessive hepatic accumulation of iron or other substances (e.g. water) from necroinflammatory activity may also be present and may represent a confounding factor in the US and/or 1H-MRS quantification of fat liver content 39,40. In the present study, however, this problem was minimized by applying a T2 correction in the 1H-MRS analyses. A further potential limitation of our study is that spectroscopy rather than histology was used as the reference technique. As stated in the Introduction section, pathologic examination of biopsied liver specimens was considered ethically unacceptable for our investigation as an invasive procedure associated with risks of morbidity and mortality 18. Histology study results have been found to correlate well with 1H-MRS hepatic triglyceride content 15. However, several clinical trials 33–35 on NAFLD have used spectroscopic magnetic resonance as an outcome measure, supporting its selection as the reference method for accurate assessment of fat content in the present study.
This study shows that our SHRI is a valid, simple, reliable, and cost-effective screening tool for the identification, assessment, estimation, and quantification of liver fat deposits and for the diagnosis of hepatic steatosis. In fact, the total time from data acquisition to index calculation is very similar to that required for conventional US analysis. Further cross-validation studies with larger and more heterogeneous population groups are warranted to evaluate the inclusion of this technique in routine screening programs. In addition, further longitudinal studies are needed to test the usefulness of SHRI in the follow-up of steatosis patients and in the evaluation of their treatment.
Dr J.P. Arrebola is under contract with the PTA-MICINN program (Spanish Ministry of Science and Innovation). The authors thank Richard Davies for editorial assistance and Centro de Diagnóstico (Granada) for providing the technical equipment. They are indebted to all of the patients who participated in the study, without whom this work would not have been possible.
Conflicts of interest
There are no conflicts of interest.
1. Clark JM, Brancati FL, Diehl AM.Nonalcoholic fatty liver disease
2. Wieckowska A, Feldstein AE.Nonalcoholic fatty liver disease
in the pediatric population: a review.Curr Opin Pediatr2005;17:636–641.
3. Angulo P.Nonalcoholic fatty liver disease
.N Engl J Med2002;346:1221–1231.
4. Marchesini G, Brizi M, Morselli-Labate AM, Bianchi G, Bugianesi E, McCullough AJ, et al..Association of nonalcoholic fatty liver disease
with insulin resistance.Am J Med1999;107:450–455.
5. Hamaguchi M, Kojima T, Takeda N, Nagata C, Takeda J, Sarui H, et al..Nonalcoholic fatty liver disease
is a novel predictor of cardiovascular disease.World J Gastroenterol2007;13:1579–1584.
6. Wong VW, Wong GL, Yip GW, Lo AO, Limquiaco J, Chu WC, et al..Coronary artery disease and cardiovascular outcomes in patients with non-alcoholic fatty liver disease.Gut2011;60:1721–1727.
7. Abdou AS, Magour GM, Mahmoud MM.Evaluation of some markers of subclinical atherosclerosis in Egyptian young adult males with abdominal obesity.Br J Biomed Sci2009;66:143–147.
8. Ndumele CE, Nasir K, Conceicao RD, Carvalho JA, Blumenthal RS, Santos RD.Hepatic steatosis
, obesity, and the metabolic syndrome are independently and additively associated with increased systemic inflammation.Arterioscler Thromb Vasc Biol2011;31:1927–1932.
9. Scorletti E, Calder PC, Byrne CD.Non-alcoholic fatty liver disease and cardiovascular risk: metabolic aspects and novel treatments.Endocrine2011;40:332–343.
10. Mehta SR.Non-invasive means of measuring hepatic fat content.World J Gastroenterol2008;14:3476–3483.
11. Powell EE, Jonsson JR, Clouston AD.Steatosis: co-factor in other liver diseases.Hepatology2005;42:5–13.
12. Dassanayake AS, Kasturiratne A, Rajindrajith S, Kalubowila U, Chakrawarthi S, De Silva AP.Prevalence and risk factors for non-alcoholic fatty liver disease among adults in an urban Sri Lankan population.J Gastroenterol Hepatol2009;24:1284–1288.
13. Syn WK, Nightingale P, Bateman JM.Nonalcoholic fatty liver disease
in a district general hospital: clinical presentation and risk factors.Hepatol Int2008;2:190–195.
14. Van Werven JR, Hoogduin JM, Nederveen AJ, van Vliet AA, Wajs E, Vandenberk P.Reproducibility of 3.0 Tesla magnetic resonance spectroscopy
for measuring hepatic fat content.J Magn Reson Imaging2009;30:444–448.
15. Roldán-Valadez E, Favila R, Martínez-Lopez M, Uribe M, Rios C, Méndez-Sánchez N.In vivo 3T spectroscopic quantification of liver fat content in nonalcoholic fatty liver disease
: correlation with biochemical method and morphometry.J Hepatol2010;53:732–737.
16. Ratziu V, Charlotte F, Heurtier A, Gombert S, Giral P, Bruckert E, et al..Sampling variability of liver biopsy in nonalcoholic fatty liver disease
17. Bravo AA, Sheth SG, Chopra S.Liver biopsy.N Engl J Med2001;344:495–500.
18. Gilmore IT, Burroughs A, Murray-Lyon IM, Williams R, Jenkins D, Hopkins A.Indications, methods, and outcomes of percutaneous liver biopsy in England and Wales: an audit by the British Society of Gastroenterology and the Royal College of Physicians of London.Gut1995;36:437–441.
19. Bohte AE, van Werven JR, Bipat S, Stoker J.The diagnostic accuracy of US, CT, MRI and 1
H-MRS for the evaluation of hepatic steatosis
compared with liver biopsy: a meta-analysis.Eur Radiol2011;21:87–97.
20. Cassidy FH, Yokoo T, Aganovic L, Hanna RF, Bydder M, Middleton MS, et al..Fatty liver disease: MR imaging techniques for the detection and quantification of liver steatosis.Radiographics2009;29:231–260.
21. Chang JS, Taouli B, Salibi N, Hecht EM, Chin DG, Lee VS.Opposed-phase MRI for fat quantification in fat-water phantoms with 1
H MR spectroscopy to resolve ambiguity of fat or water dominance.Am J Roentgenol2006;187:103–106.
22. Longo R, Pollesello P, Ricci C, Masutti F, Kvam BJ, Bercich L, et al..Proton MR spectroscopy in quantitative in vivo determination of fat content in human liver steatosis.J Magn Reson Imaging1995;5:281–285.
23. Mazhar SM, Shiehmorteza M, Sirlin CB.Noninvasive assessment of hepatic steatosis
.Clin Gastroenterol Hepatol2009;7:135–140.
24. Thomsen C, Becker U, Winkler K, Christoffersen P, Jensen M, Henriksen O.Quantification of liver fat using magnetic resonance spectroscopy
.Magn Reson Imaging1994;12:487–495.
25. Strauss S, Gavish E, Gottlieb P, Katsnelson L.Interobserver and intraobserver variability in the sonographic assessment of fatty liver.Am J Roentgenol2007;189:320–323.
26. Dasarathy S, Dasarathy J, Khiyami A, Joseph R, Lopez R, McCullough AJ.Validity of real time ultrasound
in the diagnosis of hepatic steatosis
: a prospective study.J Hepatol2009;51:1061–1067.
27. Hamer OW, Aguirre DA, Casola G, Lavine JE, Woenckhauss M, Sirlin CB.Fatty liver: imaging patterns and pitfalls.Radiographics2006;26:1637–1653.
28. Webb M, Yeshua H, Zelber-Sagi S, Santo E, Brazowski E, Halpern Z, et al..Diagnostic value of a computerized hepatorenal index for sonographic quantification of liver steatosis.Am J Roentgenol2009;192:909–914.
29. Mancini M, Prinster A, Annuzzi G, Liuzzi R, Giacco R, Medagli C.Sonographic hepatic-renal ratio as indicator of hepatic steatosis
: comparison with 1
H magnetic resonance spectroscopy
30. Xia MF, Yan HM, He WY, Li XM, Li CL, Yao XZ, et al..Standardized ultrasound
hepatic/renal ratio and hepatic attenuation rate to quantify liver fat content: an improvement method.Obesity (Silver Spring)2012;20:444–452.
31. Bohte AE, Koot BG, van der Baan-Slootweg OH, van Werven JR, Bipat S, Nederveen AJ, et al..US cannot be used to predict the presence or severity of hepatic steatosis
in severely obese adolescents.Radiology2012;262:327–334.
32. Szczepaniak LS, Nurenberg P, Leonard D, Browning JD, Reingold JS, Grundy S, et al..Magnetic resonance spectroscopy
to measure hepatic triglyceride content: prevalence of hepatic steatosis
in the general population.Am J Physiol Endocrinol Metab2005;288:462–468.
33. McPherson S, Jonsson JR, Cowin GJ, O’Rourke P, Clouston AD, Volp A, et al..Magnetic resonance imaging and spectroscopy accurately estimate the severity of steatosis provided the stage of fibrosis is considered.J Hepatol2009;51:389–397.
34. Belfort R, Harrison SA, Brown K, Darland C, Finch J, Hardies J, et al..A placebo-controlled trial of pioglitazone in subjects with nonalcoholic steatohepatitis.N Engl J Med2006;355:2297–2307.
35. Hockings PD, Changani KK, Saeed N, Reid DG, Birmingham J, O’Brien P, et al..Rapid reversal of hepatic steatosis
, and reduction of muscle triglyceride, by rosiglitazone: MRI/S studies in Zucker fatty rats.Diabetes Obes Metab2003;5:234–243.
36. Hamilton G, Middleton MS, Bydder M, Yokoo T, Schwimmer JB, Kono Y, et al..Effect of PRESS and STEAM sequences on magnetic resonance spectroscopic liver fat quantification.J Magn Reson Imaging2009;30:145–152.
37. Vehmas T, Kaukiainem A, Luoma K, Lohman M, Nurminen M, Taskinen H.Liver echogenicity: measurement or visual grading?Comput Med Imaging Graph2004;28:289293.
38. Saadeh S, Younossi ZM, Remer EM, Gramlich T, Ong JP, Hurley M, et al..The utility of radiological imaging in nonalcoholic fatty liver disease
39. Martí-Bonmatí L, Alberich-Bayarri A, Sánchez-Órtega J.Overload hepatitides: quanti-qualitative analysis.Abdom Imaging2012;37:180–187.
40. Hayasi N, Miyati T, Minami T, Takeshita Y, Ryu Y, Matsuda T, et al..Quantitative analysis of hepatic fraction by single-breath-holding MR spectroscopy wits T2 correction: phantom and clinical with histologic assessment.Radiol Phys Technol2013;6:219–225.