Access to the PDF text

Free Article !

Journal de radiologie
Vol 85, N° 3  - mars 2004
pp. 301-306
Doi : JRE-01-2004-85-3-0221-0363-101019-ART01
Materials and methods
© Masson, Paris 2004

Diagnosis and quantification of hepatic fibrosis with diffusion weighted MR imaging: preliminary results

C Aubé [1] PX Racineux [1] J Lebigot [1] F Oberti [2] V Croquet [2] [3] C Argaud [4] P Calès [2] C Caron [1]

[1] Département de radiologie,

[2] Service d'hépatogastroentérologie, CHU Angers

[3] Service d'hépato-Gastroentérologie, CH Nord deux Sèvres, Thouars

[4] General Electric Medical System.

To cite the present paper, use exclusively the following reference. Aubé C et al. Diagnostic et quantification de la fibrose hépatique par IRM de diffusion : résultats préliminaires. (full text in english on J Radio 2004;85:301-6.

Diagnostic et quantification de la fibrose hépatique par IRM de diffusion : résultats préliminaires

Objectif. Le diagnostic et la quantification de la fibrose hépatique sont des sujets particulièrement importants en hépatologie clinique. La biopsie hépatique en demeure la clé, mais souffre de nombreuses limitations. Le but de cette étude est de définir les possibilités de l'IRM de diffusion pour le diagnostic et la quantification de la cirrhose et de définir les meilleurs paramètres de séquences pour ce faire.

Matériel et méthode. Une IRM de diffusion a été réalisée chez 13 patients cirrhotiques et 14 volontaires présumés sains. Une séquence en apnée de 8 coupes a été réalisée avec 4 facteurs b (200, 400, 600, et 800 sec/mm2) et 2 TR différents (3 500 et 5 000 ms. Le coefficient de diffusion a été calculé (ADCs) pour chaque séquence et corrélé au score de Child Pugh et au taux d'acide hyaluronique sanguin.

Résultats. L'ADCs est statistiquement significativement plus bas chez les patients cirrhotiques (2,055 10–3) que chez les patients contrôles (2,915 10–3) (p < 0,05) en utilisant un facteur b de 200 sec/mm2 et un TR de 5000 ms. Il existait une corrélation significative entre l'ADCs et le score de Child Pugh (p < 0.05) et entre l'ADCs et le taux d'acide hyaluronique sanguin (p < 0,05) quand le facteur b était de 400 sec/mm2 et le TR 5 000 ms.

Conclusion. Notre étude préliminaire montre que l'IRM de diffusion pourrait être capable de diagnostiquer et éventuellement de quantifier la fibrose hépatique, particulièrement en utilisant des valeurs de facteur b égales à 200 sec/mm2 et à 400 sec/mm2.

Foie , Cirrhose , IRM diffusion , Fibrose hépatique ,


Purpose. Diagnosis and quantification of hepatic fibrosis are especially important in patients with chronic liver disease. Liver biopsy remains the gold standard for diagnosis of cirrhosis but has several limitations. The purpose of this study was to determine the usefulness of diffusion-weighted MR imaging, for the diagnosis of cirrhosis and quantification of hepatic fibrosis, and to define the best sequence parameters for this evaluation.

Methods and Materials. Diffusion-weighted imaging using a 1.5 T MR unit was performed in 14 healthy volunteers and 13 cirrhotic patients. Sets of 8 images with different b values (200, 400, 600, and 800 sec/mm2) and different TR (3500 and 5000 ms) were acquired with breath-holding. Apparent diffusion coefficients (ADCs) were calculated. Correlation between Child-Pugh scores, serum hyaluronate concentrations and ADCs were performed.

Results. ADCs were significantly lower in cirrhotic patients (2,055 10–3) compared to controls (2,915 10–3) (p<0.05) when the b value was 200s/mm2 and the TR was 5000ms. Significant correlations were observed between Child-Pugh scores and ADCs (p<0.05), and between serum hyaluronate concentrations and ADCs (p<0.05), when the b value was 400sec/mm2 and the TR was 5000ms.

Conclusion. Our preliminary study showed that the measurement of ADCs has good potential for diagnosis and quantification of hepatic fibrosis, especially when using b values of 200sec/mm2 and 400sec/mm2.

Liver , Cirrhosis , MR diffusion , Hepatic fibrosis ,

Hepatic fibrosis is a typical complication of chronic liver diseases resulting in cirrhosis and its complications: hepatocellular insufficiency, portal hypertension, increased risk of hepatocellular carcinoma with a 5 year survival rate of about 50% [1] .

Because of its prophylactic and therapeutic implications, the evaluation of hepatic fibrosis is a major focus of daily clinical hepatology.

Currently, the diagnosis and quantification of hepatic fibrosis rely on histological analysis of a biopsy sample, most frequently obtained percutaneously. Percutaneous liver biopsy is an invasive procedure with an associated morbidity of 3% and mortality of 0.03% [2] . In addition, the rate of false negative result from percutaneous biopsy is estimated at 24% [3] due to sampling errors and inter-observer variability.

Over the past few years, several markers for hepatic fibrosis were evaluated. Non-invasive markers may be clinical [4] , [5] , biological [5] , [6] , [7] , or sonographic [8] , [9] , [10] . However, these parameters only correspond to an indirect evaluation of fibrosis and not to direct quantification of hepatic fibrosis.

Diffusion weighted MRI sensitizes the MR signal related to the incoherent microscopic motion of hydrogen protons, or water molecules, within a single voxel [11] . Within a diffusion pulse sequence, proton dephasing within a predetermined amount of time, or b factor, occurs. The apparent diffusion coefficient (ADC) corresponds to the difference of signal intensity between a reference MR sequence and the diffusion MR sequence [12] . The measured ADC values are representative of the hydration and metabolic status of the imaged tissues. Therefore, the modifications in water content within the liver parenchyma in patients with fibrosis should thus be measurable with diffusion weighted MR imaging.

The purpose of this preliminary study is to evaluate the potential for diffusion weighted MR imaging (DWI) to diagnose the presence and quantify the degree of hepatic fibrosis and to determine optimal acquisition parameters to enable this evaluation.

Materials and methods


A total of 27 patients were prospectively enrolled into the study. There were 14 presumed healthy volunteers (6 females, 8 males), with a mean age of 33.7 years (range: 25-50 years) and 13 patients with cirrhosis (5 females, 8 males), with a mean age of 56.1 years (range: 40-72 years). The presumed healthy volunteers had no previous history of liver disease or risk factors for liver disease.

Cirrhotic patients were admitted for routine 6 month follow-up (n=7), treatment of esophageal varices (n=1), or acute decompensation of cirrhosis (n=3). For the latter patients, DWI was performed only after treatment and resolution of the acute episode of decompensation. Twelve patients had cirrhosis secondary to alcoholic liver disease and one patient had cirrhosis secondary to hemochromatosis. None of the patients were known to have a hepatocellular carcinoma.

The diagnosis of cirrhosis was confirmed by biopsy in all patients. The biopsy was performed more than 15 days and less than 2 months prior to MR imaging to avoid artifacts related to early post biopsy changes while allowing correlation between imaging findings and biopsy results. Several clinical and laboratory data were collected for cirrhotic patients: presence of ascites or hepatic encephalopathy, bilirubin and albumin level, and prothrombin time. Data required to establish the Child-Pugh score (ascites, hepatic encephalopathy, bilirubin and albumin level and PT) were collected to evaluate the severity of the cirrhosis. Seven patients were Child class A, 5 were Child class B and 1 was Child class C. Five patients had a small (n=3) to moderate (n=2) amount of ascites. Only one patient (Child C) had hepatic encephalopathy. The mean serum hyaluronic acid (HA) level for all cirrhotic patients was 690mg/l (N<90mg/l) with a distribution based on Child's classification as follows: 448mg/l for patients with Child class A cirrhosis, 622mg/l for patients with Child class B cirrhosis, and >1000mg/l for the patient with Child class C cirrhosis. Eleven of thirteen patients with cirrhosis had esophageal varices: 6 with grade I and 5 with grade II varices.


All MR examinations were performed on a 1.5T unit (GE Medical Systems) using the body coil. Spin echo echoplanar DWI was performed using the following acquisition parameters: TE=90ms, TR= 3500ms and 5000ms, matrix=96X128, FOV=36X36, bandwidth=62kHz. Four different b values were tested: 200, 400, 600 and 800 sec/mm2. Signal was initially acquired in a single direction (frequency encoding direction) followed by an acquisition in all 3 spatial directions (phase encoding, frequency encoding and slice select directions). The TR was selected to achieve a compromise between the number of acquired slices and the acquisition time (breathhold acquisition). For all acquisitions, a total of nine 10 mm thick images were obtained with an interslice gap of 5 mm. A total of 16 sequences (9 images/sequence) were acquired for each patient. The acquisition time for each sequence was 7-20 seconds, allowing good quality breathhold acquisitions (table I) .

Calculation of ADC values was performed on a workstation using the GE software functool – ADC. The T2* sequence was used to measure the signal at b=0. The software calculated the ADC value comparing pixel signal values at b=) and b=200, 400, 600 or 800 using the equation: ln (I (b)/I (0))=–ADC*b. Three ROI were placed over the liver within segments V and VI in order to avoid pulsation artifacts from large vessels, the gallbladder and be far enough from the diaphragm. In addition, liver biopsies were performed in these liver segments, further decreasing the risk of sampling errors. From these 3 ROI, the software calculated 3 ADC values and the mean value was used for statistical analysis. Color maps were also provided by the software (fig. 1) .

Statistical Analysis

The calculated ADC values were correlated to the presence or absence of cirrhosis, the serum HA values, and the Child-Pugh score.

Pearson's correlation coefficient and the partial correlation coefficient were used as test methods. The variables isolated during this univariate analysis were tested by a global linear discriminant analysis than step by step. The statistical analysis was performed using the SPSS software.


Signal obtained from sequences with sensitization in a single direction was too low, near 0. These 8 sequences were thus excluded from the statistical analysis. In addition, 2 cirrhotic patients with marked iron deposition within the liver were excluded due to the markedly decreased signal intensity on the T2W images. One of these patients had cirrhosis secondary to alcoholic liver disease and the other patient had cirrhosis secondary to hemochromatosis. There was an age related influence on the presence of cirrhosis due to a selection bias for both groups with the mean age of healthy volunteers at 33.7 years (range: 25-50 years) and the mean age of cirrhotic patients at 56.1 years (range: 40-72 years). The mean calculated ADC values were inversely proportional to the b value and not significantly affected by the repetition time. With the exception of the two sequences obtained with higher b values (600 and 800 sec/mm2) and low TR (3500ms), there was a decrease in calculated ADC values for cirrhotic patients compared to healthy volunteers (table II) (fig. 2) . This difference was statistically significant (p<0.05) for sequences using a TR of 5000 ms and b value of 200 sec/mm2.

The best degree of correlation, and the only one that was statistically significant, between ADC values and HA levels (r=0.71, p<0.05) and Child-Pugh score (r=0.69, p<0.05) was achieved using a TR of 5000 ms and a b value of 400 sec/mm2 (table III) . The presence of a moderate amount of ascites in 5 patients caused no significant modification of calculated ADV values.


Diffusion weighted imaging is related to the diffusion of protons within tissues. In a diffusion sequence, protons dephasing occurs within a predetermined time interval, or b value. The diffusion coefficient corresponds to the difference of signal intensity between a standard MR pulse sequence and a diffusion sequence [12] . Collagen is the main component of hepatic fibrosis [13] . This molecule is not proton-rich and its protons are tightly bound. As such, diffusion in hepatic fibrosis should be reduced and the ADC values decreased within cirrhotic liver parenchyma compared to normal liver parenchyma. Our results confirm this first statement since we have demonstrated that ADC values within cirrhotic liver parenchyma were reduced compared to ADC values within normal liver parenchyma and that the degree of reduction was correlated to the serum HA level, a marker for hepatic fibrosis. Also, there was a correlation between ADC values and the Child-Pugh score, a marker for the severity of the cirrhosis.

Müller et al. [14] have previously reported in 1994 that ADC values in 3 cirrhotic patients were reduced compared to patients with normal liver. These early results have only been evaluated by 3 additional groups [15] , [16] , [17] . All 3 groups confirmed these results by showing the presence of decreased ADC values in cirrhotic livers compared to non-cirrhotic livers, but none correlated to ADC values to a direct or indirect marker for hepatic fibrosis. In a recent publication, Taouli et al. [17] reported that diffusion within the liver parenchyma was isotropic, and reported a significant difference in ADC values between normal and cirrhotic livers even though there was some overlapping in values. Amano et al. [6] also reported that the difference of ADC values between cirrhotic and normal livers was higher with higher b values. However, Amano et al. used relatively low b values (between 3.8 and 383 sec/mm2).

DWI sequences using a low b value were mostly sensitive to parenchymal microperfusion [18] . A higher b value must be used to increase the sensitivity to diffusion and lessen the impact of perfusion. It seems thus logical that improved results would be obtained by using higher b values (200 to 400 sec/mm2 in our study). However, increasing the b value results in decreased signal. Therefore, calculated ADC values are decreased proportionally to the increase in b value [15] . This is further complicated by the fact that the liver has a short T2 [18] , [19] . Because diffusion weighted sequences are T2W sequences, there already is inherent loss of signal. Motion artifacts are an additional significant pitfall of diffusion weighted imaging of liver parenchyma. This requires the use of EPI sequences to reduce the acquisition time and allow breathhold acquisitions [18] , [20] , [21] . Magnetic susceptibility artifacts of EPI sequences are increased by higher b values, further decreasing image quality [18] .

In order to increase signal, we have used a long TR (5000 ms) and a large voxel size since spatial resolution is not mandatory for the evaluation of diffuse liver disease. Differences in calculated ADC values between normal and cirrhotic livers was more important for sequences with long TR value (5000 ms), underscoring the need for optimizing signal acquisition.

One of the main limitations of our study was the inability to use a surface coil for the acquisition of DWI data. The use of the body coil significantly decreases the acquired signal and SNR [16] . The use of surface coils, by increasing the amount of acquired signal, should provide significant results using higher b values (>400sec/mm2). Also, sequences sensitized to diffusion in only one direction could thus provide enough signal hence further reducing the acquisition time. Other limitations to our study include the small number of subjects limiting the ability to achieve statistically significant results, and the absence of quantitative evaluation of the degree of hepatic fibrosis. The use of a reference test such as the measurement of the area of fibrosis by image analysis [22] would allow a more accurate correlation between ADC values and degree of fibrosis.

Hepatic fibrosis is associated with other histopathological processes that may interfere with proton diffusion. Iron accumulation within the liver parenchyma is certainly the most important of these processes and results in magnetic field heterogeneities and marked reduction of T2 relaxation time [16] , [23] . Therefore, patients with liver diseases associated with iron accumulation (hemochromatosis, hemosiderosis) are probably not candidates to DWI of hepatic fibrosis, as demonstrated by two of our patients (fig. 3) . Testing for serum ferritin should probably be obtained before DWI is performed. A threshold serum ferritin value above which diffusion weighted evaluation of the liver is not appropriate should first be proposed.

Fatty infiltration of the liver, frequently present in patients with alcohol related liver disease [24] , and inflammation may increase the number of free protons in the liver parenchyma and have an impact on ADC values. As such, future studies should not only correlate ADC values with the degree of hepatic fibrosis but also with a quantitative score of inflammation and liver steatosis.


The use of diffusion weighted MR imaging to quantify the degree of hepatic fibrosis remains limited by several technical limitations, including the need for high performance MR imagers. In spite of these limitations, our results confirm the reduction of ADC values in cirrhotic patients and the presence of a correlation between serum HA values, a marker for liver fibrosis, and the degree of decrease in ADC values. Further evaluation is needed using high performance MR imagers, a larger patient population, and a correlation with an accurate technique for quantification of liver fibrosis. Additional correlation with other processes associated with liver fibrosis will further define the value and role of diffusion weighted MR imaging as a non-invasive technique to quantify hepatic fibrosis.


[1] Moirand R, Deugnier Y. Prise en charge et surveillance de la cirrhose. Encycl Med Chir, Hepatologie 2000; 7-034-B-12, 8 p.

[2] Piccinino F, Sagnelli E, Pasquale G, Guisti G. Complications following percutaneous liver biopsy; a multicentre retospective study on 68.276 biopsies. J Hepatol 1986;2:165-73.

[3] Nord JH. Biopsy diagnosis of cirrhosis: blind percutaneous versus guided direct vision techniques. A review. Gastrointest Endosc 1982;28:102-4.

[4] Espinoza P, Ducot B, Pelletier G, et al. Interobserver agreement in the physical diagnosis of alcoholic liver disease. Dig Dis Sci 1987;32:244-7.

[5] Tiné F, Caltagirone M, Camma C, et al. Clinical indicants of compensated cirrhosis: a prospective study. In: Dianzani MU, Gentilini P, eds. Chronic liver damage. Amsterdam: Elsevier 1990;187-98.

[6] Oberti F, Valsesia E, Pilette C, et al. Non invasive diagnosis of hepatic fibrosis or cirrhosis. Gastroenterology 1997;113: 1609-16.

[7] Imbert-Bismut F, Ratziu V, Pieroni L, Charlotte F, Benhamou Y, Poynard T. Biochemical markers of liver fibrosis in patients with hepatitis C virus infection: a prospective study. Lancet 2001; 357:1069-75.

[8] Aubé C, Oberti F, Nouri K, et al. Ultrasonographic diagnosis of hepatic fibrosis or cirrhosis. Journal of Hepatology 1999;30: 472-8.

[9] Iwao T, Toyonaga A, Oho K, et al. Value of Doppler ultrasound parameters of portal vein and hepatic artery in the diagnosis of cirrhosis and portal hypertension. Am J Gastroenterol 1997;6: 1012-7.

[10] Arda K, Ofelli M, Calikoglu U, Olcer T, Cumhur T. Hepatic vein doppler waveform changes in early stage (Child-Pugh A) chronic parenchymal liver disease. J Clin Ultrasound 1997;25:15-9.

[11] Le Bihan D, Breton E, Lallemand D, et al. Apports de l'imagerie des mouvements incohérents intra-voxeliens (IVIM) en neuroradiologie. J Neuroradiol. 1987; 14:295-312.

[12] Maki JF, Cofer GP, Johnson GA. Separation of perfusion and diffusion effects in MR imaging. Radiology 1988;169: 155.

[13] Grimaud JA, Druguet M, Peyrol S, Chevalier O, Herbage D, El Badrawy N. Collagen immunotyping in human liver: light and electron microscope study. J Histochem Cytochem 1980;28: 1145-56.

[14] Müller M, Prasad P, Siewert B, Nissenbaum M, Raptopoulos V, Edelman R. Abdominal Diffusion Mapping with Use of a Whole-Body Echo-planar System. Radiology 1994;190:475-8.

[15] Kim T, Murakami T, Takahashi S, Tsuda K, Nakamura H. Diffusion-Weighted Single-Shot Echoplanar MR Imaging for Liver Disease. AJR Am J Roentgenol 1999; 173:393-8.

[16] Amano Y, Kumazaki T, Ishirhara M. Single-Shot diffusion-weighted echo-planar imaging of normal and cirrhotic livers using a phased-array multicoil. Acta Radiologica 1998;39:440-2.

[17] Taouli B, Vilgrain V, Dumont E, Daire JL, Fan B, Menu Y. Evaluation of liver diffusion isotropy and characterisation of focal hepatic lesions with two single-shot echo planar MR imaging sequences: prospective study in 66 patients. Radiology 2003;226:71-8.

[18] Ichikawa T, Haradome H, Hachiya J, Nitatori T, Araki T. Diffusion-weighted MR imaging with single-shot echo-planar imaging in the upper abdomen: preliminary clinical experience in 61 patients. Abdom Imaging 1999;24:456-61.

[19] Le Bihan D, Turner R, Douek P, Patronas N. Diffusion MR Imaging: Clinical Applications. AJR Am J Roentgenol 1992;59:591-9.

[20] Yamashita Y, Tang Y, Takahashi M. Ultrafast MR Imaging of the Abdomen: Echo Planar Imaging and Diffusion-Weighted Imaging. JMRI 1998;8:367-74.

[21] Okada Y, Ohtomo K, Kiryu S, Sasaki Y. Breath-Hold T2-Weighted MRI of Hepatic Tumors: Value of Echo Planar Imaging with Diffusion-Sensitizing Gradient. J Comput Assist Tomogr 1998;22: 364-71.

[22] Pilette C, Rousselet MC, Bedossa P, et al. Histopathological evaluation of liver fibrosis: quantitative image analysis vs semi-quantitative scores. J Hepat 1998; 28:439-46.

[23] Gandon Y, Guyader D, Heautot JF et al. Hemochromatosis: diagnosis and quantification of liver iron with gradient-echo MR imaging. Radiology 1994;193: 533-8.

[24] Teli MR, Day CP, Burt AD et al. Determinants of progression to cirrhosis or fibrosis in pure alcoholic fatty liver. Lancet 1995;346:987-90.


Cliquez pour voir l'image dans sa taille originale

Figure 1. A 32 year old healthy volunteer. Sequence parameters: TR=5000ms, b value=200 sec/mm2. Diffusion – weighted MR image of normal liver with 3 ROI for measurement of ADCs. Mean ADC=3.123.10–3. Software gives diffusion coefficient as numeric data and colorimetric mapping.

Cliquez pour voir l'image dans sa taille originale

Figure 2. A 45 year old cirrhotic patient.Sequence parameters: TR=5000ms, b value=200 sec/mm2. Decrease of the ADCs. Mean ADC = 2.366.10–3.

Cliquez pour voir l'image dans sa taille originale

Figure 3. A 50 year old cirrhotic patient with hemochromatosis.Sequence parameters: TR=5000ms, b value=200sec/mm2. It is not possible to calculate ADC values because of low received signal. Indeed, echoplanar imaging is very susceptible to magnetic field inhomogeneity related to the iron.

© 2004 Elsevier Masson SAS. Tous droits réservés.
EM-CONSULTE.COM is registrered at the CNIL, déclaration n° 1286925.
As per the Law relating to information storage and personal integrity, you have the right to oppose (art 26 of that law), access (art 34 of that law) and rectify (art 36 of that law) your personal data. You may thus request that your data, should it be inaccurate, incomplete, unclear, outdated, not be used or stored, be corrected, clarified, updated or deleted.
Personal information regarding our website's visitors, including their identity, is confidential.
The owners of this website hereby guarantee to respect the legal confidentiality conditions, applicable in France, and not to disclose this data to third parties.
Article Outline
You can move this window by clicking on the headline