Download Characterization of focal liver lesions by ADC

Eur Radiol (2008) 18: 477–485
DOI 10.1007/s00330-007-0785-9
Melanie Bruegel
Konstantin Holzapfel
Jochen Gaa
Klaus Woertler
Simone Waldt
Berthold Kiefer
Alto Stemmer
Carl Ganter
Ernst J. Rummeny
Received: 22 February 2007
Revised: 4 August 2007
Accepted: 11 September 2007
Published online: 25 October 2007
# European Society of Radiology 2007
M. Bruegel (*) . K. Holzapfel .
J. Gaa . K. Woertler . S. Waldt .
C. Ganter . E. J. Rummeny
Department of Radiology, Klinikum
rechts der Isar, Technische Universität
München,
Ismaninger Str. 22,
81675 Munich, Germany
e-mail: [email protected]
Tel.: +49-89-41402621
Fax: +49-89-41404834
B. Kiefer . A. Stemmer
Siemens, Medical Solutions,
Erlangen, Germany
HEPATO BILI ARY-PANCREAS
Characterization of focal liver lesions by ADC
measurements using a respiratory triggered
diffusion-weighted single-shot echo-planar
MR imaging technique
Abstract The aim of this study was to
determine apparent diffusion coefficients (ADCs) of focal liver lesions on
the basis of a respiratory triggered
diffusion-weighted single-shot echoplanar MR imaging sequence (DWSS-EPI) and to evaluate whether ADC
measurements can be used to characterize lesions. One hundred and two
patients with focal liver lesions [11
hepatocellular carcinomas (HCC), 82
metastases, 4 focal nodular hyperplasias (FNH), 56 hemangiomas and 51
cysts; mean size, 16.6 mm; range 5–
92 mm] were examined on a 1.5-T
system using respiratory triggered
DW-SS-EPI (b-values: 50, 300,
600 s/mm2). Results were correlated
with histopathologic data and followup imaging. The ADCs of different
lesion types were compared, and
lesion discrimination using optimal
Introduction
Diffusion-weighted imaging (DWI) has been reported to be
useful for the detection of focal liver lesions [1–3].
Moreover, DWI offers the possibility to obtain criteria for
lesion characterization-independently from T1 and T2
relaxation times and without the need for contrast agent
administration-by quantifying diffusion effects via apparent diffusion coefficient (ADC) measurements. The diffusion coefficient is related to the molecular mobility of water
molecules and reflects tissue properties, such as the size of
the extracellular space (i.e., the rate of relatively unhindered moving water protons), viscosity and cellularity
[4–9]. Accordingly, determination of diffusion coefficients
has been shown to be helpful for the characterization of
thresholds for ADCs was evaluated.
Mean ADCs (×10−3mm2/s) were 1.24
and 1.04 for normal and cirrhotic liver
parenchyma and 1.05, 1.22, 1.40, 1.92
and 3.02 for HCCs, metastases, FNHs,
hemangiomas and cysts, respectively.
Mean ADCs differed significantly for
all lesion types except for comparison
of metastases with HCCs and FNHs.
Overall, 88% of lesions were correctly
classified as benign or malignant
using a threshold value of 1.63×
10−3mm2/s. Measurements of the
ADCs of focal liver lesions on the
basis of a respiratory triggered DWSS-EPI sequence may constitute a
useful supplementary method for
lesion characterization.
Keywords Diffusion magnetic
resonance imaging . Echo-planar
imaging . Liver neoplasms
focal [1, 10–14] and diffuse [10, 14–16] diseases within the
liver and may also have potential for the evaluation of
tumor response to therapy [17, 18].
For diffusion-weighted imaging in the abdomen,
breath-hold single-shot echo-planar imaging (DW-SSEPI) is the most commonly used technique. However,
susceptibility-induced image distortions, blurring artifacts
due to significant T2 decay during the acquisition of
relatively long echo trains, as well as motion artifacts
frequently degrade the quality of DW-SS-EPI images.
Recently, parallel acquisition techniques have been
employed to reduce the aforementioned problems [2,
19–21]. The application of respiratory triggering to DWSS-EPI may further improve image quality, spatial
resolution and signal-to-noise ratio and thus may enhance
478
the ability to characterize focal hepatic lesions on the
basis of ADC measurements.
Hence, the aim of our study was to determine ADC
values of benign and malignant hepatic lesions on the basis
of a respiratory triggered DW-SS-EPI sequence and to
evaluate whether ADC measurements can be used to
characterize lesions.
Materials and methods
Study population
During a 12-month period (from September 2005 to
August 2006), DW-SS-EPI was performed as part of a
routine liver imaging protocol in 249 patients referred to
our institution for MR imaging of the liver. In 72 patients
with primary or secondary hepatic neoplasms, chemotherapy had been performed within the last 12 months prior to
the MR examination. In order to ensure that ADC
measurements were reflective of the natural state of the
liver lesions, those patients were excluded from our
analysis. Of the remaining 177 patients, 75 were excluded,
because: (1) no focal hepatic lesion with a size of ≥5 mm
was present (n=31), (2) sufficient confirmation of the
nature of the lesions was not availabe (n=39) and (3)
distinct motion artifacts were observed on DW-SS-EPI
images (n=5; in these patients, severe ghosting artifacts
occured as a consequence of large amounts of ascites and/
or pleural effusion). Hence, our retrospective analysis
included 102 patients (age range, 30–79 years; mean age,
61 years; 46 women, 56 men). Liver cirrhosis was
diagnosed in ten patients (histopathologically, n=6;
clinically, n=4). In two patients, the clinical history and
the MR imaging findings were indicative of severe
hemosiderosis.
The MR images were analyzed by two radiologists, and
the final diagnoses of hepatic lesions were reached by
consensus involving histopathological data, findings at
PET-CT and/or follow-up imaging studies. Multiple lesions
were present in 86 of the 102 patients and different types of
lesions (for example, cysts and hemangiomas) were
coexisting in 25 patients. In patients with multiple lesions,
only two lesions of each lesion type were randomly
selected for further analysis by the study coordinator (one
of the two radiologists who established the final diagnoses). Lesions less than 5 mm in size were excluded in
order to avoid gross errors due to partial volume effects.
Thus, our study population encompassed a total of 204
hepatic lesions.
Ninety-three lesions were malignant tumors (11 hepatocellular carcinomas and 82 metastases). For all hepatocellular carcinomas and for 33 metastases, histopathologic
verification of the lesions by means of biopsy and/or
surgery was available. The diagnosis of the remaining
metastatic lesions was established on the basis of patho-
logic tracer uptake of the lesions at PET-CT (n=14), or
progression or regression in lesion size on serial crosssectional imaging studies after the commencement of
chemotherapy in patients with known extrahepatic primary
malignancies (n=35). The primary sites of the metastatic
lesions included breast carcinoma (n=9), bronchial carcinoma (n=2), colorectal carcinoma (n=35), duodenal
carcinoma (n=4), esophageal carcinoma (n=3), gastric
carcinoma (n=4), neuroendocrine carcinoma (n=15), melanoma (n=2), pancreatic carcinoma (n=2), renal cell
carcinoma (n=4) and urachal carcinoma (n=2).
There were a total of 111 benign lesions. Four lesions
were benign solid masses (focal nodular hyperplasias) and
107 lesions were nonsolid lesions (56 hemangiomas and 51
cysts). Histopathologic proof was available in three
hemangiomas. The remaining benign lesions showed
typical MR imaging findings [22, 23] in conjunction with
stability in lesion size on serial cross-sectional imaging
studies with a minimum follow-up interval of 12 months.
MR imaging
MR imaging was performed on a 1.5-T system (Magnetom
Avanto, Siemens Medical Solutions, Erlangen, Germany)
with two six-channel body phased array coils anterior and
two spine clusters (three channels each) posterior. Routine
breath-hold T2-weighted half-Fourier acquisition singleshot turbo spin-echo (HASTE), breath-hold T2-weighted
turbo spin-echo (TSE) and dynamic contrast-enhanced 3D
gradient-echo (volumetric interpolated breath-hold examination, VIBE) sequences were performed in all patients.
The DW-SS-EPI sequence used in this study is a vendorsupplied work-in-progress package. A single-shot EPI
readout is preceded by a diffusion-sensitizing block
consisting of two 180° radiofrequency pulses and four
motion probing gradient (MPG) pulses. Compared to the
conventional Stejskal-Tanner preparation this scheme
reduces the influence of eddy currents [24, 25]. The
technical parameters were as follows: echo time, 69 ms;
echo train length, 58; echo spacing, 0.69; receiver
bandwidth, 1,736 Hz/pixel; spectral fat saturation; field
of view, 263×350 mm; matrix, 144×192; number of signal
averages, 3; section thickness, 5 mm; intersection gap,
0.5 mm; 30–45 transverse sections acquired; ≈4–6-min
acquisition time. For shortening of the echo train length,
integrated parallel imaging techniques (iPAT) by means of
generalized autocalibrating partially parallel acquisitions
(GRAPPA) [26] with a twofold acceleration factor were
used. For respiratory triggering, PACE (prospective acquisition correction) was implemented. The PACE technique
interleaves the imaging sequence with a navigator sequence. The information gained with the navigator is used
to synchronize the measurement with the patient’s breathing cycle and to place the data acquisition period into the
end-expiration phase. The number of sections acquired per
479
respiratory cycle (i.e., the number of sections per block) is
adjusted to fit the individual breathing cycle of the patient.
Typically 15 sections were acquired per respiratory cycle.
The gradient factors (b-values) and spatial direction of the
MPGs are identical for all sections acquired during one
respiratory cycle and are altered only in between respiratory cycles. Three mutually perpendicular spatial directions
were encoded with three increasing b-values: 50, 300 and
600 s/mm2. In order to acquire images with a high contrastto-noise ratio for optimal conspicuity of liver lesions while
keeping “pseudodiffusion” by means of perfusion effects
low, the minimum b-value was set at 50 s/mm2. Trace
images were synthesized for each b-value, and an ADC map
was calculated from all diffusion weightings and directions.
Image analysis
Review of all MR images and of all follow-up imaging
studies (MRI, CT, PET-CT) was performed on a PACS
workstation (Easy vision, Philips, Best, The Netherlands).
The study coordinator recorded the final diagnoses of all
selected lesions and their location according to Couinaud’s
segmental anatomy. The size of each lesion was determined
by the largest diameter as displayed on DW-SS-EPI (b=
50 s/mm2) images.
The DW-SS-EPI images were quantitatively analyzed by
one radiologist who was blinded to the diagnosis of the
lesions and to the results of the other MR imaging
sequences. A satellite console of the MR unit was used
for the ADC measurements. Mean ADC values of normalappearing liver parenchyma and the spleen were obtained
from the ADC maps in each patient. For evaluation of the
liver parenchyma, regions of interest (ROIs) with an
approximate size of 100 pixels were placed in four
locations away from prominent vascular structures: (1)
segment II, anteriorly; (2) segment IVb, centrally; (3)
segment VI, posteriorly; (4); and segment VIII, centrally.
For lesion evaluation, a circular ROI encompassing as
much of a lesion as possible was first drawn on the b=50 s/
mm2 image and then transferred onto the ADC map. Two
repetitive measurements per lesion were undertaken, and
the ADC values were then averaged.
Statistical analysis
Statistical analysis was performed using SPSS software
(version 11.5, SPSS, Chicago, IL). The mean ADC values
of normal and cirrhotic liver parenchyma were compared
using the Mann-Whitney test. The paired t-test was used
for comparison of the ADC values obtained from the four
different locations in the liver. The Kruskal-Wallis test was
performed in order to assess for statistically significant
differences among the mean ADC values of the different
types of hepatic lesions, and subsequent pairwise compar-
isons of lesion groups were performed using the MannWhitney test. A p-value was considered significant at
<0.05. Bonferroni correction was used for multiple
pairwise comparisons. Additionally, optimal ADC threshold values for lesion discrimination were determined by
means of ROC analysis, and corresponding sensitivities,
specificities and accuracies were calculated.
Results
Evaluation of liver and spleen parenchyma
Mean ADC values obtained from normal, cirrhotic and
hemosiderotic liver parenchyma as well as from the splenic
parenchyma are summarized in Table 1.
For normal liver parenchyma, a mean ADC value of
1.24×10−3mm2/s was observed. Cirrhotic liver parenchyma was found to have significantly lower ADC values
(1.04×10−3mm2/s) compared with normal liver parenchyma (p<0.001). Two patients were diagnosed with severe
hemosiderosis, which displayed extremely low ADC
values (0.44×10−3mm2/s).
When comparing the mean ADC values obtained from
four different locations in the liver, no significant difference
in the values of segments IVb and VI was found (p=0.94).
However, segment VIII revealed slightly lower ADC values
compared to all other remaining segments, whereas
segment II showed notedly higher ADC values (p<0.001
for all pairwise comparisons) (Table 1).
Table 1 Apparent diffusion coefficients (ADCs) of liver and spleen
ADC [10−3mm2/s]
Liver parenchyma in patients without known
cirrhosis (n=90)
Segment II, anterior region
Segment IVb, central region
Segment VI, dorsal region
Segment VIII, central region
Average value of four liver segments
Liver parenchyma in patients with cirrhosis
(n=10)
Segment II, anterior region
Segment IVb, central region
Segment VI, dorsal region
Segment VIII, central region
Average value of four liver segments
Liver parenchyma in patients with severe
hemosiderosis (n=2)
Average value of four liver segments
Spleen (n=96)
a
Data are mean values ± standard deviation
1.44±0.28
1.21±0.13
1.21±0.15
1.12±0.18
1.24±0.15
1.19±0.27
1.01±0.22
0.98±0.23
0.97±0.25
1.04±0.23
0.44±0.05
0.82±0.11
a
480
Evaluation of focal liver lesions
Seventy-three (36%) of the 204 focal liver lesions were
located in the left lobe (segments I to IV) and the remaining
131 (64%) were located in the right lobe (segments V to
VIII). The mean size of lesions was 16.6 mm (range: 5–
92 mm).
The box plots of the ADC values of hepatocellular
carcinomas (HCCs), metastases, focal nodular hyperplasias
(FNHs), hemangiomas and cysts are shown in Fig. 1. ADC
values of metastases overlapped strongly with ADC values
of hepatocellular carcinomas (HCC) and focal nodular
hyperplasias (FNH), and to some extent with ADC values
of hemangiomas. ADC values of hemangiomas also
partially overlapped with those of FNHs and cysts. Mean
ADCs were as follows: HCC, 1.05×10−3mm2/s; metastases,
1.22×10−3mm2/s; FNH, 1.40×10−3mm2/s; hemangiomas,
1.92×10−3mm2/s; and cysts, 3.02×10−3mm2/s (Table 2).
Plots showing the 95% confidence intervals (CI) of the
mean ADC values of the different lesion types are shown in
Fig. 2. Metastases were found to have significantly lower
ADCs compared with hemangiomas (p<0.001), and
hemangiomas revealed significantly lower ADCs compared with cysts (p<0.001). ADCs of HCCs were
significantly lower than those of all types of benign lesions
(p<0.005 for all) (Fig. 3). Although HCCs showed a
Fig. 1 Box plots of the ADC values of 204 focal liver lesions.
Boxes stretch across interquartile range (IR), i.e., from lower
quartile (Q1) to upper quartile (Q2); whiskers show smallest data
point that is greater than [Q1−1.5 × IR] and largest data point that is
smaller than [Q2+1.5 × IR]; median is shown as line across each bar;
O denotes outliers. ADC values of metastases overlapped with ADC
values of hepatocellular carcinomas (HCC), focal nodular hyperplasias (FNH) and hemangiomas. ADC values of hemangiomas also
overlapped with ADC values of FNHs and cysts
Table 2 Apparent diffusion coefficients (ADCs) of focal liver lesions
ADC [10−3mm2/s] a
Lesions
Lesions
5–9 mm
≥10 mm
(n=77)
(n=127)
Hepatocellular
carcinomas
Metastases
Focal nodular
hyperplasias
Hemangiomas
Cysts
a
–
1.19±0.33
(n=27)
–
1.78±0.34
(n=19)
2.92±0.33
(n=31)
1.05±0.09
(n=11)
1.24±0.30
(n=55)
1.40±0.15
(n=4)
1.99±0.32
(n=37)
3.16 ± 0.21
(n=20)
Lesions
overall
(n=204)
1.05±0.09
(n=11)
1.22±0.31
(n=82)
1.40±0.15
(n=4)
1.92±0.34
(n=56)
3.02±0.31
(n=51)
Data are mean values ± standard deviation
slightly lower mean ADC compared with that of metastases, a statistically significant difference was not observed
(p=0.073). The ADCs of FNHs were significantly lower
than those of all other benign lesions (p<0.01 for all)
(Fig. 4). Compared with metastases, FNHs showed a
slightly higher mean ADC, but the difference did not reach
statistical significance (p=0.087).
Mean ADC values of metastases, hemangiomas and
cysts according to lesion size criteria are given in Table 2.
Fig. 2 Plots showing the 95% confidence intervals (CI) of the mean
ADC values of 204 focal liver lesions. The mean ADC values of the
different lesion types showed significant differences (p<0.01) with
the exception of mean ADCs of hepatocellular carcinomas (HCC)
and metastases (p=0.073) and mean ADCs of metastases and focal
nodular hyperplasias (FNH) (p=0.087)
481
Fig. 3 (a) b=50 s/mm2 DWSS-EPI image. (b) b=600 s/
mm2 DW-SS-EPI image. (c)
ADC map. The hepatocellular
carcinoma (HCC) in segment IV
(arrowhead) shows a merely
moderate signal loss from the
b=50 s/mm2 to the b=600 s/mm2
DW-SS-EPI image. The cyst in
segment II (arrow) displays
markedly high signal intensity on
the b=50 s/mm2 DW-SS-EPI
image and becomes isointense on
the b=600 s/mm2 DW-SS-EPI
image. On the corresponding
ADC map, the HCC shows a low
ADC value (1.18×10−3mm2/s),
whereas the cyst has a high ADC
value (3.09×10−3mm2/s)
Small lesions (5–9 mm) generally showed slightly lower
ADC values than the remaining lesions (≥10 mm), which
may be attributed to partial volume effects.
Table 3 shows sensitivities, specificities and accuracies
for lesion characterization in our study population when
optimal thresholds of ADC values were applied. Overall,
88% of lesions were correctly classified as benign or
malignant when an ADC value of 1.63×10−3mm2/s was
used as a threshold. Of the 25 misclassified lesions, 19
were found in the right lobe of the liver with the remaining
six in the left lobe.
Most of the misclassified lesions were metastases or
hemangiomas. When looking closer at these two groups,
differentiation of metastases and hemangiomas was most
accurate using a threshold of 1.57×10−3mm2/s (Fig. 5).
However, only 83% of cases were correctly classified in
this way. The 13 misclassified metastases with uncommonly high ADC values originated from duodenal, pancreatic and urachal carcinoma (n=2 lesions in each),
neuroendocrine (n=3) and colorectal carcinoma (n=5). It
should be noted that our study population included 12
further metastases of neuroendocrine carcinomas and 30
further metastases of colorectal carcinomas, which were all
found to have expectedly low ADC values. Thus, a valid
correlation between the ADCs and the primary sites of the
metastatic lesions cannot be derived from our data. Of the
ten misclassified hemangiomas in our study population,
five were smaller than 8 mm in size; however, the size of
the remaining lesions ranged from 11 to 23 mm. As shown
in Fig. 1, there was no overlap in the ADC values of
metastases and cysts in our study population. Thus, an
accuracy of 100% for the differentiation of these two lesion
entities was obtained when a threshold of 2.10×10−3mm2/s
was used.
Discussion
The use of DW-SS-EPI in combination with navigatorcontrolled respiratory triggering and parallel acquisition
techniques enabled us to acquire high-quality diffusionweighted images of the liver within a relatively short
acquisition time of approximately 4 to 6 min. Even small
hepatic lesions were clearly visible on the b=50 s/mm2
images. This is reflected by our study material, which
included 77 (38% out of 204) lesions with a size of less
than 10 mm. To our knowledge, the application of
respiratory triggering to DW-SS-EPI has not been widely
evaluated. A previous study compared respiratory triggered
and breath-hold DW-SS-EPI for liver imaging, and respiratory triggered DW-SS-EPI was found to show overall
better image quality and a significantly higher lesion-toliver contrast ratio [27]. Another publication reported that
DWI under free breathing was inferior to DWI with
simultaneous respiratory triggering regarding accuracy of
ADC measurement [28].
However, in spite of improved image quality and lesion
conspicuity with respiratory triggered DW-SS-EPI, some
technical limitations remain. We observed that liver
parenchyma in segment II displayed notedly higher ADC
values than the parenchyma in the remaining segments,
which can be explained by the increased exposure of
482
to the heart and diaphragm, and which indicated that ADCs
acquired without pulse triggering were artificially increased by motion influences. It may be assumed that
cardiac motion artifacts have an impact not only on the
ADCs of liver parenchyma, but also on the ADCs of focal
liver lesions in sections close to the heart. However, in our
study population, only 6 of the 25 lesions, which were
misclassified on the basis of their ADC values, were
located in regions particularly exposed to cardiac motion,
i.e., in the left lobe of the liver.
The limited achievable SNR is another potential source
of error in the determination of ADC values, because there
can be a systematic underestimation of ADC values for
tissues with low SNR. The theoretical background for this
issue is well explained in an article by Jones and Basser
[30]. In our measurements, noise contamination might
explain the slightly lower ADC values obtained for liver
parenchyma in the central region of segment VIII compared
to those obtained posteriorly in segment VI and anteriorly in
segment IVb, since the latter regions were located closer to
the receiver coils. In contrast, no relevant noise-related
bias may be expected for the ADC of the liver lesions,
which typically appeared hyperintense at least on b=50 and
300 s/mm2 images due to their relatively high SNR.
Previous publications showed large discrepancies regarding ADC values in the abdomen. For example,
reported mean ADCs (×10−3mm2/s) range from 0.69 to
2.28 for normal liver parenchyma, and from 0.85 to 2.85
for metastases [1, 11–14, 19, 29, 31, 32]. Besides the
aforementioned technical limitations, the choice of the bvalues has a substantial influence on the resulting ADCs.
The ADC quantifies intravoxel incoherent motion, which
integrates the effects of both diffusion and capillary
Table 3 Differentiation of focal liver lesions-optimal thresholds of
apparent diffusion coefficients (ADCs) in our study population
Fig. 4 (a) T2-weighted inversion-recovery image shows a focal
nodular hyperplasia (FNH) in the left liver lobe with a hyperintense
central scar (arrowhead). Additionally, a small cyst can be seen in
segment VI (arrow). (b) The FNH is moderately hyperintense on the
b=50 s/mm2 DW-SS-EPI image, whereas the cyst displays very
high signal. (c) On the ADC map, a relatively low ADC value was
observed for the FNH (1.35×10−3mm2/s). The cyst displays a high
ADC value (2.75×10−3mm2/s) and is clearly visible on the ADC
map despite its small size
segment II to cardiac motion artifacts. Mürtz et al.
performed ADC measurements in healthy volunteers
using a diffusion-weighted single-shot sequence both
without and with pulse triggering [29]. The authors
reported that ADCs of abdominal organs obtained with
pulse triggering were lower than those obtained with fixed
repetition times, which particularly applied to regions close
Threshold
[ADC × 10−3
mm2/s]
Metastasis vs. 1.57
hemangioma
Metastasis vs. 2.10
cyst
Malignantb
vs. benignc
lesions
1.63
Sensitivity Specificity Accuracy
(%)a
(%)a
(%)a
84
(69/82)
[75;91]
100
(82/82)
[97;100]
90
(84/93)
[83;95]
82
(46/56)
[71;90]
100
(51/51)
[95;100]
86
(95/111)
[78;91]
83
(115/138)
[76;89]
100
(133/133)
[98;100]
88
(179/204)
[83;92]
a
Data in parantheses were used to calculate proportions
Data in brackets show 95% confidence intervals
b
HCCs and metastases.
c
FNHs, hemangiomas and cysts
483
Fig. 5 (a) Arterial phase contrast-enhanced T1-weighted 3D GRE
image shows a large hypovascular metastasis in segment VII
(asterisk) and a small hemangioma with globular enhancement in
segment VIII (arrow). (b) Both lesions are markedly hyperintense on
the b=50 s/mm2 DW-SS-EPI image (arrowheads denote central bile
ducts). (c) On the ADC map, the hemangioma shows a relatively
high ADC value (2.33×10−3mm2/s), whereas the ADC of the
metastasis (1.08×10−3mm2/s) is similar to that of the surrounding
liver parenchyma. The central bile ducts are hyperintense on the
ADC map due to their fluid content
perfusion. For large b-values (≥300 s/mm2), perfusion
effects are cancelled out, whereas ADCs calculated from
very low b-values may be artificially increased by capillary
perfusion in terms of a “pseudodiffusion” [4, 8, 14].
In accordance with previous reports, we found that the
ADCs of cirrhotic liver parenchyma were lower than those
of normal liver parenchyma [1, 10, 13–15]. As expected,
cysts exhibited the highest ADC values because of the
relatively unrestricted motion of water molecules within
their fluid contents, whereas HCCs, metastases and FNHs
showed the lowest ADC values probably due to their high
cellularity. In conformity with former studies, no overlaps
were found between the ADCs of cysts and solid lesion
[11, 14]. However, a clinically more relevant problem is the
discrimination of metastases from hemangiomas, since
hemangiomas may demonstrate atypical contrast enhancement patterns similar to those of hypervascular metastases
or may hyalinize and therefore show decreased signal
intensity on T2-weighted images [33], whereas necrotic
metastases may exhibit pronounced hyperintensity mimicking hemangioma. The ADCs of metastases and
hemangiomas have been reported to be significantly
different regarding their means, but have also been
shown to overlap to some extent [1, 10, 11, 13, 14]. In
our study, we obtained an accuracy of 83% for discriminating these types of lesions. When reviewing the 13
misclassified metastases on corresponding T2-weighted
images, we found seven lesions to be markedly hyperintense, which may reflect less restricted diffusion due to
necrotic changes. However, the remaining six lesions
displayed only mild to moderate hyperintensity on T2weighted TSE images. Partial volume effects in very smallsized lesions may explain some, but not all of the
abnormally low ADC values obtained in the ten misclassified hemangiomas in our study population.
All in all, discrimination of benign from malignant
lesions by means of ADC values was successful in 88% of
our lesions. This accuracy for lesion characterization is
slightly inferior to that obtained on the basis of quantitative
measurements of T2 relaxation times [34–36], but it should
be considered that the mean size of lesions in our study
population was very small, which might have caused
problems to arise with quantitative measurements of T2
relaxation times as well.
As a limitation of the current respiratory triggered DWSS-EPI technique, cardiac motion artifacts and noise
contamination may distort ADC values to a certain degree.
Additional pulse triggering may overcome cardiac motionrelated artifacts, and the number of acquisition averages
could be increased to reduce noise contamination. In order
to keep acqusition time within reasonable frames, the
number of b-values could be lowered to two (for example,
b=50 and 600 s/mm2).
In conclusion, measurements of the ADCs of focal liver
lesions on the basis of a respiratory triggered DW-SS-EPI
484
sequence may constitute a useful supplementary method
for lesion characterization. Malignant lesions demonstrate
significantly lower ADC values than benign lesions, but
interpretation of the ADCs should be handled with care,
since overlaps between some individual lesion types such
as metastases and hemangiomas seem to be inevitable.
References
1. Ichikawa T, Haradome H, Hachiya J,
Nitatori T, Araki T (1998) Diffusionweighted MR imaging with a singleshot echoplanar sequence: detection
and characterization of focal hepatic
lesions. AJR Am J Roentgenol
170:397–402
2. Nasu K, Kuroki Y, Nawano S et al
(2006) Hepatic metastases: diffusionweighted sensitivity-encoding versus
SPIO-enhanced MR imaging. Radiology
239:122–130
3. Okada Y, Ohtomo K, Kiryu S, Sasaki Y
(1998) Breath-hold T2-weighted MRI
of hepatic tumors: value of echo planar
imaging with diffusion-sensitizing gradient. J Comput Assist Tomogr
22:364–371
4. Le Bihan D, Breton E, Lallemand D
et al (1988) Separation of diffusion and
perfusion in intravoxel incoherent
motion MR imaging. Radiology
168:497–505
5. Marks MP, de Crespigny A, Lentz D et
al (1996) Acute and chronic stroke:
navigated spin-echo diffusion-weighted
MR imaging. Radiology 199:403–408
6. Sugahara T, Korogi Y, Kochi M et al
(1999) Usefulness of diffusionweighted MRI with echo-planar
technique in the evaluation of
cellularity in gliomas. J Magn Reson
Imaging 9:53–60
7. Bammer R (2003) Basic principles of
diffusion-weighted imaging. Eur J
Radiol 45:169–184
8. Turner R, Le Bihan D, Maier J et al
(1990) Echo-planar imaging of intravoxel incoherent motion.
Radiology 177:407–414
9. Thoeny HC, De Keyzer F (2007)
Extracranial applications of diffusionweighted magnetic resonance imaging.
Eur Radiol 17:1385–1393
10. Kim T, Murakami T, Takahashi S et al
(1999) Diffusion-weighted single-shot
echoplanar MR imaging for liver
disease. AJR Am J Roentgenol
173:393–398
11. Namimoto T, Yamashita Y, Sumi S,
Tang Y, Takahashi M (1997) Focal liver
masses: characterization with diffusionweighted echo-planar MR imaging.
Radiology 204:739–744
12. Sun XJ, Quan XY, Huang FH, Xu YK
(2005) Quantitative evaluation of
diffusion-weighted magnetic resonance
imaging of focal hepatic lesions. World
J Gastroenterol 11:6535–6537
13. Taouli B, Vilgrain V, Dumont E et al
(2003) Evaluation of liver diffusion
isotropy and characterization of focal
hepatic lesions with two single-shot
echo-planar MR imaging sequences:
prospective study in 66 patients.
Radiology 226:71–78
14. Yamada I, Aung W, Himeno Y,
Nakagawa T, Shibuya H (1999) Diffusion coefficients in abdominal organs
and hepatic lesions: evaluation with
intravoxel incoherent motion echo-planar
MR imaging. Radiology 210:617–623
15. Amano Y, Kumazaki T, Ishihara M
(1998) Single-shot diffusion-weighted
echo-planar imaging of normal and
cirrhotic livers using a phased-array
multicoil. Acta Radiol 39:440–442
16. Ichikawa T, Haradome H, Hachiya J,
Nitatori T, Araki T (1999) Diffusionweighted MR imaging with single-shot
echo-planar imaging in the upper abdomen: preliminary clinical experience
in 61 patients. Abdom Imaging
24:456–461
17. Deng J, Miller FH, Rhee TK et al
(2006) Diffusion-weighted MR imaging for determination of hepatocellular
carcinoma response to yttrium-90
radioembolization. J Vasc Interv Radiol
17:1195–1200
18. Deng J, Rhee TK, Sato KT et al (2006)
In vivo diffusion-weighted imaging of
liver tumor necrosis in the VX2 rabbit
model at 1.5 Tesla. Invest Radiol
41:410–414
19. Yoshikawa T, Kawamitsu H, Mitchell
DG et al (2006) ADC measurement of
abdominal organs and lesions using
parallel imaging technique. AJR Am J
Roentgenol 187:1521–1530
20. Koh DM, Scurr E, Collins DJ et al
(2006) Colorectal hepatic metastases:
quantitative measurements using
single-shot echo-planar diffusionweighted MR imaging. Eur Radiol
16:1898–1905
21. Taouli B, Martin AJ, Qayyum A et al
(2004) Parallel imaging and diffusion
tensor imaging for diffusion-weighted
MRI of the liver: preliminary experience in healthy volunteers. AJR Am J
Roentgenol 183:677–680
22. Bartolozzi C, Cioni D, Donati F,
Lencioni R (2001) Focal liver lesions:
MR imaging-pathologic correlation.
Eur Radiol 11:1374–1388
23. Horton KM, Bluemke DA, Hruban RH,
Soyer P, Fishman EK (1999) CT and
MR imaging of benign hepatic and
biliary tumors. Radiographics 19:431–451
24. Heid O (2000) Eddy Current-Nulled
Diffusion Weighting. Proc Intl Soc Mag
Reson Med 8: abstract number 799
25. Stejskal EO, Tanner JE (1965) Spin
diffusion measurements: spin echoes in
the presence of a time-dependent field
gradient. J Chem Phys 42:288–292
26. Griswold MA, Jakob PM, Heidemann
RM et al (2002) Generalized autocalibrating partially parallel acquisitions
(GRAPPA). Magn Reson Med
47:1202–1210
27. Sandberg A, Parikh T, Johnson G,
Stemmer A, Taouli B (2006) Feasibility
of a respiratory-triggered SSEPI diffusionweighted sequence for liver imaging using
navigator echo technique: comparison
with breath-hold diffusion-weighted sequence. Proc Intl Soc Mag Reson Med 14:
abstract number 400
28. Nasu K, Kuroki Y, Sekiguchi R,
Nawano S (2006) The effect of simultaneous use of respiratory triggering in
diffusion-weighted imaging of the liver.
Magn Reson Med Sci 5:129–136
29. Mürtz P, Flacke S, Traber F et al (2002)
Abdomen: diffusion-weighted MR imaging with pulse-triggered single-shot
sequences. Radiology 224:258–264
30. Jones DK, Basser PJ (2004) “Squashing
peanuts and smashing pumpkins”: how
noise distorts diffusion-weighted MR
data. Magn Reson Med 52:979–993
31. Chow LC, Bammer R, Moseley ME,
Sommer FG (2003) Single breath-hold
diffusion-weighted imaging of the abdomen. J Magn Reson Imaging
18:377–382
32. Müller MF, Prasad P, Siewert B et al
(1994) Abdominal diffusion mapping
with use of a whole-body echo-planar
system. Radiology 190:475–478
33. Vilgrain V, Boulos L, Vullierme MP et al
(2000) Imaging of atypical hemangiomas of the liver with pathologic
correlation. Radiographics 20:379–397
485
34. Cieszanowski A, Szeszkowski W,
Golebiowski M et al (2002) Discrimination of benign from malignant
hepatic lesions based on their T2relaxation times calculated from
moderately T2-weighted turbo SE
sequence. Eur Radiol 12:2273–2279
35. Farraher SW, Jara H, Chang KJ,
Ozonoff A, Soto JA (2006)
Differentiation of hepatocellular
carcinoma and hepatic metastasis from
cysts and hemangiomas with
calculated T2 relaxation times and
the T1/T2 relaxation times ratio.
J Magn Reson Imaging
24:1333–1341
36. Fenlon HM, Tello R, DeCarvalho VL,
Yucel EK (2000) Signal characteristics
of focal liver lesions on double echo
T2-weighted conventional spin echo
MRI: observer performance versus
quantitative measurements of T2
relaxation times. J Comput Assist
Tomogr 24:204–211