In vitro and in vivo repeatability of abdominal diffusion-
1,2M E MIQUEL, PhD,1A D SCOTT, PhD,1N D MACDOUGALL, PhD,1,2R BOUBERTAKH, PhD,
3N BHARWANI, MRCP, FRCR and4A G ROCKALL, MRCP, FRCR
1Clinical Physics, Barts Health NHS Trust, London, UK,2Barts and the London Cardiovascular Biomedical Research Unit,
London, UK,3Radiology Department, Imperial College Healthcare NHS Trust, London, UK, and4Department of Imaging,
Barts Health NHS Trust, London, UK
diffusion coefficient (ADC) measurements at 1.5T using a free-breathing multislice
diffusion-weighted (DW) MRI sequence.
DW MRI images were obtained using a multislice spin-echo echo-planar
imaging sequence with b-values50, 100, 200, 500, 750 and 1000smm22. A flood-field
phantom was imaged at regular intervals over 100 days, and 10 times on the same day
on 2 occasions. 10 healthy volunteers were imaged on two separate occasions. Mono-
exponential ADC maps were fitted excluding b50. Paired analysis was carried out on
the liver, spleen, kidney and gallbladder using multiple regions of interest (ROIs) and
volumes of interest (VOIs).
The in vitro coefficient of variation was 1.3% over 100 days, and 0.5% and 1.0%
for both the daily experiments. In vivo, there was no statistical difference in the group
mean ADC value between visits for any organ. Using ROIs, the coefficient of
reproducibility was 20.0% for the kidney, 21.0% for the gallbladder, 24.7% for the liver
Good in vitro repeatability of ADC measurements provided a sound basis
for in vivo measurement. In vivo variability is higher and when considering single
measurements in the abdomen as a whole, only changes in ADC value greater than
23.1% would be statistically significant using a two-dimensional ROI. This value is
substantially lower (7.9%) if large three-dimensional VOIs are considered.
To study the in vitro and in vivo (abdomen) variability of apparent
Received 2 November 2011
Revised 7 February 2012
Accepted 14 March 2012
’ 2012 The British Institute of
Brownian motion of water in biological tissues [1, 2].
The technique has played a preponderant role in neuro-
imaging over the last two decades and it is known to
detect small changes before they are apparent on
anatomical imaging [3, 4].
In recent years DW MRI has been increasingly used in
other parts of the body, demonstrating great diagnostic
potential in cancer imaging. To date, DW MRI has been
successfully used for tissue characterisation and tumour
staging. However, the apparent diffusion coefficient
(ADC) is a potential biomarker that could be used to
monitor treatment response or evaluate post-therapeutic
changes. Details of the clinical use of DW MRI can be
found in the 2009 consensus paper  or in general and
organ-specific review articles [6–8].
in ADC values are reported in the literature depending on
the acquisition parameters, in particular the choice of b-
values (e.g. see  for ADC values in the kidney or Table 1
for the liver). The 2009 consensus and recommendation
paper  highlighted the importance of quality analysis,
validation and reproducibility studies. Although there are
some emerging reproducibility and repeatability data in the
abdomen [15, 19–22], a recent review by Taouli and Koh 
highlights the need for further work in this area. Recently,
coefficients of variability of around 14% were published for
seem to indicate that only ADC changes of over 27% 
or 30%  are significant. Substantial variations in ADC
values have also been found between different scanners and
vendors [24–26], further highlighting the difficulty of setting
up multicentre trials.
In preparation for a study on renal cell carcinoma at
our centre, we required information on the variability of
a free-breathing multislice DW MRI sequence. As these
tumours are relatively large and heterogeneous, we were
particularly interested in the variability of both large
volumes on multiple slices and smaller regions on
Address correspondence to: Dr Marc Miquel, Clinical Physics, 60
Dominion House, 4th Floor, St Bartholomew’s Hospital, London
EC1A 7BE, UK. E-mail: email@example.com
MEM is partly funded (20%) by the UK National Institute for Health
The British Journal of Radiology, 85 (2012), 1507–1512
The British Journal of Radiology, November 20121507
Methods and materials
After giving informed consent, 10 healthy volunteers
(7 females, 3 males; age 32.3¡4.6 years, range 26–42
years) were imaged, in accordance with local ethics
regulations, on two occasions (second visit 5.8¡1.9 days
later, range 5–11 days).
MRI diffusion sequence and apparent diffusion
Scans were performedon a 1.5T Achieva system (Philips
Medical Systems, Best, The Netherlands) in conjunction
with a four-element body coil array. DW MRI images were
obtained using a free-breathing multislice spin-echo echo-
planar imaging (EPI) sequence: repetition time (TR) 5300–
field-of-view (FOV) 400–450mm, rectangular FOV 75%,
matrix 1126256, 28–35 slices to cover the abdomen from
the diaphragm to the iliac crest, slice thickness 6mm, slice
gap 1mm. Six motion probing gradients with b-values of 0,
100, 200, 500, 750 and 1000smm22were applied in three
orthogonal directions and trace images were synthesised
for each b-value using the mean of three orthogonal
directions. ADC maps were calculated on a pixel-by-pixel
basis using a mono-exponential fit, and b50 was excluded
from the calculation in order to eliminate perfusion effects.
Two analyses were run, one on volumes of interest (VOIs)
and one on multiple smaller regions of interest (ROIs). VOI
analysis was performed for the following organs: entire
spleen, entire gallbladder, kidney (renal parenchyma) and
liver (part of the right hepatic lobe, approximately
segments V and VI). For the ROI analysis circular ROIs
(area 4cm2, or 1–2cm2for the gallbladder) were placed on
same slice). The numbers of ROIs used were: liver, 5;
kidney, 6 (3 in each); spleen, 3; gallbladder, 2. The ROIs
were then copied to the nearest corresponding anatomical
position on the second visit images. For bothmethods, data
analysis was carriedout on matched pairs for each organ to
investigate ADC change between visits and calculate the
coefficient of repeatability expressed as a percentage of the
The intra- and interobserver variability was assessed
for the VOI analysis on the dataset from the first visit by
analysing the difference between matched pairs of the
results obtained by one reader on two occasions and two
independent readers, respectively.
A large (5l)copper sulphatesolution([CuSo4]5
361023moll21, [NaCl]53.461022moll21) phantom was
imaged 10 times on two different occasions, and also at
regular intervals over a period of 3 months. To minimise the
17¡0.5u uC. For multiple image acquisitions on the same
occasion, the phantom was removed from the scanner and
repositioned between each acquisition.
The sequence was identical to the sequence described
above for in vivo imaging but used a reduced number of
slices (20). For analysis, a circular ROI covering 90% of the
cross-section of the bottle was selected on each slice. The
coefficient of variation (CV; %) was calculated as CV5
Table 1. Apparent diffusion coefficient values measured in normal liver at 1.5T
Taouli et al 1.60
10 v 0, 500Conventional
With parallel imaging
Mu ¨rtz et al 0.92–0.96a
12 v50, 300, 700, 1000, 1300
Kim et al 
6 v/126p3, 57, 192, 408, 517, 850
3, 57, 192, 408, 192, 408
0, 134, 267, 400
Ichikawa et al  2.28
Taouli et al 
Kwee et al 
Free breathing (7mm slice)
Free breathing (5mm slice)
Diffusion coefficient (DC)
Yamada et al 78p30, 300, 900,1100
Mu ¨ller et al 
Namimato et al  0.69
8 b-values; bmax328–454
100, 200, 500, 750, 10001.040.95–1.11Free breathing
ADC, apparent diffusion coefficient; p, patients; v, volunteers.
In studies including patients, only ADC values relating to measurements performed in normal liver are quoted here.
aValue range for 3 directions.
cEach sequence repeated three times.
M E Miquel, A D Scott, N D MacDougall et al
1508 The British Journal of Radiology, November 2012
mean) for the daily repeat experiments and over the 3-
month period. CV was also calculated for each slice and
The ADC values were stable over the studied period
(Figure 1) and no significant image artefact was observed.
On the first day, the mean ADC of the phantom was
200.1¡1.061025mm2s21with a CV of 0.5% (10 measure-
ments). On day 100, the mean ADC of the phantom was
199.4¡2.061025mm2s21with a CV of 1.0% (10 measure-
ments). The mean ADC over the 3-month period was
199.0¡2.661025mm2s21with a CV of 1.3%. The mean
intraslice CV was 3.2¡1.4% (range 1.0–6.9%). The mean
sample CV was 2.9¡1.0% (range 1.5–6.3%). A small
gradient in ADC value was always observed between one
end of the phantom and the other along the z-axis of the
scanner; an example measurement is displayed in
Figure 1. In vitro apparent diffu-
sion coefficient (ADC) measure-
ments. (a) Variability over 100 days
with horizontal lines showing the
mean (solid) and standard deviation
(dashed) over the period. (b) Daily
repeats with horizontal lines show-
ing the mean and standard devia-
tion of the 10 scans. (c) Example of
phantom ADC measurements where
points are the mean ADC values for
each slice, the solid line is the mean
for the whole phantom, the dashed
lines show the standard deviation
for the slices and dotted lines show
the standard deviation of all pixels.
Short communication: Repeatability of abdominal diffusion-weighted MRI
The British Journal of Radiology, November 20121509
In vivo study
Both inter- and intraobserver variability were small.
between pairs was 20.1¡1.5% and for interobserver it
was 20.2¡2.5%, leading to coefficients of repeatability
of 3.0% and 4.8%, respectively.
No statistical difference was found between the two
visits using a paired t-test for the two types of analysis.
The ADC values obtained for the liver, spleen, kidney
and gallbladder using the VOI method are displayed in
Table 2. The mean volumes of the VOIs were gallbladder
18¡5, liver 331¡93, kidney 148¡27 and spleen 168¡
47cm3. The mean change in ADC value (expressed as a
percentage to facilitate comparison between organs)
between the two visits for all organs and the two types
of analyses (VOI and multiple ROIs) was always small
(between 21.5% and +1.0%; Table 3).
The standard deviation of the mean change and the
coefficient of reproducibility (r51.966SD; %) are also
given in Table 3.
the mean difference
The in vitro reproducibility of the ADC measurement
was good, with a daily CV of less than 1% and a CV
of less than 1.5% over 3 months. Our results compare
favourably with those published by Delakis et al .
Although not stated, the CV for their more dilute CuSO4
phantom can be calculated to be 2.3% (b50 and b5500)
and 2.5% (b50 and b51000), whereas the CVs for their
sucrose solution are 3.6% and 3.5%, respectively. Delakis
et al  retrospectively normalised their results to a
reference temperature (21u uC), which may have contrib-
uted to the higher CV. Recently, Pierpaoli et al 
reported a CV of 2% for a polyvinylpyrrolidone phantom
and Chenevert et al [25, 26] reported a reproducibility
coefficient of 3–5% for iced water phantoms. The latter
also reported a spatial dependence along the z-axis .
In vivo apparent diffusion coefficient measurement
ADC values reported in the literature vary substan-
tially but the values reported here are similar to those
quoted by other authors using comparable protocols, in
particular the b-values. For the liver, reported values
range from 0.69 to 4.861023mm2s21(Table 1). How-
ever, our value of 1.04¡0.0561023mm2s21is very
similar to volunteer studies that have used a range of
b-values that exclude 0 and include high values, in
particular Mu ¨rtz et al  (between 1.03¡0.22 and
1.14¡0.4061023mm2s21) and Kim et al  (1.05¡
Diffusion in the kidneys was extensively reviewed by
Zhang et al , who reported a large variability in
published ADC values, ranging from 1.64¡0.09 to
4.07¡0.7461023mm2s21. Once more, the choice of b-
values has a considerable influence on the resulting ADC
values, and studies comparable with ours (encompas-
sing the whole renal parenchyma and including high
.700) all measured
than 261023mm2s21(range 1.64–1.9261023mm2s21),
similar to those found here (1.76¡0.0861023mm2s21).
ADC values for the gallbladder and the spleen are
also comparable with those reported in the literature
Table 2. Mean apparent diffusion coefficient (1023mm2s21) values, standard deviation and range of the different organs
calculated using the volume of interest method
Visit 1Visit 2
p-value ADCSD RangeADCSDRange
ADC, apparent diffusion coefficient; SD, standard deviation.
No significant difference was found between visits for any of the organs (p-value for a two-tailed matched-pair t-test).
Table 3. Paired analysis of the difference in apparent diffusion coefficient value between the two visits for the different organs
Volumes of interest analysisRegions of interest analysis
12.6 24.7 0.79
10.2 20.0 0.88
14.3 28.0 0.39
10.7 21.0 0.55
ADC, apparent diffusion coefficient; SD, standard deviation.
r5coefficient of reproducibility, p-value for two-tailed matched-pair t-test.
M E Miquel, A D Scott, N D MacDougall et al
1510 The British Journal of Radiology, November 2012
Apparent diffusion coefficient repeatability
For theVOIanalysis, thestandard deviation of the ADC
value in the spleen, liver and kidney was 4–5%, and
slightly lower in the gallbladder (Table 2). Only changes
greater than the coefficient of reproducibility would be
statistically significant. This implies that only changes in
mean ADC value over the organ greater than 6.4% for the
gallbladder, 7.7% for the kidney, 8.6% for the liver and
9.6% for the spleen would be statistically significant
(.1.966SD). For multiple two-dimensional ROIs, the
standard deviation of the mean of the percentage
difference between visits was much higher at over 10%
in all organs. This implies that, for a small ROI, only
changes .20.0% for the kidney, 21.0% for the gallbladder,
24.7% for the liver and 28.0% for the spleen would be
statistically significant. For multislice acquisitions, such as
those used here, variations in ADC values between slices
are often observed within an organ. These ADC variations
are less likely to affect three-dimensional volumes as any
differences between (and within) slices are likely to be
averaged over the large VOI. An example for the spleen is
shown in Figure 2. However, when using smaller two-
dimensional ROIs, less in-slice averaging and no interslice
averaging is present, which explains why the values
required for statistical significance are higher for ROI
analysis. Fluctuations in ADC values between slices were
more noticeable in the spleen, thus explaining its higher
(28%) coefficient of reproducibility. If we were to consider
all organs indiscriminately, only changes greater than
7.9% would be significant when considering VOIs and
greater than 23.1% when considering ROIs.
Although the coefficients of reproducibility for the ROI
analysis are larger than those for the VOI analysis, they
are in line with published literature. Kim et al 
suggested that changes of less than 30% fall within mea-
surement error for hepatic tumours, while Braithwaite et
al  consider that treatment changes of less than 27%
for abdominal ADC values at 3T ‘‘will not be clinically
detectable with confidence with one acquisition in a
single individual’’. In contrast, Messiou et al 
measured a coefficient of reproducibility of 14.8% in
bone marrow. This could be due to reduced patient
motion (respiratory and/or peristaltic) in the regions
analysed (L5 vertebral body and left iliac bone). Very
good reproducibility (4.8%) was recently presented in
metastatic and ovarian peritoneal tumours . It was,
however, somewhat mitigated by a high inter- and intra-
observer variability (11.4% and 13.7%, respectively).
Koh et al  reported a coefficient of reproducibility of
approximately 14% for solid tumours using a volume
analysis. Although this is greater than our average values,
it is nonetheless consistent; the volumes analysed in their
study were on average smaller than the ones used here
(apart from the gallbladder) and varied considerably
in size (106¡103cm3, range 10.0–400.9cm3). Both the
size and the position of lesions are known to influence
the reproducibility, with larger regions being more
In summary, the in vitro repeatability of ADC
measurements was good, with a coefficient of variation
of ,1.5%. In vivo variability was much higher and, if
considering a single ROI measurement in the abdomen,
Figure 2. Example of apparent dif-
fusion coefficient (ADC) measure-
ments through the spleen. (a) ADC
map of slice 5 through the spleen,
showing a region of interest (ROI)
used for analysis. (b) Anatomically
equivalent slice on the repeat visit.
(c) ADC values through the spleen
(black, visit 1; grey; visit 2). Solid
line, mean of the entire organ
(volume of interest analysis); dashed
lines, mean ADC value for the spleen
in this slice; individual points, mean
ADC value for the ROI.
Short communication: Repeatability of abdominal diffusion-weighted MRI
The British Journal of Radiology, November 2012 1511
only changes greater than 23.1% would be statistically
significant. This value is substantially lower (7.9%) if
volumes are considered.
1. Le Bihan D, Breton E, Lallemand D, Aubin ML, Vignaud J,
Laval-Jeantet M. Separation of diffusion and perfusion in
intravoxel incoherent motion MR imaging. Radiology 1988;
2. Le Bihan D. Molecular diffusion nuclear magnetic reso-
nance imaging. Magn Reson Q 1991;7:1–30.
3. Yoshikawa K, Nakata Y, Yamada K, Nakagawa M. Early
pathological changes in the parkinsonian brain demon-
strated by diffusion tensor MRI. J Neurol Neurosurg
4. Johnston KC, Wagner DP, Wang XQ, Newman GC, Thijs V,
Sen S, et al. Validation of an acute ischemic stroke model:
does diffusion-weighted imaging lesion volume offer a
clinically significant improvement in prediction of out-
come? Stroke 2007;38:1820–5.
5. Padhani AR, Liu G, Mu-Koh D, Chenevert TL, Thoeny HC,
Takahara T, et al. Diffusion-weighted magnetic resonance
imaging as a cancer biomarker: consensus and recommen-
dations. Neoplasia 2009;11:102–25.
6. Koh D-M, Collins DJ. Diffusion-weighted MRI in the
body: applications and challenges in Oncology. AJR Am J
7. Taouli B, Koh D-M. Diffusion-weighted MR imaging of the
liver. Radiology 2010;254:47–66.
8. Sugita R, Ito K, Fujita N, Takahashi S. Diffusion-weighted
MRI in abdominal oncology: clinical applications. World J
9. Zhang JL, Sigmund EE, Chandarana H, Rusinek H, Chen Q,
Vivier P-H, et al. Variability of renal apparent diffusion
coefficients: limitations of the mono-exponential model for
diffusion quantification. Radiology 2010;254:783–92.
10. Taouli B, Martin AJ, Qayyum A, Merriman RB, Vigneron D,
Yeh BM, Coakley FV. Parallel imaging and diffusion tensor
imaging for diffusion-weighted MRI of the liver: prelimin-
ary experience in healthy volunteers. AJR Am J Roentgenol
11. Mu ¨rtz P, Flacke S, Traber F, van den Brink JS, Gieseke J,
Schild HH. Abdomen: diffusion weighted MR imaging with
pulse-triggered single-shot sequences. Radiology 2002;224:
12. Kim T, Murakami T, Takahashi S, Hori M, Tsuda K,
Nakamura H. Diffusion-weighted single-shot echoplanar
MR imaging for liver disease. AJR Am J Roentgenol
13. Ichikawa T, Haradome H, Hachiya J, Nitatori T, Araki T.
Diffusion-weighted MR imaging with a single-shot echo-
planar sequence: detection and characterization of focal
hepatic lesions. AJR Am J Roentgenol 1998;170:397–402.
14. Taouli B, Vilgrain V, Dumont E, Daire JL, Fan B, Menu Y.
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.
15. Kwee TC, Takahara T, Koh D-M, Nievelstein RA, Luijten
PR. Comparison and reproducibility of ADC measurements
in breathhold, respiratory triggered, and free-breathing
diffusion-weighted MR imaging of the liver. J Magn Reson
16. Yamada I, Aung W, Himeno Y, Nakagawa T, Shibuya H.
Diffusion coefficients in abdominal organs and hepatic
lesions: evaluation with intravoxel incoherent motion echo-
planar MR imaging. Radiology 1999;210:617–23.
17. Mu ¨ller MF, Prasad P, Siewert B, Nissenbaum MA,
Raptopoulos V, Edelman RR. Abdominal diffusion map-
ping with use of a whole-body echo-planar system.
18. Namimoto T, Yamashita Y, Sumi S, Tang Y, Takahashi M.
Focal livermasses: characterization
weighted echo-planar MR imaging. Radiology 1997;204:
19. Damasio MB, Tagliafico A, Capaccio E, Cancelli C, Perrone
N, Tomolillo C, et al. Diffusion-weighted MRI sequences
(DW-MRI) of the kidney: normal findings, influence of
hydration state and repeatability of results. Radiol Med
20. Braithwaite AC, Dale BM, Boll DT, Merkle EM. Short- and
midterm reproducibility of apparent diffusion coefficient
measurements at 3.0-T diffusion-weighted imaging of the
abdomen. Radiology 2009;250:459–65.
21. Kim SY, Lee SS, Byun JH, Park SH, Kim JK, Park B, et al.
Malignant hepatic tumors: short-term reproducibility of
apparent diffusion coefficients
respiratory-triggered diffusion weighted MR imaging.
22. Koh D-M, Blackledge M, Collins DJ, Phadani AR, Wallace
T, Wilton B, et al. Reproducibility and changes in the
apparent diffusion coefficients of solid tumours treated
with combretastatin A4 phosphate and bevacizumab in a
two-centre phase I clinical trial. Eur Radiol 2009;19:2728–38.
23. Messiou C, Collins DJ, Morgan VA, deSouza NM.
Optimising diffusion weighted MRI for imaging metastatic
and myeloma bone disease and assessing reproducibility.
Eur Radiol 2011;21:1713–18.
24. Sasaki M, Yamada K, Watanabe Y, Matsui M, Ida M,
Fujiwara S, et al. Variability in absolute apparent diffusion
coefficient values across different platforms may be
substantial: a multivendor, multi-institutional comparison
study. Radiology 2008;249:624–30.
25. Chenevert TL, Galba ´n CJ, Londy FJ, Meyer CR, Johnson TD,
Rehemtulla A, et al. Icewater for quality control of diffusion
measurements in multi-center trials. Proc Intl Soc Magn
Reson Med 2011;19:844.
26. Chenevert TL, Galba ´n CJ, Ivancevic MK, Rohrer SE,
Londy FJ, Kwee TC, et al. Diffusion coefficient measure-
ment using a temperature-controlled fluid for quality
control in multicenter studies. J Magn Reson Imaging
27. Delakis I, Moore EM, Leach MO, De Wilde JP. Developing a
quality control protocol for diffusion imaging on a clinical
MRI system. Phys Med Biol 2004;49:1409–22.
28. Pierpaoli C, Sarlls J, Nevo U, Basser PJ, Horkay F.
Polyvinylpyrrolidone (PVP) water solutions as isotropic
phantoms for diffusion MRI studies. Proc Intl Soc Magn
Reson Med 2009;17:1414.
29. Kyriazi S, Collins DJ, Messiou C, Pennert K, Davidson RL,
Giles SL, et al. Metastatic ovarian and primary peritoneal
cancer: assessing chemotherapy response with diffusion-
weighted MR—value of histogram analysis of apparent
diffusion coefficients. Radiology 2011;261:182–9.
M E Miquel, A D Scott, N D MacDougall et al
1512The British Journal of Radiology, November 2012