Professional Documents
Culture Documents
Desbarats L.N. Differential MRI Diagnosis Between Brain Abscesses and Necrotic or Cystic Brain Tumors Using The Apparent Diffusion Coefficient and Normalized Diffusion Weighted Images 2003
Desbarats L.N. Differential MRI Diagnosis Between Brain Abscesses and Necrotic or Cystic Brain Tumors Using The Apparent Diffusion Coefficient and Normalized Diffusion Weighted Images 2003
net/publication/10616795
CITATIONS READS
79 3,045
7 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Giovanni de Marco on 05 June 2018.
Abstract
Magnetic Resonance Diffusion-Weighted Imaging (DWI) has been reported to be helpful for the differential diagnosis between abscesses
and cystic/necrotic brain tumors. However the number of patients is still limited, and the sensitivity and specificity of the method remain
to be confirmed. The primary purpose of this study was to investigate a larger sample of patients, all investigated under the same
experimental conditions, in order to obtain statistically significant data. Moreover, there is no consensus about the appropriate values of b
required to use to make an accurate diagnosis from DWI. The secondary purpose of this study was to determine the discriminating threshold
b values for raw diffusion-weighted images and for normalized diffusion-weighted images. On the basis of 14 abscesses, 10 high-grade
gliomas and 2 metastases, we show that the calculation of accurate Apparent Diffusion Coefficient (ADC) values gives a specificity rate of
100%. Without ADC calculation, we show that image normalization is required to make an accurate differential diagnosis, and we highlight
the ability of DWI to discriminate between brain abscesses and cystic/necrotic brain tumors using normalized signal intensity at lower b
values (503 s/mm2) than usual. © 2003 Elsevier Inc. All rights reserved.
Keywords: ADC; Signal intensity; Diffusion-weighted images; MRI; Brain abscess; Necrotic or cystic tumor
0730-725X/03/$ – see front matter © 2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0730-725X(03)00084-5
646 L. Nadal Desbarats et al. / Magnetic Resonance Imaging 21 (2003) 645– 650
Fig. 1. A 41-year-old woman with a surgically proven left parietal pyogenic brain abscess. a) T2-weighted image, b) precontrast T1-weighted image, c)
contrast-enhanced T1 image, d) diffusion-weighted image b ⫽ 80 s/mm2, e) diffusion-weighted image b ⫽ 724 s/mm2.
Fig. 2. A 61-year-old man with a pathologically proven right frontal malignant glioma a) T2-weighted image, b) precontrast T1-weighted image, c)
contrast-enhanced T1 image, d) diffusion-weighted image b ⫽ 80 s/mm2, e) diffusion-weighted image b ⫽ 724 s/mm2.
L. Nadal Desbarats et al. / Magnetic Resonance Imaging 21 (2003) 645– 650 647
Fig. 4a. Raw signal intensities on DWI from ROIs chosen in the central cavities of lesions versus b values. White vertical boxes and E for Abscesses, Gray
vertical boxes and F for Cystic/Necrotic tumors. The vertical box represents values between the 25th and 75th percentiles, the 5th and 95th percentiles are
displayed as symbols. n is the number of patients and N the number of measurements.
3.3. DWI means from T1- and T2-weighted images can differentiate
Raw signal intensities versus b values for abscesses and between abscesses and cystic/necrotic tumors. However, if
cystic/necrotic tumors are shown in Fig. 4a. The overlap an early and accurate diagnosis can be made, antibiotics can
between the signal intensities of the two groups, persists up be used successfully to treat brain abscesses.
to b values of 986 s/mm2. Consequently, diagnoses based on The literature includes reports of 1H MR spectroscopy
raw DWI images may lead to the wrong conclusion. studies used to characterize [10 –13] and classify brain tu-
Normalized signal intensities versus b values, for ab- mors [14 –16] and of their ability to provide differing spec-
scesses and cystic/necrotic tumors, are shown in Fig. 4b. For tral patterns for abscesses and for cystic or necrotic lesions.
normalized signal intensities, low b values such as 322 Some in vivo localized studies [17–22] describe the pres-
s/mm2 could provide an accurate diagnosis, and there was ence of end-products of bacterial break-down, such as ami-
no overlap between the normalized signal intensities of the no-acid residues, along with lactate and acetate in abscesses,
two groups at the higher b values. Using normalized DWI, whereas only lactate is present in high grade gliomas. More
it is possible to increase the accuracy of the diagnosis even recently, Grand et al. [23], have compared in vivo localized
at lower b values than usual. spectroscopy and in vitro spectroscopy, and have shown
differential spectroscopic patterns in both types of lesion
4. Discussion based on the quantification of the amino-acids. This tech-
It can be difficult to reach a differential diagnosis of nique is certainly attractive, but not available at every im-
ring-enhancing intracranial mass lesions on the basis of aging center.
clinical symptoms and conventional MRI. We found that Diffusion-weighted MRI provides images in which the
neither visual inspection nor measuring signal intensity contrast is dependent on the molecular motion of water [1]
Fig. 4b. Normalized signal intensities on DWI from ROIs chosen in the centers of lesions versus b values. White vertical boxes and E for Abscesses, Gray
vertical boxes and F for Cystic/Necrotic tumors. Vertical box represents values between the 25th and 75th percentiles, the 5th and 95th percentiles are
displayed as symbols: n is the number of patients and N the number of measurements.
L. Nadal Desbarats et al. / Magnetic Resonance Imaging 21 (2003) 645– 650 649
and may be altered as a result of disease. The clinical clinical software uses 2 b values to calculate the ADC and
potential of diffusion-weighted imaging is likely to be great- we must be very cautious about interpreting these data if we
est in the study of acute stroke, both for the early detection are to obtain a reliable diagnosis. Our findings show low
of cerebral infarcts and for monitoring subsequent ischemic ADC values in abscesses, and these low ADC values are
damage. The clinical use of the technique is now becoming attributable to the presence of pus [6]. Pus is a highly
routine for the diagnosis of stroke [24]. Another potential viscous, thick, mucoid fluid consisting of inflammatory
application is mapping the development and maturation of cells, bacteria, proteoneous exudate and fibrinogen [9]. Be-
white matter [25]. More recently, DWI has also been ap- cause of this high viscosity, diffusion water motion is se-
plied to brain tumor processes [4]. verely curtailed. The high ADC values we find in cystic or
Functional MRI, like diffusion-weighted imaging, has necrotic lesions are attributable to an intra-cavity fluid that
occasionally been used to diagnose cerebral abscesses. is less viscous than that found in abscesses. It consists of
However, so far only a small number of abscesses have been necrotic tissue debris and contains fewer inflammatory cells
reported, 5 papers report a total of 17 abscesses. In one case than abscess fluid. However, Holtas et al. [27] have recently
report, Ebisu et al. [6] showed that diffusion weighted reported a low ADC in a ring-enhancing metastasis. Parks et
image had high signal intensity in a brain abscess. Kim et al. al. [28] reported 7 metastases, 2 of which had a low ADC.
[7] reported a markedly hyperintense signal in abscesses, Holtas et al. suggest that early stage tumor necrosis with
whereas necrotic or cystic brain tumors gave a hypointense intracellular edema, but as yet no liquefaction, may give
signal. These studies were based on signal intensity images, DWI findings similar to those for abscesses at the capsule
rather than on the calculated apparent diffusion coefficient. stage. In our case, the metastases were probably in the
More recently, Desprechins et al. [9] showed that it is liquefaction stage and similar to high-grade gliomas. In our
possible to reach a differential diagnosis based on the high study, the metastases and the high-grade gliomas behaved
signal intensity in diffusion imaging and a strongly reduced similarly. They both had a high ADC, and so we included
apparent diffusion coefficient. Noguchi et al. [8], investi- both these processes in a single group.
gating 2 pyogenic brain abscesses and 6 high-grade gliomas, High DWI signal intensity for brain abscesses is usually
reported a significant difference between the ADC values, associated with a reduced ADC value and, conversely, low
but underline a potential methodological problem, since an DWI signal intensity is associated with a high ADC value
inaccurate ADC value could result if only two b values are for cystic or necrotic tumors, including metastases [8,28].
used. Our results closely match these observations. Another goal
So far, depending on the authors, publications report the of this study was to find out what b value can be used as the
use of 2 or more b values to calculate ADC, and in addition cutoff threshold between abscesses and necrotic or cystic
there is no agreement about the appropriate b value to use to tumors for raw DWI signal intensity and for normalized
obtain the most accurate diagnosis. Because of the low DWI signal intensity.
incidence of abscesses, the preliminary studies involved Up to b ⫽ 986 s/mm2, we found that the raw DWI signal
only small numbers of patients [6,8,9], however, we were intensities for abscess and cystic or necrotic tumors overlap.
able to use a larger sample, which allowed us to detect a We have to take into consideration that DWI signal intensity
statistically significant difference between the groups. is influenced not only by water diffusion, but also by the
We used 8 b values to calculate accurate ADC values and intrinsic T2 property of the tissue being examined (the T2
found that those for white matter were statistically different shine through effect) [29]. At high b values, the T2 effect is
in the abscess and high-grade glioma groups. This discrep- presumably negligible, but in clinical practice, considering
ancy may be attributable to the differing locations of the only DWI obtained with high b values, often 1000 s/mm2,
slices containing lesions. The ROIs in the white matter were can lead to a mistaken diagnosis. This is similar to the
not always at same spot, because we centered the 7 slices for situation for stroke. Interpreting DW images without ADC
diffusion-weighted imaging on the lesion. The anisotropy calculations may lead to the wrong interpretation of stroke
and non-homogeneous composition of the white matter can data with respect to stroke evolution [30,31]. On the other
account for some of the statistical difference between the hand, working with normalized signal intensities makes it
two groups, although there is a huge degree of overlap possible to eliminate the T2 shine through effect. In our
between the two ranges of values. For the central cavity of study of normalized images, starting from b ⫽ 503 s/mm2,
both lesion groups, the statistical difference was accompa- the overlap between the two ranges of signal intensities
nied by an absence of any overlap between the two ranges disappears. This could make it possible to carry out diffu-
of ADC values. This made it possible to determine a thresh- sion weighted MRI at lower b values and/or lower gradient
old CDA value below which a lesion can be classified as an strength. According to these results, image normalization is
abscess, with a specificity rate of 100%. In this study, this required to interpret the diffusion-weighted images properly
threshold value is 1.10 ⫻ 10⫺3 mm2/s. Recently, Burdette et and to avoid false negative interpretations. Commercially
al. [26] have shown that brain ADC values may be slightly available clinical software is based on 2 b values: b ⫽ 0
overestimated if a two-point technique is used instead of a s/mm2 and 1000 s/mm2. This means that it is necessary to
six-point technique. Nowadays, commercially available analyze normalized images before making a diagnosis, but
650 L. Nadal Desbarats et al. / Magnetic Resonance Imaging 21 (2003) 645– 650
post-processing software does not offer the option of rapidly [11] Kuesel AC, Sutherland GR, Halliday W, Smith IC. 1H MRS of high
normalizing the signal intensity of the images. grade astrocytomas: mobile lipid accumulation in necrotic tissue.
NMR Biomed 1994;7:149 –55.
[12] Negendank W. Studies of human tumors by MRS: a review. NMR
5. Conclusion Biomed 1992;5:303–24.
It is often impossible to reach a differential diagnosis [13] Remy C, Arus C, Ziegler A, Lai ES, Moreno A, Le Fur Y, Decorps
between brain abscesses and cystic or necrotic brain tumors M. In vivo, ex vivo, and in vitro one- and two-dimensional nuclear
magnetic resonance spectroscopy of an intracerebral glioma in rat
using conventional MRI. DWI is a useful tool to explore
brain: assignment of resonances. J Neurochem 1994;62:166 –79.
water diffusion and to make this differential diagnosis pos- [14] Hagberg G, Burlina AP, Mader I, Roser W, Radue EW, Seelig J. In
sible. Working on the raw DWI signal intensity, even at vivo proton MR spectroscopy of human gliomas: definition of met-
high b values, could lead to an erroneous clinical interpre- abolic coordinates for multi-dimensional classification. Magn Reson
tation. Image normalization seems to be a better way, mak- Med 1995;34:242–52.
[15] Howells SL, Maxwell RJ, Griffiths JR. Classification of tumour 1H
ing it possible to avoid this T2 shine-through phenomenon
NMR spectra by pattern recognition. NMR Biomed 1992;5:59 – 64.
and to reach an accurate diagnosis even at low b values. [16] Nadal L, Leray G, Desbarats C, Darcel F, Bansard JY, Bondon A, de
Unfortunately, no commercially available software allows Certaines JD. Proton and phosphorus nuclear magnetic resonance
to normalize images automatically. This means that to avoid spectroscopy of human brain tumor extracts with automatic data
the T2 shine through phenomenon, we need to calculate classification: a preliminary study. Cell Mol Biol 1997;43:659 –73.
[17] Ott D, Henning J, Ernst T. Human brain tumors: assessment with in
ADC values. Our study included the largest number of
vivo proton MR spectroscopy. Radiology 1993;186:745–57.
abscesses so far reported, and we calculated an accurate [18] Poptani H, Gupta R, Jain V, Roy R, Pandey R. Cystic intracranial
ADC using 8 b values. This allowed us to propose a thresh- mass lesions: possible role of in vivo MR spectroscopy in its differ-
old ADC value. ential diagnosis. Magn Reson Imaging 1995;13:1019 –29.
[19] Poptani H, Gupta R, Roy R, Pandey R, Jain V, Chhabra D. Charac-
terization of intracranial mass lesions with in vivo proton MR spec-
troscopy. AJNR Am J Neuroradiol 1995;16:1593– 603.
References [20] Grand S, Sam-Lai E, Esteve F, Rubin C, Hoffmann D, Rémy C,
Segebarth C. In vivo 1H MRS of brain abscesses versus necrotiv brain
[1] Le Bihan D. Diffusion and perfusion magnetic resonance imaging. tumors. Neurology 1996;47:846 – 8.
Applications to functional MRI. New York: Raven Press, 1995. [21] Kim S, Chang K, Song I, Han M, Kim H, Kang H, Han M. Brain
[2] Marks MP, de Crespingy A, Lentz D, Enzmann DR, Albers GW, abscess and brain tumor: discrimination with in vivo H-1 MR Spec-
Moseley ME. Acute and chronic stroke: navigated spin-echo diffu- troscopy. Radiology 1997;204:239 – 45.
sion-weighted MR imaging. Radiology 1996;199:403– 8. [22] Burtscher IM, Holtas S. In vivo proton MR spectroscopy of untreated
[3] Tsuruda J, Chew W, Moseley M, Norman D. Diffusion-weighted MR and treated brain abscesses. AJNR Am J Neuroradiol 1999;20:1049 –
imaging of the brain: value of differentiating between extraaxial cysts 53.
and epidermoid tumors. AJR Am J Roentgenol 1990;155:1059 – 68. [23] Grand S, Passaro G, Ziegler A, Esteve F, Boujet C, Hoffmann D,
[4] Tien R, Felsberg G, Friedman H, Brown M, MacFall J. MR imaging Rubin C, Segebarth C, Décorps M, Le Bas J, Rémy C. Necrotic tumor
of high-grade cerebral gliomas: value of diffusion-weighted echop- versus brain abscess: importance of amino-acids detected at 1H MR
lanar pulse sequences. AJR Am J Roentgenol 1994;162:671–7. spectroscopy—initial results;Radiology 1999;213:785–93.
[5] Maeda M, Kawamura Y, Tamagawa Y, Matsuda T, Itoh S, Kimura H, [24] Warach S, Chien D, Li W, Ronthal M, Edelman R. Fast magnetic
Iwasaki T, Hayashi N, Yamamoto K, Ishii Y. Intravoxel incoherent resonance diffusion-weighted imaging of acute human stroke. Neu-
motion (IVIM) MRI in intracranial, extraaxial tumors and cysts. rology 1992;42:1717–23.
J Comput Assist Tomogr 1992;16:514 – 8. [25] Chenevert TL, Brunberg JA, Pipe JG. Anisotrophic diffusion within
[6] Ebisu T, Tanaka C, Umeda M, Kitamura M, Naruse S, Higuchi T, human white matter: demonstration with NMR techniques in vivo.
Ueda S, Sato H. Discrimination of brain abscess from necrotic or Radiology 1990;177:401–5.
cystic tumors by diffusion-weighted echo-planar imaging. Magn Re- [26] Burdette J, Elster A, Ricci P. Calculation of apparent diffusion coef-
son Imaging 1996;14:1113– 6. ficients (ADCs) in brain using two-point and six-point methods.
[7] Kim Y, Chang K, Song I, Kim H, Seong S, Kim Y, Han M. Brain J Comput Assist Tomogr 1998;22:792– 4.
abscess and necrotic or cystic brain tumor: discrimination with signal [27] Holtas S, Geijer B, Stromblad L, Maly-Sundgren P, Burtscher I. A
intensity on diffusion-weighted MR imaging. AJR Am J Roentgenol ring-enhancing metastasis with central high signal on diffusion-
1998;171:1487–90. weighted imaging and low apparent diffusion coefficient. Neuroradi-
[8] Noguchi K, Watanabe N, Nagayoshi T, Kanazawa T, Shimizu M, ology 2000;42:824 –7.
Seto H. Role of diffusion-weighted echo-planar MRI in distinguish- [28] Park SH, Chang KH, Song IC, Kim YJ, Kim SH, Han MH. Diffusion-
ing between brain abscess and tumor: a preliminary report. Neurora- weighted MRI in cystic or necrotic intracranial lesions. Neuroradiol-
diology 1999;41:171– 4. ogy 2000;42:716 –21.
[9] Desprechins B, Stadnik T, Koerts G, Shabana W, Breucq C, Osteaux [29] Burdette J, Elster A, Ricci P. Acute cerebral infraction: quantification
M. Use of diffusion-weighted MR Imaging in differential diagnosis of spin-density and T2 shine-through phenomena on diffusion-
between intracerebral necrotic tumors and cerebral abscesses. AJNR weighted MR images. Radiology 1998;212:333–9.
Am J Neuroradiol 1999;20:1252–7. [30] Ulug A, Beauchamp N, Bryan R, PC v.Z. Absolute quantification of
[10] Kinoshita Y, Kajiwara H, Yokota A, Koga Y. Proton magnetic diffusion constants in human stroke. Stroke 1997;28:483–90.
resonance spectroscopy of brain tumors: an in vitro study. Neurosur- [31] Beauchamp N, Ulug A, Passe T, van Zijl P. MR diffusion imaging in
gery 1994;35:606 –14. stroke: review and controversies. Radiographics 1998;18:1269 – 83.