1217

Magnetic Resonance
Imaging in the Evaluation of
Abscesses

Susan D. Wall1 Ten patients with percutaneous biopsy or surgically proven abscesses were evaluated
Madeleine R. Fisher with magnetic resonance imaging (MRI) to describe the appearance of abscesses,
Eugenio G. Amparo define the capability of MRI to localize abscesses, and compare the capabilfties of MRI
American Journal of Roentgenology 1985.144:1217-1221.

and CT for the diagnosis and determination of the extent of an abscess. Comparative
Hedvig Hricak
CT scans were available in six cases. The most common MRI finding was an abnormal
Charles B. Higgins
area of low signal intensity, either homogeneous or heterogeneous, on the short
repetition rate (500 msec TR) images with a relative increase in signal intensfty on the
longer repetition rate (1500 or 2000 msec TR) images. MRI demonstrated a more clear
delineation of the extent of inflammatory changes than did CT, and MRI demonstrated
the abscess as a collection distinct from surrounding structures on at least one repetition
rate. Intravenous contrast medium was unnecessary with MRI to evaluate vasculature
or to define the capsule around an abscess. With CT, unless an abscess contained air
or was of low attenuation, it often blended with the surrounding structures and was
difficuft to differentiate from them. Surgical clips in the postoperative patient with an
abscess did not degrade the MR images as often occurred with CT. This study describes
the MRI appearance of abscess and indicates a potential value of the use of MRI to
evaluate abscess outside the central nervous system and spine.

Signs and symptoms of abdominal abscesses are nonspecific and onset may be
insidious, making diagnosis and localization difficult. The mortality rate associated
with intraabdominab abscesses ranges from as high as 30% in treated patients to
as high as 80%-100% in untreated patients [1 -3]. Abscesses are often occult and
patients may present with few if any signs and symptoms [4]. Therefore, the index
of suspicion for the clinician must remain high and its detection by the radiologist
is important.
Several studies have evaluated the capabilities of various imaging methods in
the diagnosis of abscesses, including computed tomography (CT), sonography,
and radionuclide scans [5-1 5]. Compared to most imaging methods, magnetic
resonance imaging (MRI) has the advantages of improved contrast resolution for
soft tissues and the versatility of direct imaging in multiple planes. The capability
of MRI for the diagnosis of abscesses has not been evaluated. Therefore, the goals
of this study were to describe the appearance of abscesses on MRI, to define the
capability of MRI for the localization of abscesses, and to compare the capabilities
of MRI and CT for the diagnosis and determination of the extent of the abscesses.

Recsived October 3, 1984; accepted after revi- Subjects and Methods

sion January 7, 1985.
I All authors: Department of Radiology, (.kiiver- Patients
sity of Califomia School of Medicine, San Francisco,
CA 94143. Address reprint requests to S. D. Wall
Ten patients with percutaneously aspirated or surgically proven abscesses were studied.
(C-309). All were referred for MRI with the din’ical history of a probable abscess and in some instances
with a positive CT scan. Therefore, sensitivity and specificity cannot be assessed in this
AJR 144:1217-1221, June 1985
0361 -803X/85/1 446-1 217 study. Nine patients manifested the three cardinal clinical features of an abscess, namely
C American Roentgen Ray Society fever, leukocytosis, and localizing pain. The tenth patient was receiving corticosteroid therapy
 1218 WALL ET AL. AJR:144, June 1985

and had an occult abscess. Four patients were women and six were tions showed that the abscesses adjacent to fat showed the
men; their ages ranged from 24 to 68 years. The abscesses were greatest contrast with adjacent structures on the SE 500/28
located within the abdomen in eight patients, in the inguinal region in (TR/TE) images. Those abscesses adjacent to muscle were
one, and in the axilla in another. MRI was performed before aspiration
best seen and showed greatest contrast on images obtained
in all cases. A positive CT study was available before MRI in five
with 1 500 msec or 2000 msec TR and 56 msec TE. In general,
patients. The sixth CT study was interpreted as negative; in retro-
the abscesses appeared as abnormal areas of low signal
spect, an abnormal area in the site of proven abscess was present.
intensity on the short-TR (500 msec) images and had a
relatively higher signal intensity on the longer-TR (2000 or
MR Imager and Technique 1500 msec) images (fig. 1). The abscesses in our study had
MRI was performed with a Diasonics MT/S system, which is a longer mean Ti and mean T2 (1 1 07 and 81 msec, respec-
superconducting magnet with a field strength of 0.5 T, operating at tively) values compared to fat (mean Ti = 243 msec, mean
0.35 T (proton resonance frequency of 15 MHz). The spin-echo (SE) T2 = 55 msec), muscle (mean Ti = 528 msec, mean T2 =
imaging technique was used with echo delay times (TE) of 28 and 56 29 msec), and normal liver (mean Ti = 377 msec, mean T2
msec for the first and second echo images, respectively. All patients = 45 msec), but shorter than those of urine (mean Ti = 2964
were imaged in the transaxial plane with a long (2000 or 1500 msec) msec, mean T2 = i 66 msec). In five patients, the abscess
and a short (1 000 or 500 msec) repetition rate (TA). Four patients collection was heterogeneous in signal intensity (fig. 2) and in
were imaged additionally in the direct coronal and sagittal planes. five patients, it was homogeneous (fig. 3) (low signal intensity)
on the shorter TR. The heterogeneous pattern of signal
CT Scans intensity was more apparent in each case on images using
longer TR (1 500 or 2000 msec) and TE (56 msec). A capsule
American Journal of Roentgenology 1985.144:1217-1221.

Six subjects had correlative CT studies, performed with a GE 8800

surrounding the abscess was identified in four patients and
scanner using intravenous and oral contrast media. The diagnosis
was of low signal intensity on both long and short TRs, but
was confirmed by surgery or percutaneous aspiration in all patients.
Drainage procedures were performed after radiographic imaging. often it was better visualized on the Ionger-TR and bonger-TE
images. Gas was seen as an absence of signal on both long
and short TRs in the abscesses of two patients.
MRI Data Analysis

The MRI data collected for each abscess included site, size,
Effect of Adjacent Structures
homogeneity, presence of a capsule or gas, intensity changes within
the abscess with different repetition rates, and the effect on surround- In four patients, the abscess was adjacent to the aorta,
ing tissues. Ti and T2 relaxation times were calculated and compared inferior vena cava, or other large vessels. In each case,
to normal values for fat, muscle, liver, and urine. Percentage of displacement or compression of these vessels was identified
contrast for the abscess and surrounding structures was calculated

-
from the intensity measurements
contrast = [(li
using the following equation: %
l)/l] x 100, where l is the intensity of the abscess,
and l is the intensity of the surrounding structure. Fat and muscle
on MRI because their recognition
luminal
Ii 6]. Inflammatory involvement
was facilitated

of surrounding
by the intra-
signal void from rapid laminar flow within them (fig. 4)
muscles was
were used as the surrounding structures. The MA images were detected on MRI by the finding of abnormally high signal
analyzed independently by three observers for the above factors. intensity in focal areas of adjacent muscles (fig. 5). Intestinal
To measure Ti relaxation times, two scanning sequences are involvement was demonstrated by thickening of adjacent
necessary. We use a short (500 or 1000 msec) and a long (1500 or bowel wall (fig. i ). Bone destruction, present in one case,
2000 msec) TA. To measure the T2 relaxation time, only one scanning was identified by disruption of the normally low signal intensity
sequence is required, preferably the longest TA (2000 msec), but two of cortical bone and by replacement of the normally high
TEs are necessary. The region of interest for calculation of these
signal intensity of bone marrow with a heterogeneous pattern
factors was specifically selected to include greater than 20 pixels,
of high and low signal intensity (fig. 5). Multiplanar imaging
because less reliable values (larger standard deviation) for Ti and T2
was useful in precise localization of two abscesses (figs. 3
times were obtained when a smaller region of interest was used. In
addition, at least three region-of-interest measurements were ob- and 5).
tamed for improved accuracy of the measured Ti and T2 values. Ti
measurements were computer-generated by integrating the signal
Comparison of MA! and CT Findings
intensity images of a selected region of interest on images obtained
at two different TAs (500 and 1500 or 500 and 2000 msec). T2 A comparative CT scan was available in six cases, and the
measurements were computer-generated by integrating the first and
results are summarized in table i In two of these cases, the
.
second SE images having the longer TA (2000 msec). The MR and
abscesses were seen as areas of homogeneous soft-tissue
CT scans were analyzed separately for the previously mentioned
data factors, and subsequently the findings were compared. density (fig. 2). The remaining four abscesses had a CT
density lower than that of the surrounding soft tissues. Of the
four patients found to have a capsule on MRI, three had a
Results correlative CT scan. Of these three, none had an identifiable
capsule on CT, even after intravenous administration of con-
MRI Appearance with Variations in TR/TE
trast medium. Gas was identified on the CT scans in the
The abscess appearance on MRI was dependent on TR same two patients in whom it was seen with MRI. In those
and TE imaging factors. Percentage of contrast determina- patients whose abscess was adjacent to vasculature as seen
 AJR:144, June 1985 MRI OF ABSCESSES 1 2i 9

Fig. 1 -A, SE 500/28. On short TR, periappen-

diceal abscess (arrows) adjacent to anterolateral
aspect of cecum is seen as heterogeneous area of
low signal intensity. B, SE 2000/28. On long TR,
abscess (straight arrows) has higher signal and is
well delineated from adjacent colon. Note small-
bowel loops anteriorly (curved arrow) and thicken-
ing of wall of cecum (C, arrowhead).

Fig. 2.-A, SE 500/28. On short TR, axillary

American Journal of Roentgenology 1985.144:1217-1221.

abscess is seen as area of heterogeneous low

signal intensity (arrow). Area of high signal intensity
surrounding abscess is compressed fat in axilla. B,
SE 2000/28. On long TR, abscess appears more
heterogeneous and is of higher signal intensity
(straight arrow). Air is present within superior as-
pect of abscess (curved arrow).

Fig. 3.-A, SE 500/28. Peridiverticular abscess is demonstrated adjacent signal intensity with prolongation of TR and TE. C, SE 1500/28. Coronal image
to sigmoid colon (5) and superomedial to transplant kidney within left iliac aids in localizing abscess (arrow), which is adjacent to lateral aspect of sigmoid
fossa. On short TR, abscess is seen as homogeneous, low-signal-intensity colon (5) and superomedial to transplant kidney.
collection (arrow). B, SE 2000/56. Abscess (arrow) has relatively increased
 1220 WALL ET AL. AJR:144, June 1985

on MRI, two had CT scans; in one case, the vessels were and suggested that it was the abscess in question in each of
identified only after intravenous administration of contrast the i 0 patients studied. CT identified an abnormal area that
medium. Involvement of adjacent musculature, demonstrated appeared to be an abscess in five of six patients.
in five cases with MRI, and bowel, shown in one case with
MRI, was not well delineated on CT. However, bone destruc-
Discussion
tion was better seen with CT in the one case where it occurred
(fig. 5). MRI delineated an abnormal area of signal intensity The results of this study indicate that MRI is capable of
demonstrating abscess collections, and that there is potential
value in the use of MRI to evaluate abscess outside the
central nervous system and spine. The MRI appearance of
the abscesses reviewed in the study was that of low signal
intensity, either homogeneous or heterogeneous, on the
short-TA images, with a relatively higher signal intensity on
the longer-TA images. However, this pattern is not specific
for abscesses. The ability to image directly in the sagittal and
coronal planes aided in the precise localization of these col-

TABLE 1 : Comparison of CT and MRI Findings

Fact CT MRI
American Journal of Roentgenology 1985.144:1217-1221.

Clarity of identification 4/6 10/i 0

Internal constituents:
Capsule 0/3 5/5
Gas 2/2 2/2
Effect on adjacent structures:
Vasculature 1/2 4/4
Fig. 4.-SE 2000/28. Abnormal area of increased signal intensity is present Muscle 0/4 5/5
within psoas muscles (solid arrows) and within erector spinae muscles (curved Bone i/i i/i
arrows) bilaterally. In right psoas. part of low-intensity capsule (arrowhead) is Extent of abscess 3/6 10/i 0
seen anterloily where it surrounds abscess. Similar high-intensity collections
displace aorta and inferior vena cava anteriorly (open arrow). At surgery, Note-Data reported as number of cases recognized by the imaging method for the
abscesses were found in each of these areas. factor/number of cases imaged that had the factor.

Fig. 5.-Patient with Crohn disease. A, SE 500/

56, prone position.
Abnormal area of heteroge-
neous low signal intensity (solid arrows), which is
slightly higher in intensity than right gluteus maxi-
mus muscle. This abscess extended into sacrum
and presacral area on left and exited pelvis through
left sacroiliac notch. There is thickening of left lateral
aspect of rectal wall (curved arrow). Bone destruc-
tin is suggested by loss of definition of low signal
intensity from cortical bone (open arrow). Bone
destruction is suggested by loss of definition of low
signal intensity from cortical bone (open arrow) and
replacement of high signal intensity from bone mar-
row by inhomogeneous signal intensity. B, SE
2000/28. High signal intensity within right gluteus
muscle (solid arrows), which may represent either
edema or infection. Thickening of left lateral aspect
of rectal wall is more obvious on this image (open
arrow). C, SE 2000/28, sagittal plane. Sacral de-
struction with surrounding high-intensity signal of
abscess (solid arrows) and part of cutaneorectal
fistula (open arrow). D, Axial CT image demon-
strates extent of sacral destruction but fails to show
entire extent of involvement within surrounding
muscles and presacral region.
 AJR:144, June 1985 MRI OF ABSCESSES i 22i

lections, particularly with the one gluteal-sacral abscess and collections before one may describe a characteristic pattern
the one diverticular abscess in this series. Additional signs of for each. Thus, although MAI seems capable of accurately
abscesses demonstrated with MRI were displacement of localizing, evaluating the extent of, and suggesting the diag-
adjacent vasculature and organs by the lesion and increased nosis of abscess when clinically questioned, the diagnosis of
signal intensity within the involved adjacent musculature. The these lesions on the basis of their pattern of signal intensity
cause of increased signal intensity within the muscle cannot will need further study.
be definitely ascertained with MRI and may represent either
edema or infection. Contrast enhancement was not necessary
for evaluation of a capsule or assessment of surrounding REFERENCES
vasculature. 1 Reed RP, Masters
. Si. Intra-abdominal abscess. Semin Roent-
Comparison of MRI and CT revealed several advantages genol 1973;8:365-374
and disadvantages of each in the evaluation of abscesses. 2. Connell TA, Stephens DH, Carison HC, Brown ML. Upper ab-
There was better delineation of the extent of inflammatory dominal abscess: a continuing and deadly problem. AiR
change on surrounding tissue with MRI as demonstrated, for 1980;i 34:759-765
instance, by the abnormally high signal intensity offocal areas 3. Ferrucci JT, vanSonnenberg E. Intra-abdominal abscess radio-
logical diagnosisand treatment. JAMA 1981;246:2728-2733
of adjacent muscles, regardless of TR or TE. On CT the
4. Goldman A, Hunter TB, Haber K. The silent abdominal abscess:
involvement of adjacent muscles was inconsistently demon-
role of the radiologist. AJR 1983;14i :21-25
strated. When apparent, affected musculature was seen to
5. Doust BD, Quiroz F, Stewart JM. tJtrasonic distinction of ab-
be enlarged, but no distinct alteration of attenuation of the x- scesses from other intra-abdominal fluid collections. Radiology
ray beam was present. In MRI, intravenous contrast medium 1977;i25:2i3-2i8
American Journal of Roentgenology 1985.144:1217-1221.

was not necessary to evaluate adjacent vasculature or to 6. Koehler PA, Moss AA. Diagnosis of the intra-abdominal and
define the capsule around an abscess. Occasionally these pelvic abscesses by computerized tomography. JAMA
definitive observations required contrast media on the CT 1980;244:49-52
studies. MRI demonstrated the abscess as a distinct collec- 7. Knochel JO, Koehler PR, Lee TG, Welch DM. Diagnosis of
tion, separate from adjacent soft tissues. However, with CT, abdominal abscesses with computed tomography, ultrasound,
and11 i-In leukocyte scans. Radiology 1980;i37:425-432
ifthe abscess failed to contain gas or did not have a decreased
8. Halber MD, Daffner RH, Morgan CL, et al. Intraabdominal ab-
attenuation number [3, 6-8, i 0, 1 i i 7], it blended with the
,
scess: current concepts in radiologic evaluation. AIR
surrounding muscle tissue or unopacified gastrointestinal 1979;i 33:9-i 3
tract making the diagnosis difficult. Finally, MRI was useful in 9. Levitt AG, Biello DR, Sagel 55, et al. Computed tomography
the evaluation of the postoperative patient, because surgical and 67Ga citrate radionuclide imaging for evaluating suspected
clips did not produce artifacts as they did on CT images. abdominal abscess. AIR 1979;i32:529-534
Several disadvantages of MRI were recognized. The first 10. Korobkin M, Callen PW, Filly A, Hoffer PB, Shimshak AR, Kressel
was the difficulty in identification of calcification. This was HY. Comparison ofcomputed tomography, ultrasonography, and
important in the cases of periaortic graft abscesses where it gallium-67 scanning in the evaluation of suspected abdominal
was difficult to differentiate aortic wall calcification from an abscess. Radiology 1978;1 29:89-93
1 1 . Kressel HY, Filly A. Ultrasonographic appearance of gas-contain-
abscess capsule, both seen as a low-intensity signal. In
ing abscesses in the abdomen. AiR 1978;i30:7i-73
addition, the extent of bone destruction was not as well
12. Whitley NO, Shatney CH. Diagnosis of abdominal abscesses
delineated on MRI as on CT. Also, both methods are limited
with major trauma; the use of computed tomography. Radiology
by the problem of differentiating unopacified bowel from an 1983;i 47: 179-i 83
abscess collection. At this time, the correlation with CT has 13. CaIlen PW. Computed tomographic evaluation of abdominal and
yielded several advantages and disadvantages of each, but pelvic abscesses. Radiology 1979;1 31 :171-175
there were too few cases in our study to allow firm conclu- 14. Halvorsen RA, Jones MA, Rice RP, Thompson WM. Anterior left
sions to be drawn as to the relative merits of the two imaging subphrenic abscess: characteristic plain film and CT appearance.
methods. AJR 1982;i 39:283-289
The sensitivity and specificity of the MRI abscess pattern 15. Chintapalli K, Thorsen MK, Foley WD, Unger GF. Abdominal
abscesses with enteric communications: CT findings. AiR
remains to be determined. At present, there are insufficient
1983;i41 :27-28
data to conclude how an abscess on MAI may differ in
i 6. Higgins CB, Goldberg HI, Hricak H, Crooks LE, Kaufman L.
appearance depending on whether it is a sterile or infected Nuclear magnetic resonance imaging of vasculature of abdominal
collection. Also, in order to differentiate abscesses from other viscera: normal and pathologic features. AJR 1983;140:12i7-
fluid collections and necrotic tumors, more work is necessary. 1224
Our preliminary results show there is a need for further 17. Sones PH. Percutaneous drainage of abdominal abscesses. AIR
evaluation of the characteristics of abscesses and other fluid 1984;i42:35-39