1235

Magnetic Resonance
Imaging of the Adrenal
Glands: A Comparison with
Downloaded from ajronline.org by 103.169.227.219 on 02/06/24 from IP address 103.169.227.219. Copyright ARRS. For personal use only; all rights reserved

Computed Tomography

Carl L. Schultz1 This investigation compared magnetic resonance imaging (MRI) with computed to-
John R. Haaga1 mography (CT) in the evaluation of normal and abnormal adrenal glands. Thirty normal
Barry 0. Fletcher1 volunteers were studied with MRI, and the results were compared with a retrospective
review of 30 normal CT examinations. CT identified both adrenal glands in all 30 patients.
RalphJ. AIfidi1
MRI identified both glands in 29 of 30 volunteers. There were no statistically significant
Mary A. Schultz2
differences between the two imaging techniques using chi-square analysis. Twenty-one
patients with abnormal adrenal gland(s) detected with CT were also studied with MRI.
The abnormalities studied included bilateral hyperplasia (three patients), adenoma (two),
myelolipoma (one), adrenal metastases (six), adrenal hemorrhage (two), and neuro-
blastoma (seven). MRI detected the abnormal adrenal gland(s) in 20 of 21 patients. MRI
was unable to detect calcifications in the lesions studied but more clearly showed the
relations of adrenal masses to the major vascular structures. MRI demonstrated corti-
comedullary differentiation in patients with adrenal hyperplasia and in some normal
volunteers. The CT and MRI features of the adrenal lesions are discussed.

Computed tomography (CT) is currently the method of choice for evaluating the
adrenal glands because of its noninvasive nature and its ability to clearly demon-
strate normal as well as abnormal adrenal glands. The ability to delineate normal
adrenal glands is important, because an adrenal tumor cannot be excluded unless
a normal adrenal gland is demonstrated on several contiguous sections. With
current techniques, both adrenal glands can be seen on routine abdominal CT
scans in at least 95% of patients [1 Failure to adequately
]. delineate an adrenal
gland with CT is usually related to a lack of retroperitoneal fat, which is necessary
to distinguish the adrenal from adjacent structures of similar density [2, 3]. Since
magnetic resonance imaging (MRI) provides superior contrast resolution, we
thought that it might be of use in the evaluation of adrenal glands. We compared
the ability of CT and MAI to demonstrate normal and abnormal adrenal glands and
determined the MRI features of various adrenal abnormalities.

Subjects and Methods

Received June 4, i984; accepted after revision
Imaging Systems
August 16, 1984.
Presented at the annual meeting of the American CT scanning was done with a Technicare 2060 scanner operating at 120 kVp and 200
Roentgen Ray Society, Las Vegas. April 1984. mAs. Each slice required 2 sec to perform. Most scans were performed with oral contrast
I Department of Radiology. Case Western Re- material and a drip infusion of intravenous contrast material (up to 300 ml of 1 4% organically
serve School of Medicine, University Hos-
University bound iodine).
pltals of Cleveland, 2074 Abington Ad. , Cleveland,
MRI was performed with a Teslacon imager (Technicare), which uses a superconducting
OH 44106. Address reprint requests to C. L.
Schultz. magnet operating at 0.30 T and two-dimensional Fourier transformation. Spin-echo (SE) and
2Fraecis Payne Bolton School of Nursing, Case inversion-recovery (IA) pulse sequences were used [4]. SE pulse sequences are defined
Westem Reserve University. Cleveland, OH 44106. according to echo time (TE) and repetition time (TA) in milliseconds. The following SE pulse
sequences were used (TA/TE): 250/30, 500/30, 1000/30, 1000/60, 1000/i 20, and 1500/
AJR 143:1235-1240, December 1984
0361 -803X/84/1 436-1235 120. IA sequences are defined in addition by inversion time (TI). The following IA pulse
© American Roentgen Ray Society sequences were used (TRJTI/TE): 1000/300/30 and 1500/450/30. Each SE or IA image
 1236 SCHULTZ ET AL. AJR:143, December 1984

required 2.2-6.7 mm to obtain, depending on the TA. Four averages Eight of the patients who underwent CT and i 4 of the
were obtained with the pulse sequences using 250 or 500 msec TA volunteers who underwent MRI were graded as having little
to increase the signal-to-noise ratio. Two averages were obtained retroperitoneal fat. All of the adrenal glands scored as fair
with the pulse sequences using 1000 or 1500 msec TA to decrease
depiction by CT were found in these patients with little retro-
the imaging time.
peritoneal fat. Seven of the eight left adrenal glands graded
as fair on MRI were found in volunteers with little retroperi-
Downloaded from ajronline.org by 103.169.227.219 on 02/06/24 from IP address 103.169.227.219. Copyright ARRS. For personal use only; all rights reserved

NormalAdrenal Glands toneal fat. Only two right adrenal glands were graded as fair
depiction on MRI, and one occurred in a volunteer with little
Thirty normal volunteers had MRI of the upper abdomen after
retroperitoneal fat.
informed consent was obtained, and the results were compared with
The adrenal glands were generally of similar density as the
a retrospective review of 30 abdominal CT scans that were inter-
preted as normal. The age range of the volunteers was 21 -71 years liver and vascular structures on the enhanced CT scans. In
(mean, 33). The age range of patients who had normal CT examina- some patients, however, there was a slight decrease in den-
tions was 17-80 years (mean, 51). Both CT and MRI were done with sity of the adrenals relative to the liver. This was helpful in
a 10-mm slice thickness and a 3-mm gap between slices. All volun- distinguishing the lateral margin of the right adrenal gland
teers had MRI with SE sequences using 250 or 500 msec TA and from the liver in some patients.
30 msec TE. Ten volunteers also had images with multiple SE and SE images with 250 or 500 msec TA and 30 msec TE
IA sequences at the level of the adrenal glands. provided the most consistent depiction of the normal adrenal
The normal CT and MRI scans were reviewed by two observers.
glands on MRI. The TA of 500 msec was preferred because
The shape of each normal adrenal gland as it was seen on each
it seemed to provide the best overall blend of contrast and
section was recorded. At the level where the adrenal was best seen,
the gland was graded according to the following criteria: (1) not seen; spatial resolution. The intensity of the normal adrenal glands
(2) fair depiction adrenal gland identified,
= but not all margins of the was greater than that of the vascular structures but less than
gland clearly distinguished from adjacent structures; or (3) good that of the liver and renal cortex on SE 250/30 or SE 500/30
depiction =all margins of the adrenal gland clearly distinguished. images. This natural contrast between the adrenal glands and
Each person was also graded according to the amount of retroperi- adjacent organs or vascular structures often enabled the
toneal fat present: (1) little retroperitonealfat or (2) moderate to large adrenals to be clearly distinguished in volunteers with a
amount of retroperitoneal fat. paucity of retroperitoneal fat. This was especially true on the
right, where 1 3 of 1 4 volunteers with little retropentoneal fat
AbnormalAdrenal Glands
had good depiction of the right adrenal gland (fig. 1).
In seven volunteers, there was a suggestion of two regions
Twenty-one patients with abnormal adrenal gland(s) on CT were of different intensity within the adrenal glands on SE images
also studied with MRI. The abnormalities studied were bilateral hy- (250 or 500 msec TA, 30 msec TE). There appeared to be an
perplasia (three patients), adenoma (two), myelolipoma (one), meta-
inner region of low intensity surrounded by a region of higher
static disease (six), adrenal hemorrhage (two), and neuroblastoma
intensity. This was believed to represent corticomedullary
(seven). The scanning parameters for both CT and MRI were generally
the same as those described above. In some patients who were differentiation, with the adrenal medulla appearing less intense
clinically suspected to have adrenal pathology, the CT scans were than the cortex.
obtained with contiguous 5-mm sections. Four patients had their CT Although MRI provided better contrast resolution between
done at other institutions. MRI was done with as many of the pulse adrenal glands and adjacent organs, the CT images showed
sequences listed above as possible. The pulse sequences used were better spatial resolution with sharper definition of the adrenal
limited by the constraints of time, with each MRI examination usually glands. Although the size of the adrenal glands was not
lasting about 1 hr or less. The abnormal CT and MRI scans were measured, it was our impression that adrenal glands some-
reviewed by two observers. The CT and MRI features of each times appeared wider on MRI due to this decreased image
abnormality were recorded, and the scans were compared regarding
sharpness.
their diagnostic features and image quality.

Abnormal Adrenal Glands

Results
MAI demonstrated the abnormal adrenal gland(s) in 20 of
Norma/Adrenal Glands
the 21 patients with abnormal adrenal gland(s) initially de-
CT identified both adrenal glands in all 30 patients (100%). tected with CT. MRI failed to demonstrate an abnormality in
MRI identified both adrenal glands in 29 (97%) of 30 volun- a child with an adrenal hemorrhage, and in another patient
teers. MRI failed to identify the right adrenal gland in one with an adrenal hemorrhage, the CT scan was considered to
volunteer whose scan was degraded by artifact from respi- be of greater diagnostic value. MRI was considered to be of
ratory motion. Good depiction, as defined by our criteria, was greater value than CT in the case of a nonfunctioning ada-
achieved of the right adrenal gland in 27 patients (90%) on noma. In 1 8 cases, the two methods were considered to be
both CT and MRI. Depiction of the left adrenal gland was of equal diagnostic value, but in four of these, the CT scans
graded as good in 26 patients (87%) with CT and in 22 were preferred over MA images by the observers.
volunteers (73%) with MRI. There were no statistically signif- The CT and MR scans showed enlarged adrenal glands in
icant differences between the imaging techniques using chi- the three patients with adrenal hyperplasia. Two of these
square analysis (x2 [1 ,n 60] 1 .91 p
= 0.05).= , patients had clinical and laboratory evidence of Cushing dis-
 AJR:143, December 1984 MRI/CT OF ADRENAL GLANDS 1237

pain. CT showed a low-density, sharply defined adrenal mass.

No enhancement was noted after a bolus of contrast material,
and the mass was believed to represent an adrenal cyst (fig.
3). MRI showed a low-intensity mass on SE 500/30 images.
The intensity of the mass was increased on the SE 1 500/i 20
image, but the lesion did not show the marked reversal of
Downloaded from ajronline.org by 103.169.227.219 on 02/06/24 from IP address 103.169.227.219. Copyright ARRS. For personal use only; all rights reserved

intensity that the gallbladder and spinal fluid demonstrated.

This indicated that the lesion was not cystic, and the MRI
appearance was believed to be consistent with a solid tumor,
most likely a nonfunctioning adenoma. Biopsy of the lesion
confirmed this diagnosis. There was no laboratory or clinical
evidence of adrenal abnormality in this patient. The function-
ing adenoma demonstrated an appearance identical to that
of the nonfunctioning adenoma. The mass appeared lower in
intensity than the liver on SE 500/30 images and was slightly
more intense than the liver with long TR and TE.
The adrenal myelolipoma demonstrated a characteristic
Fig. 1 -Normal adrenal glands, SE 250/30. A, Right adrenal gland (arrow) appearance on both CT and MRI (fig. 4). The CT scan showed
clearly distinguishedfrom liver by its low intensity, despite lack of retroperitoneal a low-density right adrenal mass (-85 H) that was somewhat
fat. B, Slightly lower section. Left adrenal gland (arrow) well seen. There seems mnhomogeneous. MRI a high-intensity
demonstrated right ad-
to be corticomedullary differentiation, with medulla appearing less intense than
cortex. renal mass with some scattered areas of decreased intensity.
The intensity of the mass was about equal to that of the
subcutaneous and retroperitoneal fat on all the pulse Se-
quences used.
Six patients with metastatic disease and abnormal adrenal
gland(s) were studied with CT and MRI. One patient with oat
cell carcinoma and another with poorly differentiated adeno-
carcinoma of the lung had large bilateral adrenal masses
demonstrated with CT and MRI. The intensity of the adrenal
metastases was slightly less than that of liver in one patient;
in the other, the intensity was equal to that of liver on SE
500/30 images. IA images demonstrated better contrast res-
olution between the liver and the adrenal metastases in both
patients. The CT images of the patient with poorly differen-
tiated adenocarcinoma showed the left adrenal to be less
dense than the right (fig. 5). No difference was noted on MRI
between the two glands with a pulse sequence of SE 500/
30; however, on SE 1 000/i 20, the left adrenal appeared
much more intense than the right. This probably reflected
greater fluid content within the left adrenal due to hemorrhage
or necrosis. All SE sequences demonstrated the relation of
Fig. 2.-Adrenal hyperplasia. A, SE 500/30. Enlarged adrenal glands (ar- the right adrenal mass to the inferior vena cava better than
rows) with corticomedullary differentiation. B, SE 250/30, surgically removed CT. Four other patients with metastatic disease showed a
hyperplastic adrenal gland. Cortex (arrowheads) is more intense than medulla
unilateral adrenal mass on CT and MRI. The diagnosis of
(arrows). Some high-intensity signal centrally and laterally was due to fat (f)
attached to specimen. metastatic disease was confirmed in one of these patients
with surgical removal of the gland; the other three had proven
metastatic disease elsewhere, so no confirmation of adrenal
metastases was warranted. These adrenal masses appeared
ease; the third had normal adrenal function. MRI showed similar to the proven cases of metastatic disease, with an
corticomedullary differentiation, with the medulla appearing intensity either equal to or somewhat less than that of the
less intense than the cortex (fig. 2A). This was best seen on liver on SE 500/30 images. The MRI appearance of these
SE 500/30 images. Corticomedullary differentiation was con- lesions was not, however, considered specific for metastatic
firmed in one patient by imaging the hyperplastic adrenal disease.
glands after surgical removal (fig. 2B). Two patients with presumed adrenal hemorrhage involving
Two adrenal adenomas were studied. One was a nonfunc- the right adrenal gland were studied with both methods. Both
tioning adenoma; the other was functioning and produced of these hemorrhages had dense calcifications within the right
Cushing syndrome in a 1 2-year-old girl. The nonfunctioning adrenal gland. One patient was an adult who had had a
adenoma was discovered on CT performed for abdominal calcified right adrenal mass for 1 5 years and negative skin
 1238 SCHULTZ ET AL. AJA:143, December 1984
Downloaded from ajronline.org by 103.169.227.219 on 02/06/24 from IP address 103.169.227.219. Copyright ARRS. For personal use only; all rights reserved

Fig. 3.-Nonfunctioning adenoma. A, CT scan. Low-density, right adrenal spinal fluid (arrow). C, SE 1500/1 20. Right adrenal mass (M) is higher in
mass (arrow), thought to be adrenal cyst. B, SE 500/30. Right adrenal mass intensity than liver but does not show marked reversal of intensity demonstrated
(M) is less intense than liver and similar in intensity to gallbladder (GB) and by gallbladder (GB) and splnal fluid (arrow).

Fig. 4.-Adrenal myelolipoma. A, CT scan. Right

adrenal mass (M) about same density as subcuta-
neous and retroperitoneal fat. K = kidney. (Cour-
tesy of J. Holyland, Cleveland, OH.) B, SE 500/30.
High-intensity mass (M) due to fat content of tumor.

tests for tuberculosis. CT demonstrated an enlarged right six children with stage IV neuroblastoma originating from the
adrenal gland with extensive calcification. MAI demonstrated adrenal gland were also studied. In these cases, MRI pro-
an enlarged adrenal gland that was less intense than liver and duced anatomic detail comparable to that seen on CT with
renal cortex on SE 500/30 images. The extensive calcification the added benefit of displaying the relation of the tumor to
was not seen on MAI, so the mass could not be distinguished the inferior vena cava and the aorta and its major branches
from an adrenal adenoma or a metastatic lesion by its MRI without contrast material. Calcification of all the tumors was
appearance. Therefore, MRI was judged to be of less diag- evident with CT, but not with MRI. Contrast resolution be-
nostic value than CT in this case. In a child with an adrenal tween tumor and normal liver tissue obtained with MRI ex-
hemorrhage, MAI failed to detect the abnormality. CT dem- ceeded that of CT and was most marked on IA images, which
onstrated a calcified adrenal gland, which was not enlarged. intensified the relatively bright hepatic tissue in comparison
Seven cases of neuroblastoma were studied with both with the dark (low intensity) neoplasm. No differences in
methods. A 26-year-old patient with a “small cell tumor con- intensity were perceived betwpen kidney and tumor. There-
sistent with neuroblastoma” demonstrated a right adrenal fore, it was necessary to rely on the relatively bright perirenal
mass with scattered calcifications on CT. The calcifications fat to determine whether or not there was direct extension to
were not detected with MRI. Involvement of the right kidney the kidney, as occurred in one of these children. In another
and a lumbar vertebra was apparent with both CT and MRI patient, hydronephrosis due to extension of the tumor mass
(fig. 6). As part of another series to be reported separately, to the renal pelvis was clearly demonstrated.
 AJA:143, December 1984 MRI/CT OF ADRENAL GLANDS 1239

Fig. 5.-Bilateral adrenal metastases in patient

with poorly differentiated adenocarcinoma of lung.
A, CT scan. Bilateral adrenal masses with left
adrenal mass less dense than right. B, SE 500/30.
Bilateral adrenal masses of equal intensity. Poor
contrast resolution between right adrenal mass
and liver. Inferior vena cava (C) is more clearly
defined than on CT scan. C, IA 1000/300/30.
Downloaded from ajronline.org by 103.169.227.219 on 02/06/24 from IP address 103.169.227.219. Copyright ARRS. For personal use only; all rights reserved

Better contrast resolution between liver and right

adrenal mass. 0, SE 1 000/1 20. Left adrenal gland
(L) much more intense than right (A), probably due
to hemorrhage or necrosis within mass.

---‘ I
&tL.,?.-

Fig. 6.-Neuroblastoma with renal invasion. A,

CT scan. Right adrenal mass with calcifications and
obliteration of perirenal fat. (Courtesy of B. J.
Archer, Cleveland, OH.) B, SE 250/30. Right adre-
nal mass with obliteration of perirenal fat; no calci-
fications apparent within mass. vertebral body, do-
creased in intensity because of tumor involvement,
was also abnormal on CT when viewed with appro-
pnate window and center settings.

A B

Discussion MAI study population. Good depiction was sometimes pre-

vented with MRI on the left by difficulty in clearly distinguishing
This study suggests that MRI can detect normal and ab- the lateral margin of the gland from adjacent pancreas in
normal adrenal glands at a rate comparable to that of CT. patients with a paucity of retroperitoneal fat. Retroperitoneal
Both adrenal glands were identified in 29 of 30 volunteers, fat aids in distinguishing the normal adrenal gland from adja-
and depiction of the adrenals was rated as good in 90% of cent structures, but the right gland was easily distinguished
the study population on the right and in 73% of the cases on from the liver by its different intensity on SE 500/30 or SE
the left. Although CT achieved a higher rate of good depiction 250/30 images. These findings are slightly different from
of the left adrenal gland than MRI, this finding was not those of Moon et al. [5], who reported that MRI clearly
statistically significant and may have been due in part to the showed 1 00% of the left adrenal glands and 86% of the right
greater number of people with little retroperitoneal fat in the adrenal glands in their study.
 1240 SCHULTZ ET AL. AJA:143, December 1984

The scanning factors used in this study (1 0-mm slices with advantages of MRI of the adrenal glands. Corticomedullary
1 3-mm slice intervals) are certainly not optimal for adrenal differentiation was demonstrated with MAI in several volun-
studies, but they represent the factors that we currently use teers and in patients with adrenal hyperplasia. However, the
for most abdominal CT scans. Adrenal pathology is often clinical utility of this finding is not readily apparent. MRI was
unsuspected clinically, so we chose to evaluate the two better able to demonstrate the relation of an adrenal mass to
methods with these “routine” factors. Certainly thin, overlap- the major vascular structures, which was of significance in
Downloaded from ajronline.org by 103.169.227.219 on 02/06/24 from IP address 103.169.227.219. Copyright ARRS. For personal use only; all rights reserved

ping sections should be used when searching for a small some patients. Disadvantages of MRI were its decreased
adrenal lesion with either CT or MRI [6]. spatial resolution and its inability to detect calcifications. We
The criteria used for the detection of an adrenal mass were consider the failure of MRI to detect calcifications a serious
the same for CT and MRI. An adrenal gland was considered drawback in adrenal imaging, since the presence and appear-
to contain a mass if the normally straight or concave margins ance of calcifications are an important feature of several
were convex, making the gland ovoid or round. No attempt adrenal lesions.
was made in this study to examine patients with MRI who In summary, MRI identified normal and abnormal adrenal
were suspected of having adrenal tumors but had normal CT glands at a rate comparable to that of CT in this study. We
scans. It is not known if MRI can detect a lesion that is not believe that either CT or MRI can be used to detect an adrenal
contour-deforming by a difference in intensity between the mass that enlarges and distorts the contour of the adrenal
lesion and the normal glandular tissue. gland. Neither method is specific for most adrenal lesions,
We prefer a spin-echo (SE) sequence with 500 msec TA and performing both studies did not seem to increase the
and 30 msec TE for delineation of normal and abnormal specificity in our series.
adrenal glands. However, this sequence may not be optimal
for delineation of small lesions that are not contour-deforming.
One can hypothesize that sequences with long TA and TE REFERENCES
(e.g., SE 1 500/1 20) could detect a lesi9n within a normal-size 1 Wilms G, Baert A, Marchal
. G, Goddeeris P. Computed tomog-
adrenal gland by a difference in intensity. These sequences, raphy of the normal adrenal glands: correlative study with au-
which are T2-weighted [7], are currently limited by decreased topsy specimens. J Comput Assist Tomogr 1979;3:467-469
spatial resolution, so the MRI detection of small adrenal 2. Karstaedt N, Sagel SS, Stanley RJ, Melson GL, Levitt AG.
lesions (such as aldosterone-producing adenomas) will re- Computed tomography of the adrenal gland. Radiology
1978;i 29:723-730
quire further technical improvements. Despite the decreased
3. Montagne JP, Kressel HY, Korobkin M, Moss AA. Computed
spatial resolution, these sequences were helpful in character-
tomography of the normal adrenal glands. AJR 1978;i30:963-
izing a large adrenal mass in several patients. Inversion-
966
recovery (IA) techniques were also useful in some patients, 4. American College of Radiology. Glossary ofNMR terms. Chicago:
since IA provided better contrast resolution between adrenal American College of Radiology, 1983
lesions and liver. 5. Moon KL, Hricak H, Crooks LE, et al. Nuclear magnetic reso-
The MRI appearance of the adrenal lesions studied did not nance imaging of the adrenal gland: a preliminary report. Radiol-
seem to be specific except for the myelolipoma, which had a ogy 1983;i47:i55-i60
characteristic CT and MRI appearance [8]. The myelolipoma 6. Sheedy PF, Hattery AR, Stephens DH, Van Heerden JA, Sheps
demonstrated an intensity similar to that of retroperitoneal fat 5G. The adrenal glands. In: Haaga JR, Alfidi RJ, eds. Computed
tomography of the whole body. St. Louis: Mosby, 1983:681-
on all pulse sequences used. Adrenal adenomas, neuroblas-
705
tomas, adrenal hemorrhage, and some cases of metastases
7. Han JS, Kaufman B, Alfidi RJ, et at. Head trauma evaluated by
all demonstrated a similar MRI appearance, with an intensity magnetic resonance and computed tomography: a comparison.
somewhat less than that of liver on short TA and TE se- Radiology 1984:150:71-77
quences (e.g., SE 500/30) and an intensity becoming greater 8. Weiner SN, Bernstein AG, Lowy 5, Karp H. Combined adrenal
than that of liver as the TA and TE were increased. adenoma and myelolipoma. J Comput Assist Tomogr
In addition to the lack of ionizing radiation, there are other 1981;5:440-442