Downloaded from www.ajronline.org by 223.255.227.25 on 01/03/17 from IP address 223.255.227.25. Copyright ARRS.

For personal use only; all rights reserved

Robert G. Levitt1
Harvey S.Glazer1
Charles L. Roper2
Joseph K. T. Lee1
William A. Murphy1
,.

cases. Several pitfalls in MRI evaluation of the mediastinum were identified. By MRI the
esophagus may be misinterpreted as an enlarged retrotracheal lymph node unless serial
scans are studied. Scattered calcifications in enlarged mediastinal and hilar lymph
nodes due to old granulomatous disease are not detectable by MRI. Small adjacent
lymph nodes shown individually by CT may appear as a single enlarged lymph node by
MRI due to partial-volume averaging. Small lung nodules may be undetected by MRI due
to respiratory motion and partial-volume averaging. Certain oatients are unsuitable for
MR scanning. Because of the requirement for patient selection and the identified pitfalls
of MRI, CT remains the radiologic procedure of choice in the staging of patients with
bronchogenic carcinoma and the evaluation of other mediastinal and hilar masses at
present However, because of the ability to show blood vessels without an intravascular
contrast agent, MRI is useful in evaluating patients with potential contrast allergy and
solving diagnostic problems not solved by CT.

Computed tomography (CT) has become the radiologic procedure of choice for
staging bronchogenic carcinoma and determining the solid or vascular nature and
extent of other mediastinal and/or hilar masses identified by routine chest radiog-
raphy [i 2]. The lack of ionizing radiation,
, the ability to do multiplanar imaging,
and the lack of need for intravenous contrast media in assessing thoracic vascular
structures makes magnetic resonance imaging (MAI) an attractive alternative to
CT scanning in the thorax. Recent reports have documented the ability of MAI to
demonstrate mediastinal invasion by tumor, and hilar and mediastinal adenopathy
[3-6]. Aegarding bronchogenic carcinoma, some authors have claimed MRI to be
as useful as CT or more so [4], while others have concluded MRI offers little
improvement in diagnosis over contrast-enhanced CT [5, 6]. These conflicting
Received October 1, 1984; accepted after revi- conclusions were based on studies of small numbers of patients with bronchogenic
sion March 12, 1985.
carcinoma.
Presented at the annual meeting of the American
Roentgen Ray Society, Las Vegas, April 1984. We undertook a comparison of MRI and CT in the staging of 37 patients with
‘Mallinckrodt Institute of Radiology. Washington
bronchogenic carcinoma and the evaluation of ii additional patients with other
Lkiversity Schoolof Medicine, 510S. Kingshighway mediastinal and/or hilar masses.
Blvd., St. Louis, MO 63110. Address reprint re-
quests to A. G. Levitt.
2t)ivis of Cardiothoracic Surgery, Washington Subjects and Methods
University School of Medicine, St. Louis, MO 63110.
Our patient population consisted of 31 men (average age, 58.2 years) and 17 women
AJR 145:9-14, July 1985
0361-803x/85/1451-0009 (average age, 56.5 years). Of these 48 patients, 37 had bronchogenic carcinoma proven by
© American Roentgen Ray Society sputum cytology, percutaneous needle biopsy, bronchoscopic washings, and biopsy or
 io LEVITT ET AL. AJA:i45, July 1985

Fig. 1 .-Mediastinal invasion by bronchogenic

-
carcinoma. A, CT scan at level of right hilum. Right-
upper-lobe tumor (t) invades mediastinum anterior
and posterior to right main-stem bronchus (arrow),
which is compressed by tumor. Superior vena cava
is encased by tumor (arrowhead). B, SE 900/30
image. Tumor of intermediate signal intensity (t)
invades mediastinum. Superior vena cava (arrow-
Downloaded from www.ajronline.org by 223.255.227.25 on 01/03/17 from IP address 223.255.227.25. Copyright ARRS. For personal use only; all rights reserved

head) and right main-stem bronchus (arrow) are

seen. aa = ascending aorta; da = descending aorta;
pa = pulmonary artery.

thoracotomy. Histology of the primary neoplasms was epidermoid Patients

were examined by MAI after informed consent under
carcinoma in 1 7 patients, adenocarcinoma in five, oat cell carcinoma guidelines by research
established protocol and approved by the
in seven, poorly differentiated carcinoma in three, and unclassified Human Studies Committee of the Washington University School of
carcinoma in three. Medicine.
An additional 1 1 patients did not have bronchogenic carcinoma, Prospective CT interpretations rendered at the completion of the
but had other superior mediastinal and/or hilar masses: Hodgkin CT examination were used in assessing CT results. These CT inter-
disease in three, fibrosing mediastinitis in two, hilar adenopathy in pretations were made with knowledge of prior radiologic and diag-
two, and lymphoma, mediastinal lipomatosis, azygos arch variant, nostic tests (e.g., bronchoscopy) but before surgical and pathologic
and normal hilum in one each. MRI and CT were performed within 72 evaluation of the extent of the neoplasm. MR studies were interpreted
hr of each other in all patients; CT preceded MRI in 44 of the 48 by two radiologists (A. G. L., H. S. G.) with knowledge of prior
patients. diagnostic tests, including CT findings but before surgical and path-
CT examinations were performed on a Siemens Somatom DR3 ologic evaluation of the extent of neoplasm.
with a scanning time of 5.2 sec or on an EMI 7070 with a scanning Criteria for unresectability of bronchogenic carcinoma for cure by
time of 3 sec. For staging bronchogenic carcinoma by CT, serial MRI and CT included direct mediastinal invasion, mediastinal lymph
contiguous scans were obtained during suspended inspiration at 1- node enlargement, extensive pleural or chest wall invasion, and
cm intervals throughout the thorax and upper abdomen using 8-i 0 metastatic tumor nodules in lung. Direct medmastinal invasion was
mm collimation. Intravenous iodinated contrast material was admin- diagnosed when tumor invaded (interdigitated with) medmastinal fat
istered by bolus technique or bolus drip-infusion technique when and/or surrounded the major mediastinal vessels or the main-stem
noncontrast scans were indeterminate concerning vascular or non- bronchi (fig. 1). Mediastinal lymph nodes that were more than 2 cm
vascular nature of soft-tissue densities within the mediastinum and in diameter were interpreted as involved with malignancy [11 (fig. 2).
hila. In patients without bronchogenic carcinoma, MRI and CT were
MAI was performed with a Siemens Magnetom using a supercon- compared in their ability to determine the extent of the mass and
ductive magnet operating at 0.35 T. Data acquisition was by the spin- tissue characteristics of the mass (i.e., signal intensity or attenuation
echo (SE) technique using one or more pulse sequences. Electrocar- value).
diographic (ECG) triggering for data acquisition became available
after the study had begun and was used in 10 cases. Display matrix
size was 256 x 256 elements; each pixel was 1 .95 x 1.95 mm. Ti Results
and T2 calculations became available near completion of the study
and were not performed. Unlike CT, the MR examination was not a MA! Technica! Considerations
complete survey of the thorax and upper abdomen. MA images were
Subjectively, mediastinal and hilar masses were most easily
obtained at 2-cm intervals using 1 -cm slice thickness through the
region of the mediastinal and/or hilar mass(es) as previously demon- identified using a pulse sequence with a short TA (300 or 500
strated by CT scans or chest radiographs. msec) and short TE (30 msec). Such Ti -weighted pulse
A single SE pulse sequence was used in 25 of our 48 patients. In sequences provided images with a significant difference in
these cases, the pulse repetition rate (TA) varied from 300 to 1800 signal intensity between normal mediastinal or hilar fat (high
msec; the spin-echo delay (TE) was usually 30 msec, but 60 and 90 signal intensity) and abnormal soft-tissue masses (lower sig-
msec TE images were obtained in some cases. Seven of the single- nal intensity). As TA or TE was lengthened, the difference in
pulse-sequence patients had data acquired during ECG gating of the signal intensity between normal fat and abnormal mass de-
pulse sequence. When the heart rate is 80 beats/mm, the effective creased, resulting in decreased detectability of the lesion
TA is 750 msec on an ECG-gated sequence; TE remained at 30
(fig. 3).
msec. Multiple SE pulse sequences were performed in 23 patients.
MR image quality was evaluated subjectively. Images in
Two gated pulse sequences were performed in 15 patients and three
gated pulse sequences in three patients. Five patients had both gated which hilar structures could be resolved and no respiratory
and ungated pulse sequences. Thirty-eight patients had MAI per- motion artifacts were present were considered high-quality
formed in the axial plane only; 10 patients had scans performed in images. Image quality was judged high quality in 42(87%) of
both axial and coronal planes. the 48 cases. ECG gating was not a requisite for high-quality
 AJR:i45, July 1985 CT/MAI OF MEDIASTINAL/HILAR MASSES ii

Fig. 2.-Mediastinal lymph node metastasis

from bronchogenic carcinoma. A, CT scan at top of
aortic arch. Right-upper-lobe mass and enlarged
pretracheal lymph node (arrows). B, SE 500/30
image. Right-upper-lobe mass and enlarged pretra-
cheal lymph node (arrows). Intermediate intensity
signal posterior to trachea (t) is esophagus (arrow-
head). s = superior vena cava; ar = aortic arch.
Downloaded from www.ajronline.org by 223.255.227.25 on 01/03/17 from IP address 223.255.227.25. Copyright ARRS. For personal use only; all rights reserved

I .s.

Fig. 3.-Effect of pulse repetition rate (TR) on

signal intensity of bronchogenic carcinoma. A, SE
500/30 image. Left hilar tumor (t) invades medias-
tinum, anterior and posterior to left atrium (Ia). Mass
produces left-upper-lobe collapse (arrowheads).
Short TA and TE pulse sequence results in good
contrast between high-intensity mediastinal fat (ar-
rows) and lower-intensity tumor. No signal intensity
difference between tumor and left-upper-lobe col-
lapse. B, SE 900/30 image at identical window
settings. As TA is increased from 500 to 900 msec,
signal intensity of tumor (t) increases relative to
signal intensity of fat, and contrast between tumor
and mediastinal fat (arrows) is reduced. There is
still no signal intensity difference between tumor
and left-upper-lobe collapse (arrowheads). aa =
ascending aorta; rv = right ventricle; da = descend-
ing aorta.

images; 32 (84%) of 38 gated examinations were of such including four cases of oat cell carcinoma), direct mediastinal
quality. ECG gating improved spatial resolution of hilar invasion in association with mediastinal lymph node enlarge-
masses in cases where gated and ungated data acquisitions ment (four cases), and multiple pleural metastases (one case).
were obtained, but was not required for identification of hilar Operative proof of unresectability for cure was obtained by
masses. Evaluation of superior mediastinal masses was not mediastinoscopy in seven cases, scalene lymph node biopsy
significantly improved by ECG gating. in two cases, and exploratory thoracotomy in two cases.
Of 48 cases, 38 (79%) were studied in the axial plane; only Both MAI and CT were interpreted as unresectable for cure
1 0 (21 %) of 48 cases were studied in both axial and coronal in 31 of 35 cases clinically or surgically determined to be
planes. In the small percentage of cases studied in both unresectable. Two cases were correctly interpreted as being
planes, coronal images did not provide information needed unresectable for cure by MRI when CT interpretations mdi-
for staging of bronchogenic carcinoma or evaluating other cated the lesions were resectable for cure. Both cases had
mediastinal or hilar masses that was not present on axial aortopulmonary window adenopathy on MR scans. These CT
scans. examinations had suboptimal drip-infusion contrast scans
through the aortopulmonary region; the adenopathy by CT
was appreciated retrospectively. In another case, MRI differ-
MA! vs. CT: Staging of Bronchogenic Carcinoma
entiated recurrent carcinoma from contiguous radiation fibro-
Two of the 37 cases of bronchogenic carcinoma were sis by differences in tissue signal intensity on T2-weighted
staged as resectable by MAI and CT; both cases were images; CT did not differentiate the two tissue types by
resectable for cure at operation. Thirty-five of the 37 cases attenuation values [7].
were staged unresectable for cure by MRI and/or CT for the MRI failed to identify tumor in one case; CT correctly staged
following reasons: direct mediastinal invasion by tumor (nine this tumor as unresectable for cure due to mediastinal inva-
cases, including three cases of oat cell carcinoma), medias- sion. This tumor was located in the anterior segment of the
tinal lymph node enlargement greater than 2 cm (21 cases, right upper lobe adjacent to the mediastinal pleura and invad-
 12 LEVITT ET AL. AJR:145, July 1985

Fig. 4.-Differentiation of proximal tumor from

distal consolidation/collapse by MRI. A, Postcon-
trast CT scan at level of aortopulmonary window.
Right-upper-lobe mass (m) invades mediastinum;
right-upper-lobe consolidation/collapse (arrow-
heads). Mass is not well differentiated from right-
upper-lobe disease by attenuation values. Calcified
pretracheal lymph node is present (arrow). B, SE
Downloaded from www.ajronline.org by 223.255.227.25 on 01/03/17 from IP address 223.255.227.25. Copyright ARRS. For personal use only; all rights reserved

1500/30 image. Invasive low-intensity mass (m) is

differentiated from high-intensity right-upper-lobe
consolidation/collapse (arrowheads). Lymph node
calcification (arrow) is seen as absence of signal
similar to superior vena cava (5) and tracheal bifur-
cation(t). Signal intensity of vertebral body is similar
to that of mass and could be mistaken for invasion
by tumor. aa = ascending aorta; da = descending
aorta.

Fig. 5.-Esophagus appeanng as enlarged

lymph node on MRI. A, CT scan at level of aortic
arch (ar). Two normal-sized pretracheal lymph
nodes (arrowheads) and esophagus (arrow). B, SE
500/30 image. Esophagus (arrow) mimics enlarged
lymph node. Note that two normal pretracheal
lymph nodes on CT appear as one enlarged lymph
node on MRI (arrowhead). s = superior vena cava.

ing the mediastinum. MR data acquisition was performed when MR images showed no difference in signal intensity
using an SE i 800 (TR)/30 (TE) pulse sequence. With this between proximal tumor and distal collapse. In two cases,
pulse sequence, the tumor had high signal intensity that could multiple discrete enlarged mediastinal lymph nodes were iden-
not be differentiated from the high signal intensity of adjacent tified on CT when they appeared as a single mediastinal mass
mediastinal fat. by MRI.
MRI showed benefits over CT in i 3 cases. Hilar adenopathy
was detected by MRI in i 2 cases, eight of which required
MA! vs. CT: Other Mediastina! and Hilar Masses
contrast-enhanced scans for detection by CT. MAI differen-
tiated proximal tumor from distal collapse or consolidation in MR and CT findings were comparable in five of 1 1 cases
five other cases (one surgically proved) due to differences in of other mediastinal or hilar masses. Both techniques showed
signal intensities, when differentiation was not possible using normal hila in one case in which a chest radiograph was
CT (fig. 4). Postcontrast CT scans were obtained in only two suspicious for enlarged hila and confirmed mediastinal and
of these five cases. hilar adenopathy in one case of Hodgkin disease and unilateral
Conversely, CT demonstrated benefits over MRI in i 2 other hilar adenopathy in two cases, one of which was shown by
cases. CT clearly identified as esophagus a structure that surgery to be due to granulomatous infection. A superior
was thought to be an enlarged mediastinal lymph node by mediastinal mass of fat density by CT, clinically thought to be
MRI in three cases (fig. 5) and thought to represent medias- mediastinal lipomatosis, was seen equally well by both tech-
tinal invasion by tumor in one case. MRI staging was not in niques.
error in these cases because of enlarged mediastinal lymph Another five mediastinal and/or hilar masses were better
nodes seen at other levels. Calcification was detected within evaluated by CT than by MRI. Two of these cases were
enlarged mediastinal lymph nodes by CT and not by MRI in fibrosing mediastinitis; calcifications within the enlarged lymph
four cases (fig. 6). Use of bolus/infusion CT contrast technique nodes were identified on CT scans only, and segmental
differentiated proximal tumor from distal collapse in two cases bronchial anatomy and segmental pulmonary vessels were
 AJR:i45, July 1985 CT/MRI OF MEDIASTINAL/HILAR MASSES i3

Fig. 6.-Calcification within lymph node not de-

tected by MRI. A, CT scan at canna. One normal-
sized (arrowhead) and one enlarged (arrow) calci-
fled pretracheal lymph node. B, SE 500/30 image
falls to demonstrate calcification within enlarged
lymph node (arrow). Calcification within normal-
sized lymph node is area of no signal (arrowhead).
Downloaded from www.ajronline.org by 223.255.227.25 on 01/03/17 from IP address 223.255.227.25. Copyright ARRS. For personal use only; all rights reserved

S = superior vena cava; aa = ascending aorta; da

= descending aorta; Ipa = left pulmonary artery;
= tracheal bifurcation.

seen on CT but not MRI. These two cases are being reported studying the chest and mediastinum, SE imaging with both
in a separate report. Poor MR imaging quality due to patient short and long TA values is helpful [9].
motion made diagnostic interpretation impossible in one case In the small percentage of patients in our series who had
of Hodgkin disease. In another case, individual enlarged su- axial and coronal images, the coronal images did not reveal
perior mediastinal lymph nodes were seen on CT, but the additional findings over axial images. Yet coronal imaging may
enlarged nodes appeared as a single mass on MRI. One case be of value in several situations. It allows imaging along the
of lymphoma with a superior mediastinal mass on CT was longitudinal axes of structures that lie in or near the coronal
not identified on MAI. Data were collected using an SE 1800/ plane, such as the trachea and main bronchi. In addition,
30 pulse sequence. The enlarged mediastinal lymph nodes coronal MAI can clarify anatomic relations difficult to resolve
could not be differentiated from mediastinal fat because the on transaxial images because of volume averaging. Specifi-
high signal intensity of the mass of nodes approximated the cally, coronal images are sometimes superior to transaxial
signal intensity of mediastinal fat. Also, more and smaller lung images in evaluating the aortopulmonary window and masses
nodules were identified by CT than by MRI in this case. at the lung apex or base [1 0, 1 1].
One mediastinal mass was better evaluated by MAI than ECG gating of data acquisition yielded high-quality images
by CT. This retrotracheal mass was identified as a high in each case performed but was not required to obtain high-
azygos arch on MRI. Postcontrast CT scans showed only quality MA images. Patient movement, deep breathing, and
slight enhancement of the retrotracheal mass, and it was not particularly coughing degraded MR scan quality more signifi-
certain whether it was a solid or vascular mass by CT. cantly than cardiac motion. ECG gating was most useful in
evaluating hilar mases where transmitted cardiac motion has
Discussion its greatest effect. Hilar structures could be demonstrated
The question of “optimal” pulse sequence for staging bron- with or without ECG gating but were better defined on gated
chogenic carcinoma and evaluating other soft-tissue medias- images. Respiratory gating can remove arclike artifacts due
tinal or hilar masses is complicated. SE pulse sequences to chest wall motion and improve MA image quality [1 2], but
using short TA and short TE provide significant tissue contrast was not used in this study.
between lower-intensity tumor and higher-intensity medias- This comparison of MAI and CT in the evaluation of me-
tinal or hilar fat, and these sequences are the most helpful in diastinal and hilar masses has several biases. Patients with
identifying mediastinal and hilar masses. As TA is increased, bronchogenic carcinoma were not randomly selected but
the signal intensity of tumor relative to mediastinal fat in- were chosen on the basis of CT findings that showed the
creases, and there is less difference in signal intensity be- tumor to be unresectable for cure. Thus, there is a heavy bias
tween tumor and fat. In this series, two cases of bronchogenic in our series toward unresectable bronchogenic carcinoma.
carcinoma and one case of lymphoma were misinterpreted as Our population of other mediastinal masses and hilar masses
mediastinal fat when an SE 1800/30 pulse sequence was is also generally recruited from patients with CT scans, so
used; these errors could have been avoided by selecting a that CT preceded MAI in 44 of the 48 cases. Also, the CT
short TA and TE sequence. examination was a complete study of the thorax and upper
However, a Iong-TR pulse sequence is advantageous in abdomen, while MRI was limited to the region of the medias-
staging bronchogenic carcinoma and evaluating mediastinal tinal or hilar mass because of time constraints on the MR
and/or hilar masses. As TA is increased, the signal/noise ratio instrument. Despite these biases, some useful information
increases and image quality improves. The resolution of bron- regarding the capabilities and limitations of thoracic MRI as
chi and mediastinal vascular structures and definition of their compared to thoracic CT was gained.
walls are all improved. The increased signal strength of hilar In our selected and biased series of patients with broncho-
and lung lesions makes them easier to detect [8, 9]. In genic carcinoma, the accuracy of MRI and CT in staging
 i4 LEVITT ET AL. AJA:145, July 1985

nonresectability for cure was comparable. CT staged 35 of cedure of choice in staging patients with bronchogenic carci-
37 cases appropriately, while MRI staged 36 of 37 cases noma at this time. We believe MRI should be reserved for
correctly. Both CT errors were due to failure to appreciate cases in which intravenous contrast material is contraindi-
aortopulmonary window adenopathy prospectively; postcon- cated or postcontrast CT scans are inconclusive.
trast scans were suboptimal in both cases, and the ade- The limitations of MRI of the mediastinum in bronchogenic
nopathy was identified retrospectively. carcinoma likewise apply to the evaluation of other medias-
Downloaded from www.ajronline.org by 223.255.227.25 on 01/03/17 from IP address 223.255.227.25. Copyright ARRS. For personal use only; all rights reserved

We identified several pitfalls in MRI evaluation of the me- tinal and hilar masses. In particular, discrete clustered normal-
diastinum which were, in some cases, avoided by prior knowl- sized lymph nodes may appear as a single large lymph node
edge of CT findings. On MR scans, the esophagus may be mass on MRI, and calcified hilar lymph nodes may also be
mistaken for an enlarged retrotracheal lymph node on a single misinterpreted as adenopathy of unknown cause. Thus, CT
image. This pitfall may be avoided by study of serial images, remains the procedure of choice in other mediastinal and hilar
which will identify the soft-tissue density as the esophagus. masses as well.
The esophagus may also mimic mediastinal invasion when
tumor is located adjacent to the esophagus. Scattered calci- ACKNOWLEDGMENTS
fications within enlarged mediastinal and hilar lymph nodes
We thank Carol Keller and Lynn Losse for help in manuscript
due to old granulomatous disease are not detectable by MRI
preparation.
[5, i 3]. Small calcifications may have no signal or low signal
intensity and lead to an interpretation of inhomogeneous
REFERENCES
signal intensity lymphadenopathy. Small lymph nodes clus-
tered together may appear as a single enlarged lymph node 1 Baron RL, Levitt RG, Sagel SS, White MJ, Roper CL, Marbarger
.

on MRI due to long data acquisition times reducing spatial JP. Computed tomography in the preoperative evaluation of
resolution. Long acquisition times when multiple scans are bronchogenic carcinoma. Radiology 1982;i45:727-732
obtained simultaneously may prevent resolution of segmental 2. Faling U, Pugatch RD, Jung-Legg Y, et al. Computed tomogra-
phy scanning of the mediastinum in the staging of bronchogenic
bronchi and small lung lesions by MRI [3, 8, 14]. On SE
carcinoma. Am Rev Resp Dis 1981;i24:690-695
images in normal patients, collections of hilar soft tissue large
3. Gamsu G, Webb WA, Sheldon P, et aI. Nuclear magnetic reso-
enough to be confused with an abnormally enlarged lymph nance imaging of the thorax. Radiology 1983;147:473-480
node are seen [8]. 4. Cohen AM, Crewston S, LiPuma JP, et al. Nuclear magnetic
In eight cases, MRI identified hilar adenopathy in patients resonance imaging ofthe mediastinum and hili: early impressions
with bronchogenic carcinoma when intravenous contrast ma- of its efficacy. AJR 1983;i 41:1163-1169
terial was required to detect the adenopathy on CT scans. 5. Ross JS, O’Donovan FB, Novoa A, et al. Magnetic resonance of
Although the presence of hilar adenopathy does not in itself the chest: initial experience with imaging and in vivo T1 and T2
indicate unresectability for cure, it may obviate surgery for calculations. Radiology 1984;152:95-101
patients whose pulmonary status or general medical condition 6. Epstein DM, Kressel H, Gefter W, et al. MR imaging of the
mediastinum: a retrospective comparison with computed tomog-
would permit a lobectomy but not a pneumonectomy.
raphy. J Comput Assist Tomogr 1984;8:670-676
MRI differentiated tumor from distal pneumonia and/or
7. Glazer HS, Levitt AG, Lee JKT, Emani B, Gronemeyer 5, Murphy
collapse in five cases in this series (one case surgically
WA. Differentiation of radiation fibrosis from recurrent pulmonary
proven), but tumor was not differentiated from collapse/con- neoplasm by magnetic resonance imaging. AiR 1984;143:729-
solidation by MRI in seven other cases. The differentiation of 730
tumor from distal pneumonia or collapse by MRI is a variable 8. Webb WA, Gamsu G, Stark DD, Moore EH. Magnetic resonance
finding in other series as well [3, 1 5], as is MRI differentiation imaging of the normal and abnormal pulmonary hila. Radiology
of tumor from fibrosis [6, 7, 1 5]. Use of multiecho sequences 52:89-94
1984;1
with long TE, not used in all our patients with tumor and distal 9. Webb WA, Gamsu G, Stark DD, Moon KL, Moore EH. Evaluation
pneumonia, would have aided differentiation of these tissues. ofmagnetic resonance sequences in imaging mediastinal tumors.
AJR 1984;143:723-727
It is important to remember that two of seven cases in which
10. Webb WA, Gamsu G, Crooks LE. Multisection sagittal and
MRI did not differentiate tumor from collapse were differen-
coronal imaging of the mediastinum and hila. Radiology
tiated by CT by bolus injection of contrast material.
1984;1 50:475-478
Although a complete MR examination of the thorax and 1 1 Webb WA, Jensen
. BG, Gamsu G, Sollitto A, Moore EH. Coronal
upper abdomen using a single short-TA, short-TE pulse Se- magnetic resonance imaging of the chest: normal and abnormal.
quence may take 45-60 mm, use of a multislice technique Radiology 1984;1 53:729-735
with multiple echoes shortens the examination considerably. 12. Aunge VM, Clanton JA, Partain CL, James AE. Respiratory
Yet long data acquisition times means small lung nodules gating in magnetic resonance imaging at 0.5 tesla. Radiology
may be undetected by MRI due to respiratory motion and 1984;151 :521 -523

partial-volume averaging. Many patients are unsuitable for 13. Gamsu G, Stark DD, Webb WA, Moore EH, Sheldon PE. Mag-
MR scanning; seriously ill patients are excluded because they netic resonance imaging of benign mediastinal masses. Radiol-
ogy 1984;151 :709-713
cannot be monitored as closely as we desire and because
14. Cohen AM, Crewston S, LiPuma JP, Bryan PJ, Haaga JA, Alfidi
metallic resuscitation equipment may not enter the scanning
RJ. NMR evaluation of hilar and mediastinal lymphadenopathy.
suite. Patients with pacemakers or certain ferromagnetic vas-
Radiology 1983;148:739-742
cular surgical clips may not be scanned. Because of the 15. Axel L, Kressel HY, Thickman D, et al. NMR imaging of the chest
requirement for patient selection and the pitfalls of MRI within at 0.12 T: initial dinical experience with a resistive magnet. AJR
the mediastinum, we believe CT remains the radiologic pro- 1983;i41 :1157-1162