Thoracic Imaging 2003 - Society of Thoracic Radiology

CT Strucural and Functional Imaging of PE
U. Joseph Schoepf, M.D.
Department of Radiology, Brigham and Women’s Hospital, Harvard Medical School
CT Imaging of Pulmonary Embolism: General
Considerations
Although increasingly sophisticated clinical algorithms for
“bed-side” exclusion of Pulmonary Embolism (PE) are being
developed, mainly based on a negative d-dimer test 1-4 there is
a high and seemingly increasing demand for imaging tests for
suspected PE.
Some still regard invasive pulmonary angiography as the
“gold standard” technique for PE diagnosis, but in reality it is
infrequently performed 5-8 . It is perceived as an invasive procedure,
although the incidence of complications with contemporary
technique is low 9, 10 . More importantly there is accumulating
evidence describing the limitations of this technique for
the unequivocal diagnosis of isolated peripheral pulmonary
emboli: Two recent analyses of the inter-observer agreement
rates for detection of subsegmental emboli by selective pulmonary
angiography ranged between only 45% and 66% 11, 12 .
Given such limitations, use of this test as an objective and readily
reproducible tool for the verification of findings at competing
imaging modalities as to the presence or absence of PE seems
questionable and the status of pulmonary angiography as the
standard of reference for diagnosis of PE is in doubt 11, 12 .
Use of nuclear medicine imaging, once the first study in the
diagnostic algorithm of PE, is in decline 13, 14 due to the high
percentage of indeterminate studies (73% of all performed 15 )
and poor inter-observer correlation 16 . Revised criteria for the
interpretation of ventilation-perfusion scans 17, 18 and novel
technologies in nuclear medicine such as single photon emission
tomography (SPECT) 19, 20 can decrease the ratio of indeterminate
scintigraphic studies, but cannot offset the limitations
inherent to a functional imaging test 21 .
Contrast enhanced magnetic resonance (MR) angiography
has been evaluated for the diagnosis of acute PE 22-23 .
However, the acquisition protocols that are currently available
for MR pulmonary angiography lack sufficient spatial resolution
for reliable evaluation of peripheral pulmonary arteries 23, 24 .
More importantly, this modality has not seen widespread use in
the acutely ill patient with suspected PE due to lack of general
availability, relatively long examination times and difficulties in
patient monitoring.
This leaves us with computed tomography (CT), which for
most practical purposes has become the first line imaging test
after lower extremity ultrasound for the assessment of patients
with suspected PE in daily clinical routine. The most important
advantage of CT over other imaging modalities is that both
mediastinal and parenchymal structures are evaluated, and
thrombus is directly visualized 25, 26 . Studies have shown that
up to two thirds of patients with an initial suspicion of PE
receive another diagnosis 27 , some with potentially life-threatening
diseases, such as aortic dissection, pneumonia, lung cancer,
and pneumothorax 28 . Most of these diagnoses are
amenable to CT visualization, so that in many cases a specific
etiology for the patients’ symptoms and important additional
diagnoses can be established 21 . The inter-observer agreement
for CT is better than for scintigraphy 29 ; in a recent study the
inter-observer agreement for the diagnosis of PE was excellent
for spiral CT angiography (k= 0.72) and only moderate for ventilation–perfusion
lung scanning (k= 0.22) 16 . CT also appears
to be the most cost-effective modality in the diagnostic algorithm
of PE compared to algorithms that do not include CT, but
are based on other imaging modalities (ultrasound, scintigraphy,
pulmonary angiography) 30 . Additionally, there are some indications
that CT may not only be used for evaluating thoracic
anatomy in suspected PE but could also to some degree allow
derivation of physiologic parameters on lung perfusion at single-slice,
electron-beam and multidetector-row CT 31-33 .
The main impediment for the unanimous acceptance of computed
tomography as the new modality of choice for the diagnosis
of acute PE have been limitations of this modality for the
accurate detection of small peripheral emboli. Early studies
comparing single-slice CT to selective pulmonary angiography
demonstrated CT’s high accuracy for the detection of PE to the
segmental arterial level 34-37 but suggested that subsegmental
pulmonary emboli may be overlooked by CT scanning. The
degree of accuracy that can be achieved for the visualization of
subsegmental pulmonary arteries and for the detection of emboli
in these vessels with single-slice, dual-slice and electron-beam
CT scanners was found to range between 61% and 79% 36, 38-
40 , limitations that are overcome by novel developments in CT
technology.
Advantages of Multidetector-Row CT for PE Imaging
In the last few years CT has seen decisive dynamic developments,
mainly brought about by the advent of multidetector-row
CT technology 41, 42 . The current generation of 4-slice, 8-slice
and16-slice CT scanners now allows for acquisition of the entire
chest with 1-mm or sub-millimeter resolution within a short single
breath-hold of now less than 10 seconds in the case of 16slice
CT. The ability to cover substantial anatomic volumes with
high in-plane and through-plane spatial resolution has brought
with it a number of clear advantages. Shorter breath-hold times
were shown to benefit imaging of patients with suspected PE
and underlying lung disease reducing the percentage of nondiagnostic
CT pulmonary angiography investigations 43 . The
near isotropic nature of high-resolution multidetector-row CT
data lends itself to 2-D and 3-D visualization. This may, in some
instances, improve PE diagnosis 44 but is generally of greater
importance for conveying information on localization and extent
of embolic disease to referring clinicians in a more intuitive display
format. Probably the most important advantage is improved
diagnosis of small peripheral emboli. Single-slice CT has shown
that superior visualization of segmental and subsegmental pulmonary
arteries can be accomplished with thinner slice widths
(e.g. 2-mm versus 3-mm) 38 . However, with single-slice CT the
range of coverage with thin slice widths within one breath-hold
was limited 38, 40 . The high spatial resolution of 1-mm or submillimeter
collimation data sets now allows evaluation of pulmonary
vessels down to 6 th order branches 45 and significantly
increases the detection rate of segmental and subsegmental pulmonary
emboli 46 . This improved detection rate is likely due to
reduced volume averaging and the accurate analysis of progressively
smaller vessels by use of thinner sections. These results
are most striking in peripheral arteries with an anatomic course
parallel to the scan plane; such vessels tend to be most affected
by volume averaging when thicker slices are used 46 . The high
spatial resolution along the scan axis of a thin-collimation multidetector-row
CT data set, however, allows an accurate evaluation
of the full course of such vessels. The inter-observer correlation
for confident diagnosis of subsegmental emboli with
high-resolution multidetector-row CT by far exceeds the reproducibility
of selective pulmonary angiography 11, 12,46 .
119
MONDAY