|
|
||||||||
State of the Art |
1 From the Department of Radiology, Hospital Calmette, Boulevard Jules Leclerc, 59037 Lille, France. Received October 6, 1998; revision requested November 24; revision received January 5, 1999; accepted April 6. Address reprint requests to M.R.J. (e-mail: mremy-jardin@chru-lille.fr).
Abstract
Spiral computed tomographic (CT) angiography of the pulmonary circulation has emerged recently as a potential useful diagnostic method for the evaluation of the pulmonary circulation. As a minimally invasive examination, this technique is becoming widely available and has progressively replaced conventional and digital pulmonary angiography as the standard diagnostic imaging modality of the pulmonary circulation. The purpose of this review is to capture the current state of the art of the technical aspects of spiral CT angiography, with special emphasis on the postprocessing techniques currently available. As CT is responsible for a considerable part of the medical radiation dose applied to the population, the current trends for dose saving will also be emphasized. With regard to the clinical applications of spiral CT angiography, its introduction into the diagnostic work-up of pulmonary embolism has considerably modified the diagnostic algorithms. Our 8-year review of the literature presented herein is expected to provide the readers with a basis for formulating an informed opinion on this topic. In addition, a large number of congenital and acquired disorders of the pulmonary circulation are also relevant to this technique, and these indications are discussed in the context of the corresponding therapeutic options. Owing to the multiple possibilities inherent to this technique, spiral CT has the potential for cost savings without reduction in image quality or diagnostic accuracy.
Index terms: Computed tomography (CT), angiography, 944.12916, 945.12916 Computed tomography (CT), helical, 944.12915, 945.12915 Computed tomography (CT), volume rendering, 944.12917, 945.12917 Embolism, pulmonary, 60.72, 944.77 Lung neoplasms, CT, 60.30 Pulmonary arteries, CT, 944.12916 Pulmonary veins, 945.12915 State-of-art reviews
Several less-invasive modern technologies have largely replaced invasive procedures such as angiography for imaging of the pulmonary arteries. These techniques have the advantage of short exposure times and the ability to create three-dimensional (3D) data sets that have greater diagnostic possibilities than do standard projection angiographic images. Contrast materialenhanced spiral computed tomography (CT) provides these advantages. Since its introduction in the early 1990s, spiral CT angiography of the pulmonary circulation has progressively gained widespread acceptance, enabling one not only to obtain uniform and nearly constant opacification of pulmonary vessels down to 23 mm in diameter, but also to analyze the peripheral pulmonary vasculature with more precise anatomic details than those available with conventional angiographic studies.
The purpose of this article is to review the current clinical applications of spiral CT in the evaluation of a variety of congenital and acquired disorders of the pulmonary circulation, except the systemic arterial supply of the lung parenchyma. Although these disorders represent an increasing indication for spiral CT angiography in clinical practice, to our knowledge each specific indication has not been systematically evaluated in the literature. Therefore, in addition to a review of the available literature, several specific indications of this imaging modality will be suggested in the context of their expected clinical interest.
TECHNICAL APPROACH
Nonenhanced CT Examination
As reported for the evaluation of many vessels, spiral CT angiography of the pulmonary arteries results from a focal spiral CT examination that is usually preceded by nonenhanced CT scanning over the entire thorax. The value of this preliminary CT examination is threefold. First, it enables analysis of the lung parenchyma, pleura, and chest wall for associated abnormalities. Second, nonenhanced scans are helpful for identifying calcified lesions such as hilar lymph nodes (which could hamper analysis of contrast-enhanced CT scans) or calcified thrombi (which could be missed on contrast-enhanced images). Third, the preliminary CT examination enables an accurate localization of the anatomic volume of interest before spiral CT scanning. This preliminary CT examination can be performed with a conventional or spiral acquisition. When using the spiral mode, scanning with a low-dose technique is strongly recommended; this technique includes the selection of a pitch of 2 (especially when a subsecond scanning time is available), use of the lowest milliamperage available, and use of breast protection for young female patients. Although 180° interpolation and narrow beam collimation is ideal for a subsecond unit with the latest x-ray tube generating higher heat capacity, it should be emphasized that image noise in old spiral units with these parameters may substantially degrade the image.
Spiral CT Protocol
As recently reported by Fishman (1), the current challenges regarding CT angiography can be divided into three categories: data acquisition, data processing, and data display. For each step, several parameters have to be considered by the radiologist at the console to optimize the final display, as further detailed.
Acquisition Parameters
Scanning a distance of 1012 cm (from the aortic arch to 2 cm below the level of the inferior pulmonary veins), enables imaging of central and peripheral pulmonary arteries of the upper, middle, and lower lobes during the CT sequence in every patient. This region-of-interest may be covered by scanning the patient in the craniocaudal direction or scanning in the caudocranial direction. In clinical practice, the direction of the table feed is not a major issue, as it does not influence the overall quality of the spiral CT examination. More important is the selection of the region to be surveyed in order to precisely include the target vasculature within the volume scanned.
The radiologist must select the collimation and the speed of the table feed with two imperatives in mind. The first is to find the optimal compromise between the patient's respiratory status and the breath-holding duration necessary for evaluation of the pulmonary vasculature in a single CT acquisition. The second objective is to scan the patient with the thinnest collimation possible, to reduce partial volume effects. This second point is of great concern for an accurate evaluation of the peripheral pulmonary arteries. During the past few years, modification of the interpolation algorithms and introduction of subsecond scanning times have considerably modified the spatial resolution of spiral CT. The older technique of scanning with 5-mm collimation and a pitch of 1 and reconstructing the images with a 360° linear interpolation algorithm resulted in an effective section thickness of 6.57 mm. Consequently, the main drawback was partial volume effects at the level of small vessels, thus limiting the evaluation of peripheral pulmonary abnormalities (25). With selection of a 180° linear interpolation algorithm and a subsecond scanning time, the effective section thickness is of 3.86 mm when scanning with 3-mm collimation and a pitch of 1.7; the effective section thickness is reduced to 2.65 mm with the use of 2-mm collimation and a pitch of 2. Applied to the evaluation of the pulmonary circulation, use of 3-mm collimation was found to improve spatial resolution at the level of segmental arteries as suggested by a lower number of peripheral arterial branches coded as nonanalyzable (6). The recent availability of subsecond scanning times has made it possible to select a thinner collimation and to scan with a pitch of 2 without any modification in the breath-hold duration or the anatomic coverage. This protocol was shown to allow marked improvement in the analysis of segmental and subsegmental pulmonary arteries (7).
It is strongly recommended that patients be scanned while they maintain strict apnea. Interpretable images might possibly be obtained while the patient is gently breathing, but this should be reserved for only those patients with severe lung impairment. However, this latter acquisition condition is associated with two main limitations: a confident evaluation of the pulmonary circulation limited to central vessels and the inability to generate high-quality multiplanar and/or 3D reformations. Whenever possible, patients should be scanned at total lung capacity owing to the direct relationship between the degree of arterial opacification and the level of pulmonary arterial resistance. Physiologic studies have demonstrated that pulmonary resistance increases at high states of lung inflation because lung inflation causes the small alveolar vessels to be compressed (8,9). The end result is an increase in the resistance to flow in all the vessels exposed to alveolar pressure, which facilitates high-quality arterial opacification. Obtaining a deep inspiration may be difficult for patients suspected of having PE. In such situations, several physiologic considerations suggest that the patient should stop breathing near end-expiration, which can also achieve high levels of pulmonary resistance and thus good arterial opacification (8,9). Intubated patients referred from intensive care units can benefit from a spiral CT evaluation of the pulmonary circulation. The overall quality of these examinations may be excellent as it is possible to manually suspend ventilation for 1520 seconds in deep inspiration while the patient is sedated, thus avoiding motion artifacts.
Contrast Material Administration
The objective of the spiral protocol is to start scanning while the target vascular structures are opacified and to ensure a constant degree of pulmonary arterial opacification during the entire spiral sequence. An adequate examination of the pulmonary arteries requires the radiologist to select the most appropriate contrast agent and timing of injection, while carefully monitoring venous access. A power injection is mandatory for obtaining a homogeneous and constant level of arterial enhancement through the entire spiral examination. However, because of the rapid administration of contrast material, care must be taken to carefully adapt the rate of injection to the site of administration, especially when selecting high-flow protocols. In most spiral CT examinations, an 18- or 20-gauge catheter is inserted into a medially located antecubital vein, through which the bolus of contrast material is safely injected at a rate varying between 2 and 5 mL/sec. When an antecubital vein is not accessible, a more peripheral venous access can be used but care must be taken to avoid extravasation from a small and fragile peripheral vein. This requires reducing the rate of injection and increasing the start delay by 46 seconds. When a central venous access is available as frequently observed for patients referred from intensive care units, verifying the maximum rate of injection acceptable for a given catheter (often 45 mL/sec) is recommended. Because of the central location of the tip of the catheter, usually at the level of the superior vena cava, a shorter scanning delay is selected, which results in a shorter injection period.
The combination of an automated injection of contrast medium with the use of spiral CT enables an excellent quality of arterial opacification while using a smaller amount of iodine than previously recommended with sequential CT. Several approaches have been reported for the evaluation of pulmonary vessels with spiral CT, either selecting a low concentrationhigh flow protocol or a high concentrationlow flow protocol. The low concentrationhigh flow protocol consists in injecting 150240 mg/mL contrast agent at a rate of 45 mL/sec, whereas a high concentrationlow flow protocol uses 300350 mg/mL iodinated contrast agent administered at 23 mL/sec. In patients with normal right ventricular outflow, a 1215-second scanning delay usually provides sufficient latitude so that the pulmonary arteries are almost always well opacified. For patients with right ventricular failure and patients with pulmonary hypertension, a longer scanning delay is required, varying from 15 to 18 seconds. In most cases, an empiric determination of the scanning delay is sufficient to obtain excellent arterial enhancement. An alternative is to monitor contrast enhancement visually or with densitometry during the early stages of contrast material injection. This method provides a mechanism by which the time of scanning initiation can be individualized on the basis of actual enhancement of anatomic structures. When a desired level of enhancement is reached for a particular structure, transition is made to a routine diagnostic spiral imaging series. This technique is mainly indicated for patients with pulmonary hypertension and/or right ventricular failure. Although back-to-back helical scans are available, the interval of time between the two acquisitions is not compatible with adequate intravascular levels of enhancement within pulmonary branches over time.
Image Reconstruction
A brief summary of the technical peculiarities of the postprocessing techniques currently applicable to the pulmonary circulation is presented below.
Transverse CT scans.From each data set, transverse CT scans are systematically reconstructed. Creation of overlapping reconstructions is of major importance when scanning with a 5-mm collimation to avoid interpretive difficulties due to volume averaging effects. Owing to the considerable improvement in spatial resolution when scanning with a 2-mm collimation, reconstruction of contiguous sections is now widely accepted. An economic parameter (ie, the number of hard-copy films necessary to photograph overlapping images in a 2-mm collimation protocol) is the second major reason for generating contiguous images. Cost-reduction efforts regarding use of film by laser printers connected to spiral CT scanners are a challenge facing many medical centers. Consequently, reconstruction of overlapping sections from data sets acquired with thin collimation should be limited to the anatomic levels showing equivocal findings, namely, helping to distinguish an endoluminal clot from a partial volume effect with the arterial wall.
Multiplanar reformation.This technique is most commonly used to obtain sagittal or coronal views of transverse CT data sets, but it also can be valuable to provide oblique or curvilinear views of CT data sets. Multiplanar reformation (MPR) images completely eliminate superimposition of pixels lying outside the selected plane, and all the attenuation values included in the reformatted plane are visualized. With most available software programs, reformation planes are defined by projecting a line on one of the transverse sections so that the resultant MPR images will be perpendicular to the transverse plane and parallel to the projected line. MPR may also be planned on frontal or profile scout views. These planar reformations, having a single obliquity relative to the x, y, and z axes, do not take into account the orientation of the more cranial and caudal parts of the structure to be imaged. Whenever the structure of interest is not strictly parallel to the chosen plane, its changing orientations will not be considered in the final image. Consequently, inaccurately selected MPR may lead to erroneous interpretations, either due to the omission of part of the lesion or to the creation of a pseudoabnormality. When a structure of interest requires optimal double oblique planar views, the adequate plane of reformation of this structure can be chosen on a 3D image. Another alternative is to determine the plane of reformation by a direct selection of the voxels located at the center of the vascular lumen.
Three-dimensional images.Acquisition of the entire thorax within a single breath hold must be considered as a volumetric acquisition that offers multiple visualization techniques, which can be classified either according to their rendering, their projectional, or their computer-based image processing capabilities. As pointed out by Brink (10), a quality 3D image requires a combination of narrow section collimation and small intersection data reconstruction. As section thickness is a key factor, one should systematically envision the utility of a 3D reconstruction before selection of the acquisition protocol to be able to generate high-quality 3D images.
Until now, maximum intensity projection (MIP) has been the most popular method of obtaining 3D CT angiographic images (11,12). The MIP technique depicts the highest pixel value along the observer's line of sight and is particularly well suited for displaying small subvolumes in which superimposition of structures does not occur or is eliminated by editing procedures. However, several drawbacks have to be mentioned. The most important one is the inability to provide images of diagnostic quality when evaluating complex anatomic situations with superimposition of structures of interest. The depth information is absent unless a cine mode display with slightly varying viewing angles is used. According to the specificity of this technique for selection of voxels of interest, it is important to keep in mind that MIP does not enable the visualization of hypoattenuating intraluminal filling defects such as pulmonary emboli unless they are in close contact with the vessel wall.
The most frequently used 3D reconstruction of the pulmonary circulation is obtained with surface-rendering techniques. The shaded surface display (SSD) depicts the outer contour of the column of contrast material as an opaque surface. The main hallmark of SSD images is the thresholding segmentation that results in a binary classification of the voxels. Two categories of voxels are taken into account for the reformatted image: voxels higher than a user-selected single threshold or voxels included within a double-thresholding segmentation. For 3D SSD images of the pulmonary vasculature generated from nonenhanced data sets, the threshold values range from -600 to -700 HU. Should contrast medium be used, the thresholding segmentation has to be changed from -700 HU to +100 to 150 HU. Four categories of pitfalls can be encountered with this technique of reconstruction (13). The first group concerns the pitfalls related to the thresholding range, explaining why the stenosed part of a tubular structure can be artificially narrowed with too high a threshold or artificially minimized with too low a threshold. In such cases, an additional 3D display should be systematically performed with a lower or a higher thresholding segmentation, respectively, to avoid interpretive errors. However, low threshold values create a complex vascular background that may prevent optimal viewing of the vessel of interest, explaining why this kind of selection is usually obtained at second intention. The second group of pitfalls is related to the section thickness, either responsible for partial volume artifacts or stair-step artifacts. The third category of pitfalls may be caused by motion during data acquisition, either because of the patient's inability to maintain strict apnea or because of organ pulsation. The last group of pitfalls is related to contrast enhancement, which may be variable along the z axis.
The most recently introduced technique for creating angiographic-like images of the pulmonary vasculature is the volume-rendering technique. In this technique, classification of the voxels estimates the probability of a tissue type being homogeneously present in a voxel and attributes to this voxel a percentage corresponding to the amount of this tissue. This classification mode is called percentage classification or probabilistic classification, and a trapezoid is attributed to each tissue type (14). Several parameters have to be selected to generate volume-rendering technique reconstructions: (a) position of the trapezoid, which determines the attenuation of voxels incorporated into the image (this selection consists of defining a window width and a window level [15]); (b) shape of the trapezoid, corresponding to the intersections between the sides of the trapezoid and the top; (c) opacity value, which determines the relative transparency of the voxels incorporated into the image, and thus the degree of vascular transparency; and (d) shading option (not systematically available on volume-rendering programs), which modifies the vessel wall smoothness. Preliminary evaluation of the volume-rendering technique at the level of the pulmonary circulation has led to the selection of the following parameters: triangular trapezoid shape, selection of voxels from -750 to -450 HU for nonenhanced vessels and from 50 to 200 HU for enhanced vessels, 80% opacity value, and unshaded algorithm (authors' unpublished data, 1997). This selection was found to create diagnostically useful images of the pulmonary circulation, easily applicable in clinical practice. With continued advances in computer technology, interactive volume-rendering techniques soon will be widely available, an imperative for further evaluation of this technique at the level of the pulmonary circulation and rigorous comparison with alternative reconstruction techniques (16).
Slabs.The delineation of peripheral pulmonary vessels is facilitated by the low attenuation of the surrounding lung parenchyma, explaining why their analysis does not systematically require injection of contrast medium. With regard to the technical particularities of a spiral CT evaluation of the peripheral vascular bed, one should mention the usefulness of generating slabs from nonenhanced focal spiral CT examinations. On cross-sectional images, a linear or tubular structure obliquely oriented or perpendicular to the plane of the CT section is segmented proportionally to the section thickness. Conversely, the stacking of several contiguous transverse images allows one to benefit from the advantages of the summation effect, thus enabling an easy identification of a pulmonary vessel in its entirety from the hilum to the periphery. Further anatomic details can be obtained when thin-slab MIP with or without automatic sliding in the acquired volume as well as targeted SSD images are obtained. The most frequently used technique is called sliding thin-slab (STS) MIP, which consists of an automatic sliding through the volume of interest (17). Because the selection of a narrow collimation is mandatory for high-quality reconstructions, care should be taken to precisely determine the volume to be scanned, usually limited to 23 cm.
Current Trends for Dose Saving
Because CT is responsible for a considerable part of the medical radiation dose applied to the population, concerted efforts are being made to minimize the dose required for a spiral CT examination without making compromises in image quality. To reach this goal at the level of the thorax, several recommendations can be made. First, a decreased kilovoltage can be easily applied to routine spiral CT angiography of the pulmonary circulation and the milliamperage setting can be individually selected. The use of subsecond scanning and the selection of a pitch of 2 are two additional efficient means of reducing the radiation dose. Moreover, dose modulation methods have been recently developed by manufacturers; these methods enable the tube current and hence the dose applied to be adjusted to the individual patient's geometry and patient's absorption during the acquisition. From preliminary evaluations, dose savings of up to 40% are obtained, depending on the patient and the examined region of the thorax (18). As recently proposed by Hopper et al (19), the use of thin layers of radiation absorbent material such as bismuth can provide breast radiation protection to women without affecting the quality of diagnostic CT images. These authors emphasized the seriousness of the dose administered to such a radiosensitive organ, reminding us that a woman's breast receives 2.03.5 rad (0.0200.035 Gy) of radiation from a thoracic CT examination. To emphasize the importance of this radiation dose, one should keep in mind that delivery of 1 rad (0.01 Gy) of radiation to a woman's breasts before age 35 years fractionally increases her risk of breast cancer by 13.6% over the expected spontaneous rate for the general population.
CLINICAL INDICATIONS: ACQUIRED DISEASES
Pulmonary Embolism
Diagnosis of Acute Pulmonary Embolism
CT features and pitfalls.Until the early 1990s, the standard diagnostic imaging procedures for evaluation of pulmonary embolism (PE) were ventilation-perfusion (V-P) scintigraphy and pulmonary angiography. During the past few years, there has been a progressive replacement of pulmonary angiography with spiral CT angiography, as the latter imaging procedure allows a noninvasive identification of intravascular clots. The CT criteria for acute PE are the tomographic equivalent of the classic angiographic signs of PE (ie, partial or complete filling defects and the "railway track" signs) (Figs 1, 2). In terms of contrast material administration, the drawbacks of the high concentrationlow flow protocol argue for the use of the low concentrationhigh flow protocol in the evaluation of PE. The injection of a highly concentrated material usually results in streak artifacts at the level of the superior vena cava, which can substantially degrade quality in the adjacent right main pulmonary artery but also the right upper lobe pulmonary artery and its segmental branches. At the level of these branches, this may hamper detection of intraluminal changes. Although the low rate of injection recommended with this protocol is usually sufficient for depicting large emboli, it is not always compatible with the identification of emboli at the level of segmental arteries and precludes any detection of subsegmental emboli. The ideal concentration of contrast material for low concentrationhigh flow protocols has not yet been definitely determined, to our knowledge, but many users agree to consider a 150200 mg/mL concentration.
|
|
A number of interpretive pitfalls exist in assessing enhanced spiral CT images, but their recognition is less and less problematic as the radiologist gains experience with spiral CT of the pulmonary vasculature (23). Four main categories of technical or pathophysiologic factors may be responsible for pseudofilling defects. In severely tachypneic patients, breathing artifacts may be observed at the level of obliquely oriented arteries due to the variable position of the vessel in the section width in two successive scans. Consequently, any abrupt decreased attenuation at the level of a given arterial section between two contiguous sections should not be mistaken for an arterial filling defect. The choice of the scanning delay may also influence the quality of arterial enhancement at the top or bottom of the scanned volume. If the scanning delay is too short, there is insufficient time to allow adequate opacification of the pulmonary arteries on the first images. On the other hand, if the scanning delay is too long, there is not enough material left at the bottom of the selected subvolume, which may hamper detection of intraluminal thrombi. Physiologic data must also be integrated in the spiral volumetric CT evaluation of vascular enhancement. Several cases of unilateral increased pulmonary vascular resistance due to extensive airspace consolidation have led to a false-positive diagnosis of PE (2,4). The asymmetric pulmonary vascular resistance results in asymmetry of arterial perfusion due to slow flow through the pulmonary arteries ipsilateral to the lung consolidation, with normally circulating contralateral pulmonary arteries. Any cause of unilateral increased vascular resistance may lead to asymmetry in pulmonary arterial opacification. The various causes of nonopacification or faint unilateral opacification of a pulmonary artery that have been previously described on conventional angiograms and lung V-P scans (24,25) can be summarized as follows: (a) "occult" pulmonary artery; (b) unilateral pulmonary artery obstruction of extrinsic, mural, or endoluminal origin; (c) unilateral increase in pulmonary vascular resistance secondary to proximal or distal bronchial obstruction, bronchiolar obstruction, parenchymal destruction, extensive airspace consolidation, pleural restrictive or expansive process, or conditions with elevated venous pressure; and (d) congenital and acquired hemithoracic shunting of blood from the systemic to the pulmonary circulation.
The best way to demonstrate on CT scans the patency of a unilateral faintly opacified pulmonary artery is to perform a second spiral CT acquisition with a longer scanning delay to suppress flow phenomena at the level of this vessel. There are acquired disorders that are associated with anatomic or surgical shunting from the systemic to pulmonary circulation and that are thus responsible for antegrade or retrograde left-to-right shunts. This situation results in regional or unilateral changes in pulmonary blood flow that is interrupted at the level of the anatomic communication between the systemic and pulmonary circulations. The most common clinical situation is represented by patients with severe chronic inflammatory disease, particularly those with bronchiectasis, in whom prominent bronchopulmonary collateral circulation may develop and produce a proximal retrograde flow. In these patients, non-visualization of arteries or dilution defects may simulate emboli (Fig 3), as previously reported on conventional pulmonary angiograms (24). In patients with a substantial regional variation in pulmonary blood flow, a long start delay (eg, 20 seconds) and selection of a high concentrationhigh flow protocol can help suppress the focal hypoattenuation in the pulmonary arterial bed related to retrograde left-to-right shunting. However, the difficulties in obtaining optimal enhancement of the pulmonary arterial vasculature on CT scans are similar to those encountered on unilateral angiograms on which poor opacification or pseudoamputation are commonly seen. In most cases, hyperselective catheterization of the lobar or segmental arteries of this region is required to demonstrate their patency. A transient or permanent increase in pulmonary arterial pressure may be responsible for a right-to-left shunt through a patent foramen ovale, which results in a lower degree of pulmonary arterial enhancement on CT scans. This situation is suspected when a massive and early enhancement of the aorta is observed in conjunction with a faint pulmonary arterial opacification.
|
|
|
|
value was 0.85 with spiral CT angiography and 0.61 with V-P scanning. In the European multicentric trial (32), the agreement among CT scan readers was also better than that found among V-P scan readers (ie, 0.72 vs 0.39) (32). This study showed a considerable dependance of the interobserver agreement in CT angiography on the technical quality of the study, whereas interobserver agreement in V-P scintigraphy was neither dependent on study quality nor on the use of Biello or PIOPED criteria. These results explain why several authors suggest that spiral CT should be the initial imaging modality of choice, especially in the group of patients known to be associated with a high rate of indeterminate V-P scintigraphic studies (eg, all inpatients and patients with chronic obstructive pulmonary disease) (6,28,29,30,39). These conclusions are reinforced by the results of the cost-effectiveness analysis recently reported by Van Erkel et al (34). A limited number of patients could thus benefit from scintigraphy at first intention (ie, patients with low clinical suspicion of PE and no prior cardiopulmonary disease). An important question should then be considered: What are the remaining indications for pulmonary angiography? They should be reserved for patients in whom clinical suspicion for PE remains high despite normal results at spiral CT and duplex ultrasonography (US) of the legs. However, one should nuance the answer by considering the technical quality of the negative spiral CT angiogram. When the technical conditions of the latter examination limit the depiction of acute PE to central pulmonary arteries, pulmonary angiography should be reasonably indicated to search for thrombi at the level of the peripheral arterial bed. However, when a spiral CT angiogram is negative down to the subsegmental level, the clinical usefulness of pulmonary angiography is more debatable. Recent improvement in scanner technology is leading the radiologists to reevaluate the role of spiral CT in the detection of subsegmental PE; the frequency of subsegmental PE as reported in the literature varies between 2% and 33% (4,6,33,34,40,41). Over the past few years, several studies have stressed that subsegmental pulmonary arteries are not accurately evaluated with spiral CT. This inability was the logical consequence of the scanning parameters chosen in early reports (ie, 5-mm collimation and 5 mm/sec table feed [pitch = 1]) responsible for partial volume effects at the level of small branches. Since then, selection of a 3-mm collimation and a 5 mm/sec table feed (pitch = 1.7) has improved evaluation of peripheral pulmonary arteries at routine CT examinations. As previously noted, a 2-mm collimation and a 4 mm/sec table feed (pitch = 2) are now available in routine clinical practice. In a study group of 370 patients, the percentage of the analyzable subsegmental arterial bed was statistically significantly higher when scanning the patients with a 2-mm collimation and a pitch of 2 than when scanning the patients with a 3-mm collimation and a pitch of 1.7 (65% vs 43%, respectively). A question often raised over the past few years is, what is the clinical relevance of peripheral emboli in acute PE? Several authors consider that tiny clots are from calf veins and do not require anticoagulation (4,4245). However, occlusion of a few subsegmental branches perfusing the most "normal" part of the lung parenchyma has been reported to lead to respiratory failure in patients with preexistent bronchopulmonary disease (6,33).
Follow-up of Acute PE and Management of Chronic Thromboembolic Disease
In addition to facilitating the diagnosis of acute PE, spiral CT angiography may help us understand changes within central pulmonary arteries after acute PE. In a study group of 19 patients with acute PE initially identified with spiral CT, resolving clots were found in 13 of 19 patients (68%) at 6-week follow-up (46). Residual abnormalities consisted of eccentric, wall-adherent filling defects or filling defects with central contrast material, consistent with recanalized clots. In a study group of 62 patients, follow-up spiral CT angiograms over a mean period of 11 months showed an incomplete resolution of acute PE in 39% of cases and newly developed chronic PE in 13% of cases (47) (Fig 4). The morphologic changes directly available on spiral CT angiograms are expected to improve our knowledge of the evolution of acute PE toward chronic thromboembolic disease. To date, to our knowledge only a few scintigraphic studies have addressed this issue, and the development of chronic changes has been reported with a frequency varying between 2% and 18% of cases (4850). The most conspicuous abnormality of chronic PE is the presence of complete filling defects at the level of stenosed pulmonary arteries. Several additional features are also suggestive of chronic PE, namely, eccentric thrombi, evidence of recanalization, and arterial stenosis or web. The creation of reformatted images through the longitudinal axis of obliquely oriented vessels can overcome some of the difficulties encountered with transverse sections in the identification of focal arterial stenosis (51). In addition, reformatted images provide a delineation of mural thrombi on a single image, which is usually considered a useful complement to pulmonary angiography whenever a surgical treatment is planned.
|
|
CT can also depict changes at the level of the systemic arterial circulation, the collateral supply of the occluded pulmonary arterial bed. The most common findings on CT scans consist of dilated bronchial arteries, which can be detected as tortuous and/or dotted vascular sections in their mediastinal and/or hilar pathways (54,55). Enlargement of the internal mammary arteries is also easily detectable on CT scans, especially in cases of asymmetric or unilateral systemic supply. In addition, the possibility of identifying lung parenchymal abnormalities, such as areas of ground-glass attenuation and/or bronchial dilatation in cases of chronic PE, should be emphasized (56).
Pretherapeutic Management of Bronchial Carcinoma
Pulmonary Arterial Invasion from Bronchial Tumors
Precise staging is the key to rational management of bronchial tumors. Incremental CT has fallen short in predicting tumor extension in the mediastinum and the hilum, including vascular invasion. An improvement in CT-surgical correlations is expected with spiral CT with routine use of 25-mm thin sections (Fig 5). From clinical experience, we recommend performance of a contrast-enhanced spiral CT examination of the hila after analysis of the nonenhanced CT scans of the entire thorax. Careful interpretation of these sections helps determine the most appropriate injection protocol with special attention directed toward the target vasculature, the iodinated concentration, and the optimal start delay. The routine protocol for spiral CT angiography of the pulmonary circulation virtually guarantees an optimal opacification not only of the mediastinal and hilar pulmonary arteries but also of the venoatrial confluence and of the cardiac chambers. Strictly comparative performances of spiral CT acquisition are mandatory, before and after neoadjuvant chemotherapy, for a better estimation and, if needed, for a response quantification. This has to be taken into account as neoadjuvant chemotherapy has an initial response rate of approximately 50%, and subsequent resection rates in the responders may approach 80%.
|
A hilar lymph node dissection can be influenced by the spiral CT findings. If an upper lobectomy is contemplated, lymph nodes smaller than 10 mm in diameter, outside the lobectomy field, can be discovered by their compression or invasion of the descending pulmonary artery. They run the risk of being unseen on an incremental CT study because they hardly modify the external pulmonary artery morphology. For the same reason, mediastinal lymph node imaging can also be improved by means of spiral CT angiography of the mediastinal vessels (60). Radiographically occult squamous cell carcinoma involving main stem bronchi and lobar bronchi can benefit from photodynamic therapy. These lesions may be extremely small and are very often beyond the ability of CT detection. But once they are endoscopically detected and pathologically proved, the two major problems are to determine whether the lesion is limited to the bronchial wall and to eliminate an N2 disease. Another point of interest is that, if there is an incomplete response after photodynamic therapy, the patient can undergo surgical resection. Spiral CT with thin sections and contrast medium can depict the peribronchial tissues and be used in the extrabronchial follow-up of the treated disease. Khanavkar et al (61) pointed out that centrally located squamous cell carcinomas represented a predisposing factor to hemorrhage and tried to identify patients at high risk of developing fatal hemoptysis after endobronchial brachytherapy. They emphasized the following anatomic factors: involvement of major arteries, considerable bronchial wall destruction, and mediastinal invasion. Posttreatment necrotic cavitation of a tumor located a few millimeters from a major vessel can also be considered as a potentially lethal situation. The preventive effect of spiral CT angiography should be seriously considered in the follow-up of this treatment, as well as after vaso-occlusive therapy. Whereas bronchial carcinoids can be highly vascular lesions, they may only be depicted with a moderate homogeneous contrast enhancement on incremental CT scans (62). The transmural extension of this tumor can reach the peribronchial tissues and vessels, as well as the adjacent pulmonary parenchyma. If an endobronchial biopsy is anticipated, the highly vascular component should be demonstrated to prevent any further hemorrhagic complications. A sleeve resection can be performed depending on the peribronchial extent. Both findings can be searched for on pulmonary and systemic spiral CT angiograms.
Venoatrial Extension of Bronchial Carcinoma
In addition to invasion and obstruction of a pulmonary vein by a lung tumor, embolization into the left atrium may occur (Fig 6). The most common tumor able to invade the left atrium in this way is bronchogenic carcinoma. However, metastatic carcinoma and pulmonary sarcoma of the lung have also been reported with a similar extension. A preoperative diagnosis is imperative since the consequence of intraoperatively dislodging a macroscopic tumoral thrombus from the pulmonary vein can be multiple systemic tumor emboli and can be fatal in the short term (63). In patients with large central carcinomas in contact with one or several major pulmonary veins, a preoperative transthoracic or transesophageal echocardiogram and a spiral CT angiogram are recommended because a carcinomatous thrombus of a pulmonary vein may go unnoticed at surgery and become dislodged during surgical manipulations. The CT signs of venous invasion are those of a pulmonary veno-obstructive syndrome, including pulmonary edema and hemorrhage, pulmonary venous infarct, interlobular septal thickening, increased bronchial wall thickening, and pleural effusion (64). Spiral CT angiography can additionally show a massive enlargement in diameter of the invaded vein with an absence of opacification or a more or less notable filling defect of the adjacent part of the left atrium. The focal increase in pulmonary vascular resistance can also result in absent or delayed opacification of the pulmonary arteries in the area of venous obstruction, followed later by an unusual intense opacification of these pulmonary arteries due to pulmonaryarterial stasis. On incremental CT scans, a space-occupying lesion in the left atrium could be easily missed on an insufficiently contrast-enhanced CT scan (65).
|
|
|
In the particular group of patients who have undergone lung transplantation, spiral CT can be helpful for the evaluation of anastomosis-related stenoses of the arteries and veins. The frequency of pulmonary artery stenosis has been recently estimated by Griffith et al (69) who observed obstructions in five of 60 single-lung transplant recipients and in none of 74 double-lung transplant recipients. In one case, obstruction was associated with an intraluminal thrombus. The association of systemic hypoxia, normal chest radiographic findings, and perfusion scan abnormalities represents an indication for spiral CT angiography whenever the patient is able to support breath holding for a few seconds. It is not easy to make the differential diagnosis between a true stenosis and a distorted or kinked anastomosis resulting from excessive lengths of donor and recipient artery cuff. The latter can spontaneously improve over a follow-up period, in contrast to a true stenosis, which leads to a surgical revision or to endovascular procedures. Owing to pulmonary-to-bronchial artery retrograde flow, pulmonary artery stenosis can compromise the bronchial anastomosis. In the absence of contraindications to contrast medium administration, a spiral CT study initially indicated for the follow-up of airway anastomosis can also allow evaluation of arterial and venous anastomoses. When the transplant recipient has to be treated with surgery, percutaneous angioplasty, and/or endoprosthesis (70), the precise shape, site, and length of the stenosis can be better appreciated with two-dimensional and 3D reformations because they permit unlimited viewing angles (Fig 8). Perfusion scanning and 3D spiral CT angiography can also avert the use of more invasive and repeated angiographic follow-up.
|
|
|
Imaging studies can contribute to the care of patients with pulmonary hypertension by helping to (a) detect its presence, (b) indicate possible causes, (c) grade its severity, and (d) evaluate the functional state of the right ventricle. Routine chest radiography, CT, MR imaging, pulmonary function studies, and echocardiography have all been studied with variable success in predicting the presence or severity of pulmonary hypertension in patients with underlying cardiopulmonary disease. More than 10 years ago, a main pulmonary artery diameter of 29 mm or greater on a CT scan was reported to have a sensitivity of 69% and a specificity of 100% for predicting pulmonary hypertension in patients with cardiac or intrinsic pulmonary vascular disease (72). Because CT of the chest is often used in the diagnosis and treatment of patients with parenchymal lung disease, Tan et al (73) undertook a retrospective study aimed at determining the diagnostic utility of CT-determined diameter of the main pulmonary artery for predicting pulmonary hypertension in this patient population. The diameter was measured at the widest portion of the main pulmonary artery within 3 cm of the bifurcation; in addition, the right pulmonary arterial diameter and left pulmonary arterial diameter were measured at the widest portions after bifurcation of the main pulmonary artery. The authors concluded that a main pulmonary artery diameter of 29 mm or greater enables identification of patients with even a mild degree of pulmonary hypertension and therefore should enable identification of those patients who may have the greatest pulmonary vascular limitation to exercise.
In addition to the enlargement of proximal pulmonary arterial branches, CT may help identify morphologic changes suggestive of pre- or postcapillary pulmonary hypertension. In favor of precapillary pulmonary hypertension, one may note small peripheral pulmonary arteries and veins with the concurrent presence of normal lung parenchyma and left atrium. Presence of postcapillary pulmonary hypertension is usually suggested by the identification of CT features of interstitial and/or alveolar edema, including septal lines and abnormal lung attenuation varying from ground-glass opacities to consolidation. These parenchymal abnormalities are observed with the concurrent presence of enlarged pulmonary veins, sometimes associated with pleural effusion.
During the past few years, several studies have shown that a mosaic pattern of lung attenuation can be seen in patients with pulmonary hypertension of various causes (74). In an attempt to determine whether infiltrative, airway, or vascular lung disease is the cause of mosaic attenuation on thin-section CT scans of the lung, one may pay attention to the size of the pulmonary vascular sections within the areas of ground-glass attenuation and search for air trapping. The areas of increased lung attenuation can be attributable to blood flow redistribution when the size and number of vessels in the areas of increased attenuation are increased compared with those in the areas with normal or decreased attenuation in the absence of air trapping (7577). As recently demonstrated by Worthy et al (78), it should be emphasized that, whereas the diagnoses of infiltrative lung disease and airway disease are frequently correct and confident, vascular disease is more difficult to diagnose on thin-section CT scans as the cause of areas of ground-glass attenuation.
Primary Pulmonary Artery Sarcomas
Primary pulmonary artery sarcomas are infrequent tumors that are very often confused with acute and chronic thromboembolic disease. Apart from the particular endovascular development of the tumor, which can be seen as a floating mass resembling a endoluminal clot within the pulmonary artery trunk and/or its proximal branches, it should be emphasized that this tumor is often associated with thrombus formation. The absence of predisposing factors for PE, the absence of "thrombus" dissolution despite anticoagulation, and atypical distribution of the filling defects as seen at two-dimensional or Doppler US, pulmonary angiography, MR imaging or CT, and V-P scanning should alert the radiologist (79). Peripheral pulmonary infiltrates can correspond to pulmonary infarcts due to thrombi or embolization of tumor material from the original site.
The diagnosis can be strongly suspected when chest radiography, CT, or MR imaging shows a lobulated and heterogeneous hilar mass that originates from the pulmonary artery trunk or from the main right or left pulmonary arteries and expands the artery (80). Such polypoid intraluminal masses can be partly hemorrhagic, necrotic, and hypervascularized with enhancement on spiral CT scans or after administration of gadopentetate dimeglumine on T1-weighted MR images. This finding helps eliminate a thrombus and indicates a neoplastic process (81,8284). The pretherapeutic evaluation of this tumor can be performed with spiral CT, first during the passage of the contrast medium bolus, but also a few minutes later to search for contrast medium uptake by the tumor (Fig 9). In a minority of patients, the tumor arises from the right or left main pulmonary artery with or without contralateral spreading. Direct transmural growth in the hilum, the mediastinum, the lung parenchyma, and the pericardium can occur or it can be simultaneously multicentric. Such lesions can mimic a tumor from another origin, particularly a bronchial carcinoma invading the pulmonary artery. Extravascular infiltration and nodal invasion are also possible (85), and it should be noted that osteo-, chondro-, and carcinosarcomas may exhibit osteoid and bony components (79).
|
|
|