Although the chest radiograph is usually the first imaging examination performed in patients with suspected PE, it may be completely normal even in the presence of near fatal PE. When abnormal, the findings are usually nonspecific. Chest radiographic findings are uncommon due to the dual blood supply from bronchial arteries. The main use of a chest radiograph is to exclude other disease processes that mimic the clinical signs and symptoms of PE, such as pneumonia, pneumothorax, rib fractures, aortic dissection, pleural effusions, pericardial effusion, tumor, and hiatal hernia (8). The radiograph is used in conjunction with the V/Q scan in the evaluation of PE, because it is used in the interpretation scheme for V/Q scans (9).

The chest radiograph findings are given in Table 21.2. Historical findings of PE on chest radiography are rarely encountered (Table 21.2), such as the Westermark sign, which is pulmonary oligemia distal to an obstructing embolus (Fig. 21.1), and the Fleischner sign, which is a large central pulmonary artery due to central thrombus with abrupt tapering, also known as the knuckle sign. These signs are mainly seen when there is PE without infarction. The Hampton hump, a wedge-shaped pleural based opacity with the apex pointing toward the hilum and the base abutting the pleura, particularly along the diaphragmatic pleura, is highly suggestive of pulmonary infarction. Pulmonary infarcts maintain their shape as they resolve, which may take several months. Occasionally, an infarct is mistaken for a solitary pulmonary nodule. Although most infarcts resolve completely, many leave an area of linear scar and occasionally a small nodule.

The nonspecific radiograph findings in PE with infarction include an elevated hemidiaphragm, pleural effusion, and atelectasis. Pleural effusions are quite common and are often small and unilateral. These often occur early and are frequently hemorrhagic. Occasionally, they can be large or bilateral. Effusions may be an isolated finding in up to one-third of patients with PE and are seen in approximately two-thirds of patients when accompanied by pulmonary hemorrhage or infarction.

In the original Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) study, the sensitivity and specificity of the V/Q lung scan was determined using pulmonary angiogram as the gold standard in the evaluation of suspected acute PE (10). The study concluded that clinical assessment combined with the V/Q scan established or excluded the diagnosis of acute PE in only a small number of patients. A normal chest radiograph was uncommon, found in 12% of patients. Radiographic signs poorly correlate with the diagnosis of PE and did not help to confirm or exclude the diagnosis of PE.

The principle underlying V/Q scintigraphy in the diagnosis of PE is the detection of decreased or absent perfusion, with corresponding normal ventilation. Perfusion scintigraphy is performed by intravenous injection of technetium-99m–labeled albumin macroaggregates ranging from 10 to 100 μm in size. These microparticles lodge in the precapillary arterioles of the pulmonary vascular bed. Eight static views are obtained (anterior, posterior, both laterals, and both anterior and posterior obliques). Ventilation scintigraphy is most often obtained after inhalation of radiolabeled gas such as xenon-133 or aerosols such as technetium-99m diethylenetriaminepentaacetic acid or technetium-99m pyrophosphate. Again, eight images are obtained as above, preferably in the upright position and especially with xenon. A normal V/Q scan is illustrated in Fig. 21.2.

Although the diagnosis of PE is based on a mismatched perfusion defect (i.e., perfusion defect in an area that is normally ventilated), the test is not specific for the diagnosis of PE. Most V/Q scan interpretations require interpretation in conjunction with a current chest radiograph. The current criteria used in the diagnosis of PE are the revised PIOPED criteria (11,12) (Table 21.3). Using these criteria, a V/Q scan is labeled as normal, very low probability, low probability, intermediate/indeterminate probability (Fig. 21.3), and high probability (Fig. 21.4) for the diagnosis of PE. Although a normal V/Q scan virtually excludes a PE and a high probability scan confirms a PE, in the PIOPED study only 14% of patients had a normal V/Q scan and 13% of patients had a high probability V/Q scan. The other 73% of patients had an indeterminate V/Q scan, a nonconclusive result requiring further evaluation with other studies, such as pulmonary angiography.

Pulmonary angiography is considered the reference test for the diagnosis of PE, against which all other tests are compared. Pulmonary angiography is recommended when there is an indeterminate V/Q scan, a discordance between the V/Q result (low probability, very low probability), and a high clinical suspicion for PE, indeterminate CTPA, or in cases of acute massive PE. It is particularly useful when catheter-guided thrombolysis or catheter-guided thrombectomy with direct injection of thrombolytic agent into the thrombus is to be performed. An invasive test, it requires a venous puncture usually at the groin and placement of a catheter through the right atrium and right ventricle into the main pulmonary artery, followed by selective placement in the right and left pulmonary arteries. Passage through the heart may induce an arrhythmia, including right bundle branch block. Therefore, patients with left bundle branch block should have a pacemaker in during the procedure. Contrast media is then injected and images obtained in the anteroposterior and oblique projections of each lung. It is a relatively safe invasive test, with a low mortality rate of 0.5% and morbidity rate of 3% to 5%, but is an extremely underused modality, used in only 12% to 14% of patients with nondiagnostic V/Q scans (4,13,14,15). Pulmonary angiography allows direct visualization of the clot burden. The angiographic signs of PE are a filling defect within an opacified pulmonary artery (Fig. 21.5) or occlusion of a pulmonary arterial branch (16,17). Absent or reduced opacification of small arterial branches is sometimes seen, corresponding to a perfusion defect on perfusion scintigraphy. A good quality negative pulmonary angiogram virtually excludes PE. However, angiography has its limitations. It is an invasive test that is underused and not available at all centers.

PE has incidentally been encountered on thoracic CT examinations as early as 1978. With the advent of new, faster, helical CT scanners, the ability of CT to detect PE as a primary test has received a great deal of interest (18,19). Techniques and scanners have evolved over the last decade; since the introduction of four-row multidetector and greater detector CT scanners, CT has taken on a more dominant role in the evaluation of suspected PE, right down to the subsegmental level (20). CTPA has been almost as reliable as pulmonary angiography in the detection of PE up to the segmental level; with multidetector CT, this reliability may also extend to the subsegmental level (21). Up to one-third of cases at the subsegmental level are misdiagnosed, even at pulmonary angiography (10). With thinner collimation, subsegmental emboli can be seen in 61% to 70% of cases (19,22). There is ongoing debate about the clinical significance of isolated subsegmental PE and whether these should be treated. Treatment may especially benefit patients with cardiopulmonary compromise; in patients with normal cardiopulmonary physiology there is more debate.

Iodinated contrast, 120 to 150 mL, is usually injected via the antecubital vein, at 4 to 5 mL/s. The timing of intravenous contrast bolus is crucial in obtaining diagnostic quality images. The technique has to be optimized for the CT scanner being used. Scanning is done from caudal to cranial direction to minimize streak artifact from the superior vena cava and to avoid breathing motion artifact from the diaphragm, during the latter part of the scan, if the patient cannot breath-hold.

Using single detector CT, the patient is scanned from the domes of the diaphragm to the top of the aortic arch in a single breath-hold following a scan delay of 25 seconds. Images are obtained at 2.5 mm to 3 mm collimation and reconstructed at 1.5 to 2 mm intervals. Using the 4 row multidetector CT scanner, because of the scanner speed, the entire chest can be scanned in 17 to 22 seconds, following a scan delay of 20 seconds at 1.25 mm collimation with images reconstructed at 0.625 mm intervals. Using 8-row detector multidetector CT the scan time is reduced again by half and on 16-row detector multidetector CT, to one-fourth, or approximately 5 seconds—of great benefit in patients with shortness of breath.

CTPA is interpreted on a workstation and read using a scrolling mode with altering of the window level and width dynamically for optimal evaluation of the pulmonary arteries. The scrolling mode is used to pan through the main, lobar, segmental, and subsegmental pulmonary arteries; hence, it is extremely important to be familiar with the pulmonary arterial anatomy.

Acute PE is principally diagnosed by visualizing a low attenuation filling defect within a well-opacified pulmonary artery (Figs. 21.6 and 21.7). Other CT findings of acute PE are listed in Table 21.4. Some of these signs include the “railway track” sign (Fig. 21.8D), the vessel cutoff sign, and the rim sign (Fig. 21.8C), where there is a filling defect due to thrombus with a rim of contrast around it. If there is occlusive thrombus, the corresponding artery may be larger and completely filled with low attenuation thrombus. In the case of nonocclusive thrombus, the rim sign may be seen or low attenuation thrombus can be seen in the center of the artery as a nonadherent thrombus or occasionally at the periphery. Ancillary signs of PE, although nonspecific by themselves, can be helpful in case of subtle thrombus. These can manifest as small pleural effusions, atelectasis, or pulmonary infarct distal to a PE (Fig. 21.8), oligemia or mosaic attenuation (Fig. 21.9) due to differential perfusion, and, in case of massive PE, signs of right heart strain with enlargement of the right ventricle and straightening of the interventricular septum.

The source of thrombus can occasionally be seen on the chest CT, such as in the superior vena cava, brachiocephalic veins, other neck veins, or the right atrium (Fig. 21.10). Pulmonary infarcts appear as triangular areas with a broad base in contact with the pleural surface and apex directed centrally, with a feeding vessel at the apex. Classic pulmonary infarcts do not enhance, whereas atelectasis or consolidation due to pneumonia often enhances. An advantage of CT is its capability of detecting thoracic abnormalities that mimic signs and symptoms of PE, aiding appropriate and immediate management of these patients (23). These abnormalities manifest as pleural, parenchymal, pericardial, or chest wall disease, seen in up to 40% to 70% of patients with suspected PE, such as pneumonia or abscess (Fig. 21.11), tumor (Fig. 21.12), septic emboli, malignant or nonmalignant pleural effusions, pericardial effusion (Fig. 21.13), pneumothorax, and chest wall tumors.

Pitfalls in the diagnosis of PE on CT are listed in Table 21.5. It is important to use the scrolling mode, so as not to mistake a vein for an artery, an adjacent perihilar node for an arterial filling defect, or mucoid-filled dilated bronchi for an arterial filling defect. Poor contrast bolus causing inadequate opacification of the pulmonary arteries, extensive breathing motion artifact, cardiac motion artifact, low signal-to-noise ratio due to large



patient size, and artifact from tubes and lines in critically ill patients can sometimes pose a challenge in the evaluation of PE in these patients (24,25,26).