CT angiography allows the direct visualization of emboli as arterial filling defect that partially or totally occludes the enhanced artery (Ghaye et al. 2002). A partial arterial defect surrounded by contrast material produces the so-called “polo mint” sign on images acquired perpendicular to the long axis of the vessel (Fig. 2) and the “railway track” images sign on longitudinal images of the vessel (Fig. 3). A partial filling defect can place peripherally forming acute angles with the arterial wall (Fig. 4). When the embolus totally occludes the artery, the involved vessel may be enlarged (Fig. 5) compared with the corresponding patent vessel of the opposite site (Han et al. 2003). A vessel “cut-off” sign is seen when the distal artery is not opacified owing to the presence of occlusive disease. Central filling defects can sometimes already be highlighted on CT scans without contrast medium as hyperdense images at the level of the main pulmonary arteries (Fig. 6a, b). The visualization of the thrombus on chest CT without contrast medium is related to the “age” of the clot and the haematocrit of the patient at the time of the survey. The gradual retraction of the thrombus is associated to the reduction of its water content resulting in an increase of density values up to 50–80 UH (Kanne et al. 2003).

In the presence of PE, parenchymal CT findings can be seen as areas of ground glass opacity or consolidation representing pulmonary haemorrhage that usually resolve within a week. Pulmonary infarction is more frequently seen in lower lobes as a wedge-shaped area of consolidation with the apex towards the hilum and with central hypodense areas representing secondary pulmonary lobules spared from the disease process (Fig. 7a). It is often possible to highlight the presence of an intraluminal filling defect in the arterial branching afferent to the pulmonary infarction (Fig. 7b).

Pulmonary infarction is seen in about 10 % of patient with PE, and it gradually reduces in size in 3–5 weeks retaining its triangular morphology. The excavation of pulmonary infarction is rare; it can also be found in the absence of infectious processes especially if the pulmonary infarction has a transverse diameter >4 cm and generally occurs after 2 weeks of onset of the infarct (Fig. 7d). Linear parenchymal bands, atelectasis, focal oligaemia and small pleural effusion are less frequent features.

CT can also highlight findings suggestive of right ventricular failure as dilation of the right ventricle, with right ventricular cavity transverse diameter greater than the left, with or without the presence of reflux of contrast medium in the hepatic veins and deviation of the interventricular septum towards the left ventricle (Fig. 8).

Several conditions can be responsible of acute PE misdiagnosis (Wittram et al. 2004). Respiratory motion artefacts are the most common cause of indeterminate CT pulmonary angiography. These artefacts are best seen with lung window settings and can create the “seagull” sign. Images obtained in large patients have more quantum noise making the evaluation of segmental and subsegmental vessel very difficult. A soft tissue reconstruction algorithm should be used to avoid high attenuation around vessels. We recommend using pulmonary embolism-specific settings with a window width and level of 800 and 100 HU, respectively. Streak artefacts from dense contrast material within the superior vena cava or from a central venous catheter can obscure emboli or can mimic PE. These artefacts can be distinguished from pulmonary emboli for their radiating nature and the nonanatomic morphology (Pena and Dennie 2012). These artefacts can be reduced by flushing the superior vena cava with saline solution using dual chamber injectors. A poor contrast opacification of pulmonary arteries can be also a cause of misdiagnosis of PE. Hilar lymph nodes can be confused with emboli even if, nowadays, thanks to the use of thinner collimations, lymph nodes are easily differentiated from emboli (Fig. 9a, b). Low-density mucus-filled bronchi and pulmonary veins can also mimic pulmonary filling defect. The review of sagittal and coronal reformatted images can be helpful in difficult cases.

Most cases of pulmonary emboli resolve without sequelae. In a small percentage of patients with PE, especially when large emboli are present, these can undergo organization, retraction and recanalization, with the possibility of determining frameworks of pulmonary hypertension and cor pulmonale (Hoeper et al. 2006). The most frequent CT findings are complete occlusion of a vessel that is smaller than adjacent patent vessel, a peripheral filling defect with obtuse angles with the vessel wall (Fig. 10), web or flap within a contrast material-filled artery, presence of calcifications in the context of the parietal thrombus or calcification of the pulmonary arteries (Fig. 11a, b) and contrast material flowing through thickened often smaller arteries due to recanalization (Fig. 12b). The main pulmonary artery is typically enlarged (diameter, >29 mm) for the onset of pulmonary arterial hypertension, and bronchial artery dilation is also present (Fig. 12a). Pulmonary parenchymal CT findings in chronic PE consist of “mosaic pattern”: lung areas with a different density in a patchy distribution resulting from irregular perfusion of lung parenchyma (Fig. 13) (Pena and Dennie 2012).

Chronic pulmonary embolism in the appropriate clinical setting should be differentiated from pulmonary artery sarcoma (PAS). It is a rare tumour with poor prognosis, and it should be considered as a possible diagnosis in the absence of thromboembolic risk factors and when there is no evidence of cancer elsewhere. Patients are generally in their 50s and do not show any response to anticoagulation.

Pulmonary artery sarcoma is an uncommon cause of intraluminal filling defect causing often a unilateral hilar enlargement (Cox et al. 1997; Hu et al. 2009). CT angiography of the chest shows tissue which tends to occupy the full diameter of the vessel with vascular involvement of the common trunk (85 % of cases) and the main branches (left pulmonary artery, 65 %; right pulmonary artery, 71 %); inhomogeneous enhancement of the intraluminal pulmonary tissue due to haemorrhage, necrosis and calcification (Fig. 14); increase in vessel calibre; and extra-luminal extension of the tissue. Pulmonary nodules can also be an expression of metastases (Long et al. 2010).

Over the last decade, different imaging modalities have been established as useful tools in the diagnosis of PE, with various ranges of diagnostic performance being reported. Choice of the imaging modality varies by site, with usage of a particular modality influenced by a combination of site history, availability of equipment, physician personal preference and patient suitability (Phillips et al. 2015).

For 30 years ventilation/perfusion (V/Q) lung scanning has been the preferred non-invasive imaging procedure in patients with suspected PE. However, a complex probability scoring system and a relative high number of non-diagnostic scans have dampened enthusiasm about this procedure.

In the past decade, CTPA has proven to be a more accurate test with relatively few inadequate studies and has been welcomed by clinicians (Stein et al. 2006). Nevertheless, concerns about radiation exposure and side effects of contrast medium administration with CTPA indicate that a role for V/Q scintigraphy in the evaluation of patients with suspected PE is still to consider (Freeman 2008).