Imaging evaluation of suspected pulmonary embolism

Michael B. Gotway, Keith J. Edinburgh, Vickie A. Feldstein, Jonathan D. Lehman, Gautham P. Reddy, W. Richard Webb

Research output: Contribution to journalArticlepeer-review

26 Scopus citations


Venous thromboembolism (VTE) is a common disorder that is difficult to diagnose clinically but carries significant morbidity and mortality if untreated. Additionally, although demonstrated to be of benefit in cases of proven deep vein thrombosis (DVT) and pulmonary embolism (PE), anticoagulation therapy is not without risk. Because the clinical exam is known to be unreliable for the detection of both DVT and PE, many imaging modalities have been used in the diagnostic imaging algorithm for the detection of VTE, including chest radiography, ventilation/perfusion (V/Q) scintigraphy, pulmonary angiography, and recently, spiral computed tomography (CT) and magnetic resonance imaging (MRI). Chest radiographic findings in acute PE include focal oligemia, vascular enlargement, atelectasis, pleural effusions, and air space opacities representing pulmonary hemorrhage or infarction. The chest radiograph can occasionally be suggestive of PE but is more often nonspecifically abnormal. The main use of the chest radiograph in the evaluation of suspected PE is to exclude entities that may simulate PE and to assist in the interpretation of V/Q scintigraphy. Lower extremity venous compression ultrasonography (CU) is both sensitive and specific for the diagnosis of femoropopliteal DVT, and the value of negative CU results has been established in outcomes studies. However, the reliability of CU for the detection of isolated calf vein thrombosis is not well established, and the clinical significance of such thrombi is debatable. Additional methods such as color and spectral Doppler analysis are also useful in the diagnostic evaluation of DVT but are best considered as adjuncts to the conventional CU examination rather than as primary diagnostic modalities themselves. Compression ultrasonography and Doppler techniques are useful in the evaluation of suspected upper extremity DVT; spectral Doppler waveform analysis is particularly useful to assess for the patency of veins that cannot be directly visualized and compressed with conventional gray-scale sonography. V/Q scintigraphy has been the initial modality obtained in patients suspected of PE for a number of years. Although many studies have investigated the role of V/Q scintigraphy in the evaluation of VTE, the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) study has provided the most useful information regarding the utility of V/Q scintigraphy in this setting. A high probability scan interpretation is sufficient justification to institute anticoagulation, and a normal perfusion scan effectively excludes the diagnosis of PE. A normal/near normal scan interpretation also carries a sufficiently low prevalence of angiographically proven PE to withhold anticoagulation. Although the prevalence of PE in the setting of low probability scan interpretations is low and several outcomes studies have demonstrated a benign course in untreated patients with low probability scan results, patients with inadequate cardiopulmonary reserve do not necessarily have good outcomes. Such patients deserve more aggressive evaluation. Patients with intermediate probability scan results have a 20% to 40% prevalence of angiographically proven PE and thus require further investigation. The radionuclide investigation of DVT includes such techniques as radionuclide venography and thrombus-avid scintigraphy. Although these methods have not been as thoroughly evaluated as CU, studies thus far have indicated encouraging results, and further investigations are warranted. Pulmonary angiography has been the gold standard for the diagnosis of PE for decades. Studies have indicated that angiography has probably been underutilized by referring physicians for the evaluation of suspected PE, likely because of the perception of significant morbidity and mortality associated with the procedure. However, recent investigations have demonstrated that advances in catheter and wire design as well as low-osmolar, non-ionic contrast material and recognition of the danger of non-selective contrast injection in the setting of elevated pulmonary arterial pressures have significantly improved the safety of pulmonary angiography. Thus the morbidity and mortality of untreated PE or inappropriate anticoagulation in the absence of PE likely outweigh the risks of this procedure. Spiral CT has recently been shown to be an effective tool for the diagnosis of acute and chronic PE. Spiral CT is diagnostic for PE in the central pulmonary vessels, and sensitivity and specificity are high in the proximal vasculature. However, the sensitivity of CT is diminished when subsegmental vessels are considered. The clinical significance of subsegmental emboli is uncertain, and some investigators feel that such emboli are likely insignificant in patients with adequate cardiopulmonary reserve. Outcomes investigations with CT will be required to firmly establish the value of a negative CT interpretation. The utility of MRI in the evaluation of suspected PE has also been investigated. A limited number of studies have indicated sensitivities and specificities similar to CT. However, the technical considerations involved in MR imaging of the pulmonary vasculature will likely limit this modality to situations in which CT and scintigraphy are not feasible or available. Although no imaging modality is ideal for the evaluation of suspected VTE, an algorithm for the ordered approach to patients suspected of this disorder is presented. Patients with symptoms of unilateral DVT or patients with bilateral lower extremity symptoms but with risk factors for DVT should be examined with CU supplemented with color and spectral Doppler imaging as needed. Otherwise-healthy outpatients and patients with normal chest radiographs may be first evaluated with scintigraphy. Nondiagnostic scans may be further investigated with CU if symptoms of DVT are present; in the absence of DVT symptoms, spiral CT may be chosen. In patients with symptoms of DVT but a negative CU examination, CT may be obtained: For patients without symptoms of DVT but a negative spiral CT, a negative CU examination would provide further reassurance that the patient will do well without treatment. In patients with abnormal chest radiographs or in hospitalized patients, V/Q scans would likely be nondiagnostic and thus spiral CT may be chosen as the initial modality to evaluate such patients for potential VTE. Pulmonary angiography may be reserved for patients with inadequate cardiopulmonary reserve and non-diagnostic scintigraphy and negative spiral CT and CU examinations.

Original languageEnglish (US)
Pages (from-to)133-184
Number of pages52
JournalCurrent Problems in Diagnostic Radiology
Issue number5
StatePublished - 1999

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging


Dive into the research topics of 'Imaging evaluation of suspected pulmonary embolism'. Together they form a unique fingerprint.

Cite this