CT Pulmonary Angiography for Pulmonary Embolism

CT pulmonary angiography (CTPA) is the first-line imaging test for the diagnosis of pulmonary embolism (PE), a potentially fatal condition caused by thrombotic occlusion of the pulmonary arteries. Also called computed tomography pulmonary angiography, CT angiography of the pulmonary arteries, or simply CTPA, this modality has become the reference standard for detecting PE in most clinical settings ^[4]B2a. Pulmonary embolism itself is also known as pulmonary thromboembolism, and its subtypes, acute, chronic, subsegmental, massive, and submassive, are defined by symptom duration, thrombus location, and hemodynamic impact.

Clinical Significance

PE is a major contributor to cardiovascular morbidity and mortality worldwide, with an annual incidence of approximately 1 per 1,000 adults. The case fatality of untreated PE is high, but prompt diagnosis and anticoagulation reduce mortality substantially. CTPA has transformed the diagnostic approach by enabling rapid, accurate detection of thrombus from the main pulmonary arteries down to subsegmental branches ^[2]D5. Its high specificity (reported at 0.94-0.99 in network meta-analyses) and sensitivity (0.76-0.93) make it a reliable rule-in and rule-out test ^[1]B2a.

Historical Development

Before CTPA, conventional pulmonary angiography was the gold standard, but it was invasive and carried procedural risks. Ventilation/perfusion (V/Q) scintigraphy offered a noninvasive alternative but had high rates of indeterminate results. The introduction of multidetector CT in the late 1990s allowed for rapid, high-resolution imaging of the pulmonary arteries, and CTPA quickly replaced older modalities as the first-line test. Its adoption increased the detection of subsegmental PE, a finding that was often missed by V/Q scanning and whose clinical significance remains debated ^[2]D5.

Current Role

Today, CTPA is the preferred initial imaging modality for suspected PE in nonpregnant adults and is also used in pregnancy when chest radiography is abnormal or when alternative diagnoses (e.g., aortic dissection, pneumonia) are suspected ^[5]D5. It is the standard for diagnosing PE in hospitalized patients, including those with COVID-19 , where the incidence of CTPA-confirmed PE was 2.2% in non-ICU wards during the first wave ^[8]B3b. The test is widely available, can be performed emergently, and provides additional information about lung parenchyma, mediastinum, and chest wall. However, it exposes patients to ionizing radiation and iodinated contrast, prompting ongoing efforts to reduce dose and volume through low-kVp protocols and iterative reconstruction ^[4]B2a.

Classification of Pulmonary Embolism

These classifications guide management decisions and are informed by CTPA findings such as thrombus location, right ventricular strain, and clot burden. The technical principles and protocol of CTPA are designed to optimize pulmonary artery opacification while minimizing patient risk, as discussed in the following section.

Pearl: A negative CTPA of adequate technical quality effectively rules out PE in patients with low or moderate pretest probability; in high-probability patients, a negative study should prompt consideration of alternative diagnoses or further testing.

Building on the diagnostic role of CTPA, the technical protocol directly determines image quality and diagnostic confidence. Optimal pulmonary artery opacification, minimal motion artifact, and low radiation and contrast doses are achieved through careful selection of scan parameters, contrast injection strategy, and reconstruction techniques.

Contrast Injection Protocols

Pulmonary artery attenuation above 200 HU is the standard diagnostic threshold. A prospective randomized trial comparing three contrast protocols on dual-layer spectral CT found that 40 mL at 4 mL/s and 30 mL at 3 mL/s achieved >200 HU in over 90% of conventional images, whereas 20 mL diluted 1:1 with saline at 4 mL/s reached this threshold in only 70% ^[9]A1b. However, low-energy virtual monoenergetic images (LEVMI) at 40-55 keV raised attenuation to a minimum of 269 HU across all protocols, making even the 20 mL dose diagnostically acceptable ^[9]A1b. A separate randomized trial of 328 patients compared dilution (1:1 with saline), low flow (low injection rate), and low concentration (140 mgI/mL) strategies using a fixed low contrast dose on spectral detector CT; no significant difference in pulmonary trunk attenuation was found, though the low-concentration group showed higher attenuation across all anatomical levels ^[11]A1b. These data support flexibility in contrast administration when spectral CT is used.

Scan Acquisition Parameters

Low tube voltage protocols (≤80 kVp) are the cornerstone of dose reduction. A systematic review and meta-analysis of 35 studies demonstrated that low-kVp CTPA combined with iterative reconstruction reduces CT dose index (CTDIvol) and effective dose by 50% to over 80% while preserving diagnostic image quality ^[4]B2a. Tube current (mA) is adjusted to maintain acceptable noise, typically using automatic exposure control. For dual-source CT, a multiphase acquisition using 15 mL of contrast (vs. 40 mL for single-phase) with nonrigid temporal registration provided similar or superior subjective image quality and diagnostic confidence, with significantly reduced contrast usage ^[14]A1b.

Breathing Maneuvers and Timing

Transient interruption of contrast (TIC), a streaming artifact caused by inspiratory inflow of unopacified blood, degrades image quality in up to 12.3% of standard end-inspiratory breath-hold CTPA ^[12]A1b. A prospective randomized trial of 150 patients found that a device-assisted Mueller maneuver (suction against resistance using a prototype) completely eliminated TIC (0% vs. 12.3%, p=0.005) and improved overall contrast opacification (p<0.001), though breathing artifacts were more frequent (48.1% vs. 30.1%, p=0.038) without clinical consequence ^[12]A1b. The Mueller maneuver is an effective strategy to prevent TIC and ensure uniform pulmonary artery enhancement.

Dose Reduction Strategies

Dual-low dose protocols combine low tube voltage (≤80 kVp) with reduced contrast volume (≤60 mL) and iterative reconstruction. Deep learning reconstruction (DLR) with contrast-enhancement boost (CE-boost) further improves image quality. In a two-center prospective trial of 130 patients, DLR-boost images from a dual-low dose protocol (reduced radiation and contrast) achieved significantly higher signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) than hybrid iterative reconstruction (HIR) from a routine-dose protocol, with comparable CT values and an area under the curve (AUC) of 0.986 for PE detection ^[10]A1b. These techniques enable substantial dose reduction without compromising diagnostic accuracy.

Advanced Reconstruction and Spectral Techniques

Spectral detector CT allows reconstruction of low-energy virtual monoenergetic images (40-55 keV) that boost iodine attenuation, enabling diagnostic studies with contrast doses as low as 20 mL ^[9]A1b. Iodine maps and Z-effective maps provide perfusion information; although the lowest contrast dose (20 mL) produced lower-quality iodine maps, it still detected perfusion defects in all PE cases ^[9]A1b. Deep learning reconstruction combined with CE-boost offers an alternative pathway to dose reduction without spectral hardware ^[10]A1b.

Pearl: For patients at risk of contrast-induced nephropathy, a dual-low dose protocol using 80 kVp, 30 mL contrast, and deep learning reconstruction can achieve diagnostic images while reducing radiation by >50% and contrast by 40% ^{[4]B2a[10]A1b}.

Building on the technical protocol, patient preparation and contrast administration directly determine image quality and safety. Pre-procedure evaluation, contrast delivery strategy, and breath-hold technique must be optimized for each patient.

Pre-procedure Assessment

Renal function must be assessed before contrast administration. Estimated glomerular filtration rate (eGFR) <30 mL/min/1.73 m² identifies patients at highest risk of contrast-induced nephropathy (CIN). A history of prior contrast reaction, asthma, or atopy should be elicited; premedication with corticosteroids and antihistamines is indicated for patients with prior moderate-to-severe allergic reactions. Intravenous access of at least 18-20 gauge in an antecubital vein is preferred to accommodate flow rates of 3-5 mL/s. Poor access or small veins increase the risk of contrast extravasation and suboptimal opacification ^[16]A1c.

Contrast Administration Strategies

Standard CTPA protocols use 40-60 mL of iodinated contrast (300-370 mg I/mL) injected at 4-5 mL/s, followed by a saline flush. However, substantial dose reduction is achievable without sacrificing diagnostic quality. A systematic review and meta-analysis found that low tube voltage (≤80 kVp) combined with iterative reconstruction allows contrast volumes as low as 15-20 mL while maintaining attenuation above 200 HU ^[4]B2a. In a prospective trial, a high-pitch dual-source protocol with 80 kVp and a dual-flow contrast/saline bolus delivered a mean of 17 mL of Iohexol 350 mg I/mL (range 12-20 mL), yielding 100% adequate or better examinations ^[17]C4.

Spectral detector CT enables further reduction. A randomized trial comparing 40 mL, 30 mL, and 20 mL of contrast (diluted 1:1 with saline) found that all three protocols achieved diagnostic pulmonary artery attenuation on low-energy virtual monoenergetic images (≥269 HU), though conventional images with 20 mL fell below 200 HU in 30% of cases ^[9]A1b. Another trial showed that dilution, low flow, and low concentration strategies produced equivalent pulmonary trunk attenuation at 40 keV, with the low-concentration group (140 mg I/mL) actually yielding higher attenuation across all anatomical levels ^[11]A1b. Deep learning reconstruction (DLR) combined with contrast-enhancement boost (CE-boost) further improves image quality at low doses: in a two-center trial, DLR-boost images from a dual-low dose protocol (reduced radiation and contrast) had significantly higher signal-to-noise and contrast-to-noise ratios than routine hybrid iterative reconstruction, with an AUC for PE detection of 0.986 ^[10]A1b. Multiphase CTPA with nonrigid temporal registration can use as little as 15 mL of contrast while providing similar or better subjective image quality than single-phase protocols using 40 mL ^[14]A1b.

Breath-Hold and Breathing Maneuvers

Standard end-inspiratory breath-hold is the conventional instruction, but the Mueller maneuver (forced inspiration against resistance) improves contrast dynamics. In a randomized trial of 150 patients, device-assisted Mueller maneuver (using a prototype Contrast Booster™) eliminated the transient interruption of contrast (TIC) phenomenon, which occurred in 12.3% of patients with standard breath-hold (p=0.005). Mean pulmonary trunk attenuation was 338 HU with Mueller vs 314 HU with standard command (p=0.157), and the pulmonary trunk-to-aorta ratio was significantly higher (3.86 vs 2.26, p=0.001). Breathing artifacts were more frequent with Mueller (48.1% vs 30.1%, p=0.038) but without clinical consequence ^[12]A1b.

Contrast-Induced Nephropathy Prevention

For patients with eGFR <30 mL/min/1.72 m², prophylactic hydration with isotonic sodium bicarbonate or normal saline (1 mL/kg/h for 6-12 hours before and after the study) is recommended. Low-contrast-dose protocols (≤20 mL) substantially reduce CIN risk and may allow CTPA in patients previously considered ineligible. The use of low kVp and iterative reconstruction is particularly beneficial in this population ^{[4]B2a[16]A1c}.

Special Considerations

Body habitus affects image quality: obese patients may require higher tube voltage (100-120 kVp) or increased contrast volume to maintain adequate attenuation. Reduced cardiac output delays contrast bolus arrival; a longer scan delay or bolus tracking with a lower trigger threshold (e.g., 100 HU in the pulmonary trunk) can compensate ^[16]A1c.

Pearl: For patients at high risk of contrast-induced nephropathy, a low-kVp (≤80 kVp) CTPA protocol with ≤20 mL of contrast and iterative reconstruction can provide diagnostic images while minimizing renal injury, consider this the default approach in eGFR <30 mL/min/1.73 m².

With optimal contrast opacification achieved, the interpreting radiologist must systematically evaluate the pulmonary arteries for filling defects and assess for ancillary findings that confirm or refute the diagnosis of acute pulmonary embolism (PE). Accurate interpretation requires knowledge of normal anatomy, direct and indirect signs of PE, severity markers, and common pitfalls that lead to misdiagnosis ^[16]A1c.

Normal Pulmonary Artery Anatomy and Variants

The main pulmonary artery arises from the right ventricle and bifurcates into right and left main branches at the level of the carina. The right pulmonary artery courses anterior to the right main bronchus, while the left pulmonary artery passes over the left main bronchus. Segmental arteries (10 on the right, 8 on the left) and subsegmental branches must be traced to the lung periphery. Normal anatomic variants include a separate origin of the right upper lobe artery (truncus anterior) and accessory fissure arteries. Lymph nodes, bronchi, and pulmonary veins can mimic filling defects if not carefully distinguished ^[16]A1c.

Direct Signs of Acute Pulmonary Embolism

The definitive sign of acute PE on CTPA is a filling defect within the contrast-opacified pulmonary artery lumen. Three morphologic patterns are recognized:

• Complete occlusion: The artery is entirely filled by thrombus, often with a convex margin (the "doughnut sign" on axial images).
• Partial occlusion: A central or eccentric filling defect surrounded by contrast, producing the "railway track" or "tram track" sign when the defect is longitudinal.
• Saddle embolus: A large thrombus straddling the bifurcation of the main pulmonary artery, extending into both right and left main branches. This is a high-risk finding associated with hemodynamic instability.

Acute thrombi typically appear low-attenuation (30-50 HU) and may expand the affected artery. In contrast, chronic thrombi are often eccentric, calcified, or show web-like bands ^[16]A1c.

Indirect Signs and Severity Assessment

Indirect signs of PE include:

• Right heart strain: Enlargement of the right ventricle (RV) relative to the left ventricle (LV) on axial or four-chamber views. An RV/LV diameter ratio >1.0 is the most validated marker of right ventricular dysfunction and predicts adverse outcomes ^[16]A1c.
• Interventricular septal bowing: Leftward deviation of the septum indicates elevated right ventricular pressure.
• Contrast reflux: Opacification of the inferior vena cava or hepatic veins suggests right heart failure.
• Mosaic perfusion: Heterogeneous lung attenuation due to regional oligemia distal to occluded vessels, more common in chronic thromboembolic disease.

Severity can be quantified using the obstruction index (e.g., Mastora or Qanadli scores), which assign points based on the location and degree of occlusion. These scores correlate with RV dysfunction and mortality but are not routinely used in clinical practice ^[16]A1c.

Alternative Diagnoses and Incidental Findings

CTPA often reveals alternative causes of the patient's symptoms. Common findings include:

• Pneumonia: Consolidation with air bronchograms.
• Aortic dissection: Intimal flap in the thoracic aorta.
• Pericardial effusion: Fluid density surrounding the heart.
• Pulmonary edema: Ground-glass opacities and septal thickening.
• COVID-19 pneumonia: Bilateral peripheral ground-glass opacities with vascular enlargement, which can mimic PE on clinical presentation ^[28]D5.

Incidental pulmonary nodules, mediastinal lymphadenopathy, and pleural effusions are also frequently detected and require appropriate follow-up.

Pitfalls and Mimics

Interpretation errors are common and can be minimized by recognizing the following mimics ^[16]A1c:

• Motion artifact: Cardiac or respiratory motion creates blurring or pseudo-filling defects, especially at the lung bases.
• Beam-hardening artifact: Dense contrast in the superior vena cava or subclavian veins can obscure adjacent pulmonary arteries.
• Lymph nodes: Enlarged hilar or mediastinal lymph nodes may compress or simulate intravascular filling defects.
• Mucus plugs: Bronchial mucus can mimic an intraluminal defect if the bronchus is closely apposed to the artery.
• Pulmonary vein thrombus: Thrombus in the pulmonary veins (e.g., from atrial fibrillation ) can be mistaken for arterial PE.
• Stripe sign: A thin line of contrast between thrombus and the vessel wall, indicating chronic rather than acute PE.
• Atelectasis and fissures: Linear opacities from collapsed lung or interlobar fissures can simulate a filling defect on iodine maps ^[21]B2b.

Chronic Thromboembolic Disease

Chronic PE presents with different imaging features: eccentric, crescentic filling defects that are adherent to the vessel wall, web-like bands or intimal irregularities, and complete occlusion with distal vessel tapering. The affected artery may be smaller than adjacent normal vessels. Chronic thromboembolic pulmonary hypertension (CTEPH) is diagnosed when precapillary pulmonary hypertension is confirmed by right heart catheterization and imaging shows chronic thromboembolic changes despite at least 3 months of therapeutic anticoagulation ^[26]D5. V/Q scan is the preferred screening test for CTEPH, as a normal scan essentially rules out the disease ^[26]D5.

Role of Dual-Energy CT and Iodine Maps

Dual-energy CTPA generates iodine subtraction maps that depict lung perfusion. In acute PE, perfusion defects correspond to the vascular territory of the occluded artery. A prospective study reported sensitivity of 81.3% (95% CI 76.4-85.4%) and specificity of 78.9% (95% CI 74.1-83.1%) for segmental perfusion defects compared with SPECT ^[21]B2b. Pitfalls include atelectasis, fissures, and beam-hardening artifacts, which can cause false-positive perfusion defects ^[21]B2b. Iodine maps may improve detection of subsegmental PE and help assess hemodynamic significance.

Pearl: The most actionable step in CTPA interpretation is to measure the RV/LV ratio on axial images; a ratio >1.0 identifies patients at increased risk of adverse outcomes and should be reported in every positive study ^[16]A1c.

Interpretive criteria provide the framework; the diagnostic performance of CTPA is defined by pooled estimates from large meta-analyses and prospective trials. As the established gold standard, CTPA serves as the reference against which all alternative modalities are measured. In the largest meta-analysis of deep learning (DL)-based PE detection (24 studies, 22,984 patients), DL algorithms compared to CTPA demonstrated a pooled AUROC of 0.895, sensitivity 0.894 (0.856-0.923), and specificity 0.871 (0.831-0.903) ^[29]A1a. Positive and negative predictive values were 0.832 (0.794-0.869) and 0.902 (0.874-0.929), respectively ^[29]A1a. These figures underscore CTPA's high accuracy while also highlighting that even expert interpretation leaves room for improvement, particularly in specificity.

Impact of CT Technology and Reconstruction

Modern CT technology and advanced reconstruction algorithms further enhance CTPA's diagnostic performance. Deep learning reconstruction (DLR) combined with contrast-enhancement boost (CE-boost) in a dual-low-dose protocol achieved an AUC of 0.986 for PE detection, significantly higher than hybrid iterative reconstruction (HIR) alone ^[10]A1b. Low-dose CTPA with iterative reconstruction (IR) maintains subjective image quality comparable to standard-dose protocols, with only a non-significant 4.0% and 5.5% reduction in signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR), respectively, while reducing radiation exposure substantially ^[34]A1b. Third-generation dual-source CT with dynamic reconstruction (4D noise reduction TMIP-CTA) yields significantly higher SNR and CNR and improved subjective image quality scores compared to standard CT angiography ^[35]A1b. Spectral detector CT allows low-contrast-dose protocols without compromising vascular attenuation, with no difference in pulmonary trunk attenuation between dilution, low-flow, or low-concentration strategies ^[11]A1b. These technological advances ensure that CTPA maintains its diagnostic edge even as radiation and contrast doses are minimized.

Comparison with Alternative Modalities

CTPA's diagnostic performance consistently exceeds that of alternative imaging modalities when directly compared. In meta-analyses, MRI for PE diagnosis shows a pooled sensitivity of 0.89 (0.79-0.94) and specificity of 0.94 (0.89-0.97), with an AUC of 0.97 ^[33]A1a. Non-contrast MR pulmonary angiography (NC-MRPA) yields sensitivity 0.88 (0.83-0.91) and specificity 0.97 (0.93-0.98), with an AUC of 0.92 ^[31]A1a. In contrast, V/Q scintigraphy has lower sensitivity (0.81, 0.76-0.85) and specificity (0.84, 0.74-0.91), with an AUC of 0.87 and a non-diagnostic rate of 34.7% (vs. 3.31% for NC-MRPA) ^[31]A1a. Transthoracic lung ultrasound (LUS) shows sensitivity 0.80 (0.71-0.86) and specificity 0.87 (0.81-0.92), with an AUC of 0.91 ^[20]B2a. Multi-organ point-of-care ultrasound (POCUS) in critically ill patients has high sensitivity (0.90, 0.85-0.94) but lower specificity (0.69, 0.42-0.87), with a diagnostic odds ratio of 25.3 (4.43-82.9) ^[30]A1a. These comparisons confirm that CTPA remains the most accurate single test, particularly for its high specificity and low false-positive rate.

Subsegmental Emboli Considerations

Detection of subsegmental pulmonary embolism (SSPE) remains a challenge for all imaging modalities. CTPA's sensitivity for SSPE is lower than for more proximal emboli, though modern multi-detector CT with thin collimation (≤1 mm) and advanced reconstruction algorithms improves visualization. The high pooled sensitivity of DL algorithms (0.894) suggests that AI-assisted interpretation may reduce missed SSPE, but prospective validation is lacking ^[29]A1a. The clinical significance of isolated SSPE is debated, and overdiagnosis must be balanced against the risk of unnecessary anticoagulation.

Pearl: CTPA's diagnostic performance is excellent for central and lobar PE (sensitivity >95%), but sensitivity declines for subsegmental emboli; AI-assisted interpretation may narrow this gap, though prospective validation is still needed ^[29]A1a.

The diagnostic performance data above establish CTPA as the reference standard for acute PE, yet its clinical utility depends on balancing several advantages against well-recognized limitations.

Advantages of CTPA

CTPA offers several practical advantages that have driven its adoption as the first-line imaging test for suspected PE. It is widely available in emergency departments and radiology suites, with acquisition times of seconds to minutes, enabling rapid triage of unstable patients. Beyond PE detection, CTPA frequently identifies alternative diagnoses such as pneumonia, aortic dissection, or pericardial effusion, a key advantage over ventilation-perfusion scintigraphy (V/Q), which has a pooled non-diagnostic rate of 34.7% ^[31]A1a. Diagnostic accuracy is high: spectral CT pulmonary angiography (S-CTPA) achieves a pooled sensitivity of 94.2% (95% CI 88.3-100%) and specificity of 88.5% against a V/Q reference standard ^[36]A1a. Even with reduced contrast doses, spectral CTPA maintains diagnostic vascular attenuation: protocols using as little as 20 mL of iodinated contrast (diluted with saline) still detect perfusion defects in all PE cases ^[9]A1b. Low-dose protocols employing iterative reconstruction (IR) or deep learning reconstruction (DLR) preserve image quality while reducing radiation exposure by up to 74% ^{[19]A1b[34]A1b}. DLR combined with contrast-enhancement boost (CE-boost) further improves signal-to-noise ratio and diagnostic accuracy (AUC 0.986) at dual-low radiation and contrast doses ^[10]A1b.

Limitations and Mitigation Strategies

Despite these strengths, CTPA carries inherent limitations that require careful consideration.

Radiation exposure is a primary concern, particularly in young patients and those requiring repeat imaging. However, modern low-dose techniques substantially mitigate this risk. A prospective trial using 70 kVp with automatic tube current modulation and IR reduced effective dose significantly (CTDIvol and DLP both p<0.001) while maintaining subjective image quality comparable to standard-dose protocols ^[34]A1b. Ultra-low-dose CTPA with DLR achieved a 74% reduction in effective dose without compromising diagnostic performance, and AI-assisted reading further improved sensitivity from 79.8% to 91.7% and specificity from 95.5% to 99.2% ^[19]A1b.

Contrast-induced acute kidney injury (CI-AKI) and contrast allergy remain contraindications in some patients. For those at risk, alternative strategies include using low-contrast-dose spectral CTPA protocols (e.g., 30 mL at 3 mL/s or 20 mL diluted with saline) that still achieve diagnostic attenuation in virtual monoenergetic images ^{[9]A1b[11]A1b}. Dual-energy CT can also provide iodine maps to assess perfusion defects with reduced contrast volumes ^[16]A1c. When contrast is absolutely contraindicated, non-contrast MR angiography (NC-MRPA) offers a feasible alternative with pooled sensitivity 0.88 and specificity 0.97, and a low non-diagnostic rate of 3.31% ^[31]A1a.

Overdiagnosis of subsegmental PE is a recognized pitfall, as small filling defects may be clinically insignificant but lead to anticoagulation and its attendant risks. Technical factors such as motion artifact, poor opacification, and beam hardening can mimic or obscure emboli ^[16]A1c. Understanding these pitfalls and using optimal contrast timing and reconstruction algorithms reduce false positives. AI-assisted reading may also help by flagging equivocal findings for expert review ^[19]A1b.

Incidental findings on CTPA (e.g., pulmonary nodules, mediastinal masses) can prompt unnecessary follow-up imaging and procedures. While some findings are clinically important, the rate of incidental findings requiring workup is non-negligible. Radiologists should report incidental findings with management recommendations based on established guidelines to balance benefit and harm.

Controversies and Guideline Disagreement

Pearl: The decision to perform CTPA should incorporate patient-specific risks (radiation, contrast, renal function) and local expertise; when contrast is contraindicated, NC-MRPA offers comparable accuracy with fewer non-diagnostic scans than V/Q scintigraphy ^[31]A1a.

Despite the high accuracy of CTPA, its diagnostic yield depends critically on appropriate patient selection; overuse leads to unnecessary radiation, contrast exposure, and cost ^{[40]A1c[43]D5}. The Choosing Wisely campaign explicitly recommends against performing chest CT angiography in patients with a low clinical probability and negative results of a highly sensitive D-dimer assay ^[43]D5.

Pretest Probability Assessment

Validated clinical prediction rules, the Wells criteria and revised Geneva score , stratify patients into low, intermediate, or high pretest probability before any imaging ^{[40]A1c[42]A1c}. The Wells criteria are the most studied; pooled analysis of implementation studies shows that their use increases CT angiography yield from 9% to 12% (relative risk 1.3, 95% CI 1.1-1.6) ^[44]A1a.

For very low-risk patients, the PERC rule (Pulmonary Embolism Rule-Out Criteria) can safely avoid further testing. In the PROPER randomized trial, a PERC-based strategy reduced CTPA use from 23% to 13% (difference -10%) with no significant increase in missed thromboembolic events during follow-up (0.1% vs 0%) ^[45]A1b.

Role of D-Dimer

In patients with low or intermediate pretest probability, high-sensitivity D-dimer is the appropriate initial diagnostic test ^{[40]A1c[42]A1c}. For patients >50 years, use an age-adjusted threshold: age × 10 ng/mL rather than the generic 500 ng/mL cutoff ^[40]A1c. This adjustment reduces false-positive results without missing clinically significant pulmonary embolism. In patients with a high pretest probability, D-dimer should not be obtained, proceed directly to imaging ^{[40]A1c[42]A1c}.

Indications for CTPA

CTPA is indicated in patients with high pretest probability ^[40]A1c or in those with intermediate probability and a positive D-dimer ^[42]A1c. When CTPA is contraindicated (e.g., severe contrast allergy, impaired renal function), ventilation-perfusion scan (V/Q) is an alternative ^{[40]A1c[42]A1c}. The American College of Radiology Appropriateness Criteria provide guidance for imaging in specific scenarios, including evaluation of known chronic thromboembolic disease ^[38]A1c.

In pregnancy, the recommended sequence is chest radiography first; if the chest radiograph is normal, lung scintigraphy is preferred; if nondiagnostic, CTPA is performed ^[41]A1c. This approach minimizes cumulative radiation exposure to the mother and fetus.

Avoiding Overuse

The Choosing Wisely campaign and multiple guidelines emphasize that imaging is unnecessary when pretest probability is low and D-dimer is negative ^{[40]A1c[43]D5}. The American Society of Hematology recommends that for patients at low (unlikely) VTE risk, D-dimer as the initial test reduces the need for diagnostic imaging ^[42]A1c.

Pearl: In patients with low pretest probability and a negative PERC or a negative age-adjusted D-dimer, CTPA is unnecessary and should be avoided, this strategy reduces imaging by approximately 10% without missing clinically important pulmonary embolism ^[45]A1b.

Having established the clinical indications for CTPA, the choice among imaging modalities depends on diagnostic accuracy, patient factors, and resource availability. CTPA remains the reference standard for acute PE, but alternative tests serve specific roles when contrast is contraindicated, radiation exposure is a concern, or bedside assessment is needed.

Ventilation-Perfusion Scintigraphy

V/Q scintigraphy has been the traditional alternative for patients with contraindications to iodinated contrast. A meta-analysis of nine studies reported a pooled sensitivity of 0.81 (95% CI 0.76-0.85) and specificity of 0.84, with an AUC of 0.87 ^[31]A1a. However, the proportion of non-diagnostic tests is high: 34.7% ^[31]A1a. By comparison, CTPA has a pooled sensitivity of 0.94 (95% CI 0.89-0.97) and specificity of 0.98 (95% CI 0.97-0.99) ^[18]B2a. When a high-probability V/Q scan is considered positive and low/non-diagnostic/normal scans negative, specificity rises to match CTPA, but sensitivity drops ^[18]B2a. V/Q scanning avoids nephrotoxic contrast and delivers lower radiation to the breasts, making it preferred in young women and pregnant patients (see Special Populations).

Point-of-Care Ultrasound

Multi-organ POCUS (cardiac, lung, and venous ultrasound) has emerged as a rapid bedside tool in critically ill patients. A meta-analysis of four studies (594 patients) found a pooled sensitivity of 0.90 (95% CI 0.85-0.94) and specificity of 0.69 (95% CI 0.42-0.87), with a diagnostic odds ratio of 25.3 ^[30]A1a. The high sensitivity suggests POCUS can rule out PE when negative, but the low specificity means a positive result requires confirmation with CTPA. Lung ultrasound alone, without multi-organ assessment, has a pooled sensitivity of 87.0% (95% CI 79.5-92.0%) and specificity of 81.8% (95% CI 71.0-89.3%) ^[51]A1a. POCUS is most useful in unstable patients where CTPA cannot be performed immediately.

Lower Limb Compression Ultrasonography

Proximal CUS for deep vein thrombosis has low sensitivity for PE: 41% with high specificity 96% (95% CI 94-98%) ^[47]A1a. A positive proximal CUS confirms PE (positive likelihood ratio 11.9), but a negative result cannot exclude it (negative likelihood ratio 0.6) ^[47]A1a. Whole-leg CUS (including calf veins) has higher sensitivity (79%) but lower specificity (84%), limiting its utility ^[47]A1a. CUS is reserved for patients with contraindications to CTPA when a positive result would change management .

Non-Contrast MR Angiography

NC-MRPA avoids both iodinated contrast and ionizing radiation. A meta-analysis of eight studies reported a pooled sensitivity of 0.88 and specificity of 0.97 (95% CI 0.93-0.98), with an AUC of 0.92 ^[31]A1a. The non-diagnostic rate is only 3.31%, far lower than V/Q ^[31]A1a. NC-MRPA offers superior specificity to V/Q and comparable sensitivity, making it a feasible alternative when contrast is risky ^[31]A1a.

Spectral CT

Spectral CT pulmonary angiography (S-CTPA) provides both anatomic and perfusion information. A meta-analysis of four studies found pooled sensitivity of 94.2% and specificity of 88.5% using V/Q as reference ^[36]A1a. Positive and negative predictive values were 87.8% and 94.5%, respectively ^[36]A1a. Data remain limited to small studies with methodological issues, and evidence is strongest for chronic PE/CTEPH rather than acute PE ^[36]A1a.

Prognostic Information from CTPA

Beyond diagnosis, CTPA provides prognostic data. Right ventricle dilation (right-to-left ventricle diameter ratio >1.0) on CTPA is associated with increased 30-day mortality in all-comers with PE (OR 2.08, 95%) and in haemodynamically stable patients (OR 1.64, 95%) ^[49]A1a. Central embolus localization also predicts 30-day mortality (OR 2.24, 95%) ^[50]A1a. This prognostic information is not available from V/Q or POCUS, adding value to CTPA in risk stratification.

Summary of Diagnostic Accuracy

In patients with prior VTE, the standard algorithm (clinical prediction rule, D-dimer, CTPA) remains safe but less efficient: only 15% can avoid CTPA, and the three-month VTE incidence after negative workup is 1.4% ^[48]A1a. For chronic thromboembolic pulmonary hypertension , CTPA has excellent performance with sensitivity 0.98 and specificity 0.99 when read by expert radiologists ^[46]A1a.

Pearl: When choosing an alternative to CTPA, consider that NC-MRPA offers the best balance of sensitivity (0.88) and specificity (0.97) with a low non-diagnostic rate (3.3%), while V/Q scintigraphy has a 34.7% non-diagnostic rate that often necessitates additional imaging ^[31]A1a.

While CTPA remains the first-line test for suspected PE, its application in specific populations requires tailored protocols to balance diagnostic accuracy with safety. The following subsections outline evidence-based modifications for pediatrics, pregnancy, elderly, and immunocompromised patients, drawing on recent dose-reduction and contrast-sparing studies.

Pediatrics

Children present unique challenges: lower body weight, higher radiosensitivity of developing tissues, and often the need for sedation or anesthesia. The ultra-low dose (ULD) CTPA technique using deep learning reconstruction (DLR) is particularly applicable. In a two-center trial, ULD-DLR images achieved subjective and objective image quality superior to routine-dose hybrid iterative reconstruction (HIR) and reduced effective dose by 74% ^[19]A1b. For pediatric patients, a protocol using 70 kVp with automatic tube current modulation and iterative reconstruction maintains diagnostic quality while cutting radiation dose significantly ^[34]A1b. Contrast volume can also be reduced: as low as 15 mL (via multiphase injection) or 20 mL (diluted with saline) still yields diagnostic pulmonary artery attenuation in low-energy virtual monoenergetic images on spectral CT ^{[9]A1b[14]A1b}. Sedation protocols should follow institutional guidelines; the scan itself is rapid (typically <10 seconds) and may be performed without sedation in cooperative older children. The differential diagnosis of acute dyspnea in children includes asthma, foreign body, and congenital heart disease, so CTPA should be reserved for cases with moderate-to-high pre-test probability after D-dimer or other screening where appropriate.

Pregnancy

Pregnancy raises concerns about fetal radiation exposure and contrast agent safety. Although the fetal dose from a standard CTPA is low (typically <1 mGy), the stochastic risk of childhood cancer, though small, warrants minimization. Low-dose protocols using 70 kVp and iterative reconstruction reduce fetal dose further while preserving image quality ^[34]A1b. Alternatively, lung scintigraphy (V/Q scan) delivers a comparable or lower fetal dose and may be preferred in patients with normal chest radiograph and no history of asthma or chronic lung disease, if available. Regarding contrast, iodinated contrast crosses the placenta but is not known to be teratogenic; however, it can cause transient neonatal hypothyroidism, so thyroid function should be checked in the newborn. The use of low-contrast volume protocols (e.g., 20-30 mL of contrast agent) is feasible with spectral CT or DLR-boost, maintaining diagnostic attenuation and perfusion defect detection ^{[9]A1b[10]A1b}. The mother should be well hydrated before and after the study to reduce contrast-induced nephropathy risk. In pregnant patients with suspected PE, CTPA should not be withheld if the diagnosis is essential, as untreated PE carries high maternal mortality.

Elderly

Elderly patients (age ≥65 years) often have renal impairment, reduced cardiac output, and atypical presentations of PE (e.g., syncope, delirium) that lower pre-test suspicion. The prevalence of contrast-induced nephropathy (CIN) is higher in this group due to age-related decline in glomerular filtration rate and comorbidities such as diabetes and hypertension . To mitigate CIN risk, use the lowest feasible contrast volume: 15-30 mL of iodinated contrast, combined with DLR or iterative reconstruction, provides diagnostic image quality ^{[9]A1b[14]A1b[10]A1b}. Pre-hydration with normal saline (e.g., 1 mL/kg/h for 12 hours before and after) is recommended in patients with eGFR <45 mL/min/1.73 m², though the benefit is modest. The radiation dose can be reduced with low-kVp protocols (70-80 kVp) without compromising image quality ^[34]A1b. In elderly patients with poor breath-hold capacity, the scan time should be minimized; multiphase CTPA with temporal registration may allow shorter breath-hold or even free-breathing acquisition ^[14]A1b. The diagnostic yield of CTPA in mechanically ventilated elderly ICU patients is similar to the general emergency department population (21% in one study), but elevated D-dimer and high clinical suspicion remain the best predictors ^[53]B3b.

Immunocompromised

Immunocompromised patients (e.g., post-transplant, chemotherapy, HIV) are at increased risk for both PE and alternative diagnoses such as opportunistic infections, drug-induced pneumonitis, or pulmonary edema. CTPA is valuable not only for PE diagnosis but also for detecting incidental findings: in postoperative screening, 12.7% of patients had incidental abnormalities, including 0.8% with malignant tumors or findings requiring surgery ^[54]B3b. In immunocompromised patients, the threshold for CTPA should be lower because false-negative D-dimer can occur with chronic inflammation, and clinical signs may be masked by immunosuppression. The contrast dose should be minimized to reduce renal burden, using 20-30 mL protocols with DLR-boost or spectral imaging ^{[9]A1b[10]A1b}. Radiation dose reduction is also important given the potential for repeated imaging; ULD-DLR can reduce effective dose by 74% with maintained diagnostic accuracy ^[19]A1b. Radiologists should carefully evaluate for extrapulmonary findings (e.g., nodules, consolidation, pleural effusion) that may alter management .

Pearl: In all special populations, the combination of low-kVp (70-80 kVp) and low-contrast volume (20-30 mL) with deep learning reconstruction or iterative reconstruction maintains diagnostic image quality while reducing radiation dose and contrast-induced nephropathy risk, making CTPA safer and more widely applicable.

Having considered the unique challenges in special populations, attention now turns to the technical and interpretative pitfalls that can compromise CTPA accuracy, along with the complications that affect patient safety. Understanding these pitfalls is essential to mitigate error, improve diagnostic quality, and optimize outcomes ^[16]A1c.

Technical Pitfalls

Poor pulmonary arterial opacification is the most common technical limitation. It arises from suboptimal contrast bolus timing, low cardiac output, or inadequate contrast volume and injection rate ^[16]A1c. Motion artifact from respiration or patient movement degrades image quality, particularly in segmental and subsegmental arteries ^{[16]A1c[59]D5}. Streak artifact from dense contrast in the superior vena cava (SVC) can simulate emboli in the right upper lobe pulmonary artery, a classic false-positive trap ^{[16]A1c[59]D5}. Beam-hardening artifacts from the shoulders, chest wall, or central venous catheters further obscure vessel detail ^[16]A1c. Dual-energy CT with iodine subtraction maps can mitigate some of these artifacts, but atelectasis, fissures, and beam-hardening remain sources of misinterpretation even with this technique, accounting for 48.7% of false results in one study ^[21]B2b.

Interpretative Pitfalls

Mimics of acute PE are numerous. Normal anatomic structures (e.g., pulmonary veins, lymph nodes, bronchial arteries) can be mistaken for filling defects when they cross or abut pulmonary arteries ^[16]A1c. Non-vascular diseases such as peribronchial cuffing, mucus plugs, or tumor may simulate intraluminal thrombus ^[59]D5.

Distinguishing acute from chronic PE is critical for management . Chronic thromboembolic disease manifests as web-like filling defects, complete occlusions with tapered ends, calcified thrombus, and evidence of pulmonary hypertension (e.g., right ventricular hypertrophy, dilated pulmonary artery) ^[26]D5[59]D5. Acute PE typically presents as a central filling defect with vessel expansion, whereas chronic thrombi are eccentric, adherent, and often associated with recanalization ^[26]D5. Failure to recognize chronic changes can lead to inappropriate thrombolysis or missed diagnosis of chronic thromboembolic pulmonary hypertension (CTEPH) ^[26]D5.

Subsegmental emboli are frequently overcalled. Small filling defects at the segmental or subsegmental level may be artifact, and the clinical significance of isolated subsegmental PE remains debated ^[16]A1c. Correlation with clinical probability and ventilation-perfusion scan findings is advised ^[16]A1c.

Contrast-Induced Nephropathy

Contrast-induced nephropathy (CIN) is a frequent and serious complication. In a prospective emergency department cohort, CIN occurred in 14% (95% CI 10-20%) of patients after CTPA, and its development was associated with a markedly increased risk of severe renal failure or death within 45 days (relative risk = 36, 95% CI 3-384) ^[57]B2b. A randomized trial of prophylaxis strategies reported an overall CIN rate of 23.7% (61/257), with no significant difference between N-acetylcysteine plus normal saline (23.5%), sodium bicarbonate plus normal saline (21.2%), and normal saline alone (26.4%) ^[56]A1b. Rates of moderate or severe renal failure were also similar across groups (9.4%, 10.6%, and 17.2%, respectively) ^[56]A1b. These findings underscore that CIN is at least as common as the diagnosis of PE itself in many populations, and that standard prophylaxis regimens do not eliminate the risk. Clinicians should assess baseline renal function, consider alternative imaging (e.g., ventilation-perfusion scan ), and weigh the risk of CIN against the benefit of diagnosis ^[57]B2b.

Other Complications

Allergic-like contrast reactions (urticaria, bronchospasm, anaphylaxis ) occur with typical rates (0.2-0.7% for nonionic contrast) and require prompt recognition and treatment ^[16]A1c. Contrast extravasation at the injection site can cause compartment syndrome; use of a power injector with a secure IV line is essential ^[16]A1c.

Radiation exposure from CTPA is a concern, especially in younger patients and those requiring repeated studies. Effective doses range from 3-10 mSv, with higher doses in larger patients due to increased tube current ^[16]A1c. The risk of radiation-induced cancer is small but real, particularly in children ^[58]D5. Dose-reduction strategies such as iterative reconstruction and tube current modulation should be employed ^[16]A1c.

Pearl: Always evaluate the right upper lobe pulmonary artery on multiple planes and compare with the contralateral side to avoid mistaking streak artifact from SVC contrast for a filling defect; if in doubt, review the non-contrast localizer or consider dual-energy iodine maps ^{[16]A1c[21]B2b}.

Building on the recognition of CTPA's pitfalls and complications, major professional societies have issued evidence-based guidelines to standardize its appropriate use, emphasizing pretest probability assessment, judicious D-dimer testing, and selective imaging.

Core Principles Across Guidelines

All major guidelines converge on a stepwise diagnostic strategy: estimate pretest probability using a validated clinical prediction rule (Wells criteria, revised Geneva score), then use D-dimer testing to triage low- or intermediate-probability patients. The American College of Physicians (ACP) recommends not obtaining D-dimer or imaging in patients with low pretest probability who meet all Pulmonary Embolism Rule-Out Criteria (PERC) ^[40]A1c. For intermediate-probability patients, a high-sensitivity D-dimer is the initial test; imaging is warranted only if D-dimer is elevated ^[40]A1c. In patients older than 50 years, an age-adjusted D-dimer threshold (age × 10 ng/mL) should replace the generic 500 ng/mL cutoff to improve specificity ^[40]A1c. The American Thoracic Society/American College of Chest Physicians Choosing Wisely statement reinforces that chest CT angiography should not be performed in patients with low clinical probability and negative high-sensitivity D-dimer ^[43]D5.

CTPA is reserved for patients with high pretest probability or those with intermediate probability and a positive D-dimer ^[40]A1c. The Canadian Society of Thoracic Radiology/CAR guidelines further stress that CTPA image quality must be optimized through appropriate contrast timing, rate, and volume, and that dual-energy CT may be considered when routine CTPA is problematic (e.g., impaired renal function, altered cardiac anatomy) ^{[16]A1c[39]A1c}.

Pregnancy-Specific Recommendations

The ATS/STR pregnancy guideline recommends chest radiography as the first radiation-associated procedure. If the chest radiograph is normal, lung scintigraphy (V/Q scan) is preferred over CTPA to minimize maternal breast radiation; CTPA is reserved for nondiagnostic V/Q results or when V/Q is unavailable ^[41]A1c. This contrasts with the general population where CTPA is first-line, reflecting the higher radiation sensitivity of breast tissue in pregnancy.

Cost-Effectiveness Considerations

A systematic review informing the American Society of Hematology guideline found that strategies incorporating D-dimer before imaging are cost-effective for PE diagnosis. The cost-effectiveness of CTPA versus V/Q scan was inconclusive, but both are cost-effective when performed after D-dimer or ultrasound ^[60]B2a.

Controversies and Guideline Disagreement

Clinical Prediction Tools and Patient Resources

Validated tools include the Wells criteria (dichotomized as PE unlikely ≤4 points vs. PE likely >4 points), revised Geneva score, and PERC rule (for low pretest probability). The age-adjusted D-dimer (age × 10 ng/mL for patients >50 years) is endorsed by ACP ^[40]A1c. Patient information resources are available through the Choosing Wisely campaign (www.choosingwisely.org) and the American Thoracic Society patient education series.

Pearl: In patients >50 years with intermediate pretest probability, using an age-adjusted D-dimer threshold (age × 10 ng/mL) reduces false positives and unnecessary CTPA by approximately 15-20% compared to the fixed 500 ng/mL cutoff ^[40]A1c.

Building on these guideline recommendations, several emerging technologies promise to further refine CTPA's diagnostic performance and expand its clinical utility. Advances in dual-energy CT, artificial intelligence (AI), and deep learning reconstruction are poised to address current limitations in radiation exposure, contrast dose, and interpretive accuracy.

Dual-Energy CT and Spectral Imaging

Dual-energy CT (DECT) enables material decomposition and iodine mapping, providing both anatomic and functional information. A meta-analysis of nine studies (751 participants) evaluating DECT for detecting chronic thromboembolic pulmonary hypertension (CTEPH) reported a pooled sensitivity of 87.5% and specificity of 91.2%, with an AUC of 0.95 ^[62]A1a. DECT-derived perfused blood volume (PBV) and iodine density (ID) can help differentiate acute from chronic thromboembolic disease and assess regional perfusion defects, though protocol standardization remains a challenge ^[62]A1a. Spectral CT systems (e.g., dual-layer detector, rapid kVp switching) further extend these capabilities, allowing retrospective multi-energy analysis without altering acquisition protocols.

Artificial Intelligence and Deep Learning

AI-based detection of PE on CTPA has been extensively studied. A meta-analysis of 24 studies (22,984 patients) found pooled estimates of AUROC 0.895, sensitivity 0.894 (0.856-0.923), and specificity 0.871 (0.831-0.903) ^[29]A1a. U-Net architectures demonstrated higher sensitivity (0.899 vs. 0.893), while convolutional neural networks (CNNs) achieved higher specificity (0.926 vs. 0.900); both differences were statistically significant ^[29]A1a. A separate systematic review of 12 studies (341,112 images) identified CNNs as the most common AI modality, with models such as EmbNet achieving per-scan sensitivity of 88-90.9% and reducing false positives to 0.45 per scan ^[63]D5.

When combined with deep learning reconstruction (DLR), AI assistance markedly improves radiologist performance. In a prospective two-center study of ultra-low-dose CTPA, AI-assisted reading increased sensitivity from 79.8% to 91.7% and specificity from 95.5% to 99.2% for ultra-low-dose hybrid iterative reconstruction (HIR) images, and from 90.5% to 96.4% (sensitivity) and 95.8% to 100.0% (specificity) for DLR images ^[19]A1b. Interpretation time decreased by 19.7% for HIR and 15.6% for DLR scans ^[19]A1b.

Ultra-Low Dose Protocols and Deep Learning Reconstruction

DLR enables substantial radiation and contrast dose reduction while preserving or improving image quality. In a two-center trial, a dual-low-dose protocol (reduced radiation and contrast) combined with DLR and contrast-enhancement boost (CE-boost) achieved the highest signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) among all reconstruction methods, with an AUC of 0.986 for PE detection ^[10]A1b. The effective dose in ultra-low-dose protocols can be reduced by 74% compared with routine-dose CTPA without compromising diagnostic accuracy ^[19]A1b.

Emerging Therapeutic Integration

CTPA is increasingly used to guide patient selection for advanced endovascular therapies. The ongoing STORM-PE trial (NCT05684796) randomizes patients with intermediate-high-risk PE to anticoagulation alone or anticoagulation plus computer-assisted vacuum thrombectomy (CAVT), using CTPA-measured right ventricular to left ventricular (RV/LV) ratio at 48 hours as the primary outcome ^[61]D5. Such trials underscore CTPA's evolving role not only as a diagnostic tool but also as a quantitative biomarker for treatment response.

Pearl: AI-assisted CTPA interpretation, particularly when combined with deep learning reconstruction, can achieve near-perfect accuracy at ultra-low radiation doses, but prospective validation across diverse populations and standardized reporting metrics are needed before routine clinical adoption ^{[29]A1a[63]D5}.