1. Introduction

The CT pulmonary angiogram (CTPA) is the undisputed first-line imaging modality for the diagnosis of acute pulmonary embolism (PE) in the emergency and inpatient setting. Since its widespread clinical adoption in the late 1990s, CTPA has displaced conventional catheter-based pulmonary angiography as the reference standard, offering superior availability, lower invasiveness, shorter examination time, and the added diagnostic yield of evaluating pulmonary parenchyma, mediastinum, and cardiac chambers simultaneously.^[1] In institutions operating modern 64-slice or wider detector arrays, a complete CTPA acquisition can be performed in under three seconds of breath-hold — a critical advantage in dyspnoeic or haemodynamically compromised patients.

Despite its clinical primacy, the diagnostic accuracy of CTPA is critically dependent on protocol execution. A sub-threshold contrast bolus, premature scan trigger, or a single deep inspiratory breath from an anxious patient can catastrophically degrade pulmonary arterial opacification, producing artefactual filling defects indistinguishable from true emboli or, equally dangerously, masking genuine thrombus through pseudoenhancements. This article provides a comprehensive technical, clinical, and interpretive framework for the CT pulmonary angiogram, combining evidence-based protocol parameters with a structured three-tier pitfall analysis for radiographers, radiologists, and non-radiology clinicians.^[3]

Mastery of the CTPA protocol extends far beyond understanding kVp and flow rate. It demands a thorough understanding of cardiopulmonary haemodynamics, the physics of bolus kinetics, the anatomy of the pulmonary vascular tree to subsegmental level, and the pathophysiological mechanisms that generate the diverse range of findings seen on this single acquisition. The following sections deliver precisely that understanding.

2. Pulmonary vascular anatomy & HU values

A thorough working knowledge of pulmonary vascular anatomy is essential for interpreting CTPA at every level of the arterial tree — from the main pulmonary trunk down to subsegmental fourth-order branches — particularly when evaluating for isolated subsegmental PE, which carries its own distinct clinical controversy and management implications.^[4]

2a. Gross anatomy of the pulmonary arterial system

The pulmonary trunk arises from the right ventricle (RV) infundibulum and bifurcates into the right and left main pulmonary arteries at the level of the carina, typically at the T4–T5 vertebral level. The right pulmonary artery passes posterior to the ascending aorta and superior vena cava before dividing into truncus anterior (supplying the upper lobe) and the interlobar artery (supplying middle and lower lobes). The left pulmonary artery arches over the left main bronchus and divides into upper and lower lobe branches.

The right lung is classically divided into ten bronchopulmonary segments and the left into eight to ten, depending on the anatomical variant of the lingula and medial basal segment. Each segment is supplied by a named segmental pulmonary artery that accompanies its corresponding segmental bronchus, running in the centre of each bronchopulmonary unit within the pulmonary bronchovascular bundle. Familiarity with this segmental map is indispensable for correctly localising PE burden and reporting using the standardised nomenclature recommended by the European Society of Radiology (ESR).^[5]

2b. Right ventricular anatomy and strain indicators

The right ventricle is a thin-walled, crescent-shaped structure adapted for low-pressure pulmonary circulation. In the setting of massive or sub-massive PE, sudden elevation of right heart afterload causes RV dilation, septal bowing toward the left ventricle, and a rise in the RV to LV short-axis diameter ratio. An RV/LV ratio exceeding 1.0 on axial CTPA images is the key imaging surrogate for RV strain and independently predicts adverse outcomes, including haemodynamic decompensation and in-hospital mortality.^[6]

2c. Complete HU reference table for CTPA

2d. Pulmonary venous anatomy

The pulmonary veins drain oxygenated blood from the lung parenchyma into the left atrium via four main trunks: superior right, inferior right, superior left, and inferior left. On CTPA performed in the pulmonary arterial phase, the pulmonary veins will appear partially or fully unenhanced, confirming adequate timing. Late-filling pulmonary veins that appear opacified on a well-timed CTPA should prompt careful reassessment of scan timing and consideration of pulmonary venous obstruction.^[8]

3. Scanning technique

Executing a diagnostically optimal CT pulmonary angiogram requires sequential mastery of seven technical parameters. Each step below is protocol-critical: a failure at any single stage can render the entire acquisition non-diagnostic and expose the patient to unnecessary radiation and contrast without clinical return.^[9]

1. Patient preparation and cannula selection: Establish IV access with a minimum 20G cannula in the antecubital fossa (18G preferred for 5.0 mL/s flow rates). Verify renal function, allergy history, and contraindications to iodinated contrast. Prescreening with eGFR and the Wells / Geneva Score should already be documented on the referral. Warm the contrast to 37°C to reduce viscosity and improve bolus consistency. Remove metallic artefact sources (ECG leads if positioned in the scan field).
2. Positioning and centering: Position the patient supine with arms raised above the head to eliminate photon starvation artefact through the upper thorax and axillary regions. Ensure the chest is truly centred at the gantry isocentre — lateral off-centering by even 2–3 cm increases noise substantially due to the non-linear relationship between CT dose and position. Verify that the entire lung apex-to-base is within the planned scan range using the topogram.
3. kVp and dose modulation: Select 100 kVp for standard adult patients (<110 kg). The reduced kVp increases iodine photoelectric attenuation by approximately 25–30% compared to 120 kVp, producing higher arterial contrast at lower radiation dose.^[10] Apply automatic tube current modulation (ATCM) across the projection angles. For obese patients (>110 kg), consider 120 kVp with weight-based dose adjustment. Modern dual-energy CT platforms can run a single 80/150 kVp split acquisition enabling virtual monoenergetic image (VMI) reconstruction at 40–60 keV for superior arterial conspicuity.
4. Bolus tracking setup: Place the ROI monitoring circle centrally within the main pulmonary trunk on the tracking slice (typically at the level of the main PA bifurcation). Set the HU trigger threshold at 130 HU. Configure a 4–6 second fixed scan delay after trigger detection to allow the bolus peak to transit from the monitoring slice to the distal pulmonary arterial tree. Do not use the ascending aorta as a tracking vessel for CTPA — it will trigger scan initiation too early, before the bolus has fully traversed the right heart.
5. Breathing instruction (critical step): Coach the patient carefully before injection. Instruct them to breathe in gently and hold — explicitly avoid the phrase “take a big deep breath in.” A forced maximal inspiration drives a large Valsalva-like surge of unopacified blood from the inferior vena cava into the right heart immediately before or during scan acquisition, diluting the contrast bolus and producing the characteristic Transient Interrupt of Contrast (TIC) artefact. In clinical practice, this remains the single most common cause of non-diagnostic CTPA.^[11]
6. Acquisition parameters and reconstruction: Acquire with pitch 1.3 and rotation time 0.35 seconds, enabling sub-second total acquisition for a standard thorax. Reconstruct axial images at 1.0 mm slice thickness with 0.7 mm increment using both a soft-tissue kernel (B30/B35 equivalent) and a sharper detail kernel for lung windows. Coronal and sagittal reformats at 2.0 mm improve subsegmental branch visualisation. Apply deep learning reconstruction (DLR) where available — studies consistently demonstrate 40–50% noise reduction at equivalent dose, expanding subsegmental diagnostic confidence.^[12]
7. Post-acquisition quality check: Before releasing the patient, the radiographer should measure pulmonary trunk HU on the acquired dataset. Document the attenuation value. If the pulmonary trunk reads below 210 HU, or if TIC artefact is visually identified as a central lucency within the right pulmonary artery corresponding to breath timing, document findings, communicate with the reporting radiologist, and assess the need for repeat acquisition using a modified breathing instruction or reduced flow rate protocol.

3a. Scanner comparison: protocol parameters by detector configuration

3b. Dual-energy and photon-counting CTPA protocols

3c. Deep learning reconstruction (DLR) in CTPA

The integration of deep learning image reconstruction (DLR) into CTPA workflows represents one of the most impactful technical advances of the past five years. DLR algorithms — including GE HealthCare’s TrueFidelity, Siemens Healthineers’ ADMIRE (level 5), Philips IntelliSpace Portal AI Recon, and Canon’s Advanced intelligent Clear-IQ Engine (AiCE) — are trained on large CT datasets to suppress image noise without the resolution-blurring drawback of iterative reconstruction.^[13]

In the CTPA context, DLR consistently achieves 40–55% noise reduction at equivalent dose settings, directly translating to improved visualisation of fourth-order subsegmental pulmonary arterial branches. A 2022 prospective study demonstrated that DLR at a 50% dose reduction setting maintained diagnostic equivalence to full-dose filtered back projection for PE detection down to 3 mm-diameter vessels.^[14] For departments transitioning to low-voltage (80 kVp) CTPA protocols — particularly relevant in younger or smaller patients — DLR makes these dose-reduction strategies clinically viable by restoring acceptable image quality despite the higher intrinsic noise floor of low-kVp acquisitions.

4. Contrast media protocol

The contrast injection protocol for CTPA is among the most precisely specified in all of CT imaging. The physics of pulmonary arterial opacification demand a short, high-concentration bolus that reaches peak enhancement in the pulmonary trunk simultaneously with scan acquisition — a requirement that tolerates only narrow margins of error in either timing or bolus geometry.^[15]

4a. Rationale for high flow rate and concentrated contrast

The pulmonary arterial transit time from cubital vein to pulmonary trunk is typically 8–12 seconds in a patient with normal cardiopulmonary haemodynamics. This window is narrow. Using a 5.0 mL/s flow rate with 60 mL of contrast delivers the entire iodine load into the right heart within approximately 12 seconds, creating a compact, high-concentration bolus that drives peak pulmonary arterial opacification to the target of >300 HU in the trunk and >210 HU in segmental branches. Slower flow rates (e.g., 2–3 mL/s) extend bolus duration, reduce peak iodine concentration, and substantially increase the risk of scan timing mismatch and incomplete opacification.^[16]

4b. The saline chaser — purpose and mechanics

The 100 mL saline chaser serves two critical functions. First, it flushes residual contrast from the peripheral IV line, antecubital vein, and subclavian venous segment into the superior vena cava and right heart, ensuring 100% of the injected iodine mass reaches the pulmonary circulation — effectively adding 5–8% to the functional iodine delivery. Second, and more importantly, it acts as a mechanical piston that compresses and compacts the contrast bolus, reducing bolus spread and maintaining a tighter, higher-peak concentration profile as the iodine transits the right heart chambers.^[17]

The saline chaser also prevents the phenomenon of SVC streak artefact — dense contrast pooled in the brachiocephalic veins and SVC can project high-attenuation artefacts across the right upper lobe pulmonary artery and may be misread as thrombus. By chasing the contrast with saline, SVC clearing occurs before or during the acquisition window.

4c. Modified protocols for compromised patients

5. Radiation dose in CTPA

CTPA is a moderately high-dose CT examination, reflecting the combination of a large scan volume (apex to base of lung), the use of 100 kVp rather than further-reduced kilovoltage, and the rapid rotation time required to freeze respiratory and cardiac motion. Understanding dose metrics and their benchmarks is essential for justification, optimisation, and clinical governance, particularly given the often younger demographic of patients presenting with suspected PE (DVT in pregnancy, young women on oral contraceptive pills).^[18]

5a. Diagnostic reference levels (DRLs) for CTPA

5b. Five proven dose reduction strategies for CTPA

1. Low-kVp acquisition (100 kVp as standard): Moving from 120 kVp to 100 kVp for standard adult patients reduces dose by approximately 30–40% while increasing iodine attenuation by 25–30%, netting a dose reduction without sacrificing diagnostic quality. For patients under 70 kg, 80 kVp combined with DLR can achieve dose reductions exceeding 50%.^[19]

2. Automatic tube current modulation (ATCM): Angular and z-axis ATCM algorithms (Care Dose4D, Auto mA, D-DOM) reduce tube current during beam projections through less attenuating tissue and increase it through denser regions, maintaining consistent image quality at the minimum achievable dose throughout the scan volume.

3. Optimised scan range and direction: CTPA should be acquired cranio-caudally (apex to base) to allow the contrast bolus to transit toward the scan direction, minimising the risk of breathing artefact at the lung bases. Restrict the inferior extent of the scan field to the costophrenic angles only — avoid extending unnecessarily to the upper abdomen unless there is a specific clinical indication documented by the referrer.

4. Deep learning reconstruction (DLR): DLR applied at moderate-to-high strength settings enables a 30–50% reduction in tube current (mAs) while maintaining or improving image quality relative to standard filtered back projection. This is the single most impactful dose reduction technology now available to departments with modern CT scanners.^[20]

5. Avoid non-diagnostic repeats through quality protocols: Every non-diagnostic CTPA that requires repeat acquisition doubles the patient dose and contrast load. Robust patient coaching, post-injection HU spot-check protocols, and standardised injection parameters that eliminate TIC artefact before it occurs represent the most clinically meaningful dose reduction strategy at the system level.

6. Top 10 pathologies detected on CTPA

The CT pulmonary angiogram’s diagnostic scope extends well beyond PE. The simultaneous evaluation of lung parenchyma, mediastinum, pleura, and cardiac chambers means that a single acquisition can diagnose or exclude a wide spectrum of thoracic pathologies. The following ten conditions represent the most clinically consequential and frequently encountered entities on CTPA, each carrying specific imaging characteristics, HU signatures, and protocol-dependent detection variables.^[21]

7. Pitfalls for radiographers — scanning technique errors

The primary scanning pitfall for CTPA, as identified in the protocol matrix, is Transient Interrupt of Contrast (TIC): a forced deep inspiratory breath draws a large bolus of unopacified blood from the inferior vena cava (IVC) into the right atrium and right ventricle, diluting the contrast bolus mid-acquisition and producing a central filling defect that precisely mimics — and is indistinguishable from — a true acute saddle or bilateral main pulmonary artery embolism. This single pitfall is responsible for the majority of non-diagnostic CTPA studies and unnecessary anticoagulation initiated on the basis of a false-positive finding.^[22]

8. Pitfalls for radiologists — interpretation errors

The primary interpretation pitfall for CTPA, as named in the protocol matrix, is the misidentification of hilar lymph nodes or obliquely oriented subsegmental bronchi running adjacent to pulmonary vessels as a partial volume effect mimicking PE. In the peripheral and subsegmental pulmonary arterial tree, bronchi and arteries run in intimate parallel, and at standard 5 mm slice thickness — or even at 1–2 mm — an obliquely oriented bronchus sectioned tangentially alongside its companion vessel can create a crescent of apparent low attenuation that falsely fulfils the imaging criteria for subsegmental pulmonary embolism.^[23]

9. Pitfalls for non-radiology physicians

The CTPA report landing in a clinician’s inbox is the end product of an elaborate technical and interpretive chain — one whose nuances are not always transparent to the requesting physician. The following clinical pitfalls represent the most consequential decision-making errors made by emergency physicians, general internists, and chest physicians in the context of CTPA interpretation and management escalation.

10. Pitfall comparison summary

The following side-by-side framework consolidates the three distinct pitfall domains — scanning, interpretation, and clinical — into an actionable cross-disciplinary reference. Each team member in the CTPA diagnostic chain bears responsibility for a specific set of failure modes, and effective quality improvement requires that each group understands not only their own pitfalls but also those of the adjacent professions.

11. AI & automation in CT pulmonary angiography

The CTPA pathway has emerged as one of the most mature and commercially active domains for artificial intelligence application in diagnostic radiology. The combination of high clinical urgency, standardised acquisition geometry, well-defined imaging targets (intraluminal filling defects), and the binary nature of PE diagnosis versus no-PE creates an ideal environment for supervised deep learning classification models — and the evidence base for clinically deployed AI PE tools now extends across multiple prospective validation cohorts.^[24]

11a. FDA-cleared and CE-marked AI tools for CTPA

The following commercially available AI tools have received regulatory clearance for CTPA PE detection and triage as of 2025–2026. All operate as non-independent decision support tools requiring radiologist oversight and sign-off.

11b. AI-assisted RV strain quantification

Beyond filling defect detection, a second generation of CTPA AI tools focuses on automated cardiac chamber segmentation and RV/LV ratio measurement — the haemodynamic parameter that most directly drives treatment escalation decisions. Manual RV/LV measurement is subject to significant inter-reader variability (ICC 0.71–0.82 in published studies); AI-automated measurement consistently achieves ICC values above 0.94 against expert cardiologist reference measurements, demonstrating the potential for AI to standardise one of the most clinically consequential CTPA metrics.^[25]

11c. Protocol auto-optimisation and quality assurance

A third, emerging frontier applies machine learning to the upstream acquisition process itself: AI-driven post-acquisition quality assessment tools automatically measure pulmonary trunk HU, flag studies with opacification below threshold (signalling TIC artefact or bolus timing failure), and notify the performing radiographer before the patient leaves the suite. Early validation data from two European academic centres demonstrates that implementing AI-based protocol quality monitoring reduces non-diagnostic CTPA rates from approximately 8–12% to below 3% — a clinically and economically significant improvement that reduces repeat contrast exposure and avoids diagnostic delays in the acute PE pathway.^[26]

13. Conclusion

The CT pulmonary angiogram remains the cornerstone investigation of the acute PE diagnostic pathway — but it is a protocol that is uniquely vulnerable to a narrow category of technical failures that can render the entire acquisition either dangerously false-positive or non-diagnostic. Mastery of the seven critical parameters outlined in this article — kVp selection, bolus tracking ROI placement, flow rate and access confirmation, breathing instruction precision, post-acquisition HU verification, reconstruction optimisation, and DLR integration — translates directly into diagnostic confidence, patient safety, and departmental efficiency.

The ten pulmonary pathologies reviewed in this article — from acute saddle PE and RV strain to pulmonary artery sarcoma and tumour embolism — demand that both radiographers and radiologists approach each CTPA with a structured, protocol-driven mentality rather than a purely reflexive one. Recognising that a CTPA is not merely a “rule-out PE” examination, but a comprehensive evaluation of the thoracic vasculature, cardiopulmonary architecture, and lung parenchyma, ensures that the full diagnostic value of the acquisition is captured and communicated to the clinical team.

The three-tier pitfall framework presented here — Transient Interrupt of Contrast for radiographers, subsegmental bronchial partial volume mimicry for radiologists, and failure to act on RV/LV ratio for clinicians — represents the most consequential error clusters in current CTPA practice. Quality improvement initiatives, structured reporting templates, standardised patient coaching scripts, and the emerging generation of AI-powered protocol quality monitoring tools together constitute a compelling suite of interventions for departments seeking to eliminate preventable diagnostic failures in one of the most time-critical examinations in acute radiology.

Departments that invest in standardising their CTPA workflows — from the injector to the reporting workstation — do not merely improve image quality metrics. They protect patients from the twin dangers of missed PE and false-positive anticoagulation or thrombolysis, and they fulfil the core professional mandate of evidence-based, patient-centred radiological practice. Register with SATMED Health to access protocol resources, consumable solutions, and AI integration pathways aligned to every component of the CTPA workflow.