Added value of an early delayed-phase acquisition for flow-related artifacts mimicking pulmonary embolism in patients with pulmonary opacities or atelectasis: A retrospective paired, hypothesis-generating study

Background Flow-related artifacts on CT pulmonary angiography (CTPA) can mimic pulmonary embolism (PE), particularly in segments affected by pulmonary opacity or atelectasis. Purpose To assess whether a targeted early delayed-phase acquisition resolves flow-artifact-related indeterminate findings and whether artifact frequency differs by parenchymal pattern. Materials and methods This single-center retrospective paired study included dual-phase CTPA examinations performed between April 1, 2025, and April 1, 2026. Forty-one examinations met clinical eligibility criteria; 8 were excluded — 4 because of inadequate early-phase pulmonary arterial opacification and 4 because of motion artifact precluding reliable segmental assessment — leaving 33 patients for analysis. CTPA was performed on a 128-slice scanner using iodixanol 320 mg I/mL (40 mL at 5 mL/s) followed by a 40-mL saline chaser, with automated bolus tracking (threshold, 60 HU) over the main pulmonary artery. Two radiologists with at least 5 years of experience independently reviewed the examinations at the segment level, and a three-member multidisciplinary panel (the two radiologists plus a pulmonologist) adjudicated the final patient-level PE diagnosis by consensus. In the analytic cohort, suspected flow artifacts constituted arterial-phase indeterminate segments. Paired changes were assessed with the McNemar test; because the primary analysis was performed at the segment level with multiple segments contributed per patient, and cluster-adjusted (generalized estimating equations [GEE] or mixed-effects) modeling could not be reconstructed from the data as recorded, all segment-level results are reported as exploratory and hypothesis-generating rather than confirmatory. Results The cohort included 20 men and 13 women (mean age, 63.03 ± 8.3 years); 6 patients had acute PE. The delayed phase followed the arterial phase by a protocol-defined target of 13 s, with a mean actual delay of 8.81 ± 3.6 s. Among 173 opacity- or atelectasis-associated segments, 98 (56.6%) were flow-artifact-related indeterminate segments. Artifact frequency was highest in ground-glass opacity (25/25, 100%), followed by consolidation (45/51, 88.2%), collapse (28/88, 31.8%), and tumor-associated opacity (0/9, 0%). Delayed imaging clarified 57 of 98 indeterminate segments (58.2%), while the remaining 41 stayed indeterminate and were not counted as positive for embolism; resolution was greatest in ground-glass opacity (84.0%), followed by collapse (53.6%) and consolidation (46.7%) (McNemar χ² = 57, P < .001, unadjusted for within-patient clustering). Conclusion In this small, single-center, cluster-unadjusted paired series, a targeted early delayed phase was associated with resolution of flow-artifact-related indeterminate findings, most notably in ground-glass opacity. Given the exploratory statistical approach, these findings should be regarded as hypothesis-generating; cluster-aware validation in a larger, prospectively structured dataset is required before the diagnostic-yield estimates can be considered confirmatory.