Abstract
BACKGROUND: During computed tomography pulmonary angiography (CTPA) examination, contrast accumulation in the superior vena cava often generates beam hardening artifacts (BHAs), which can interfere with accurate diagnosis. This study aimed to investigate the diagnostic value of dual-energy computed tomography (DECT) quantitative parameters in differentiating pulmonary embolism (PE) from BHA. METHODS: A total of 68 patients with PE who underwent DE-CTPA were retrospectively included. Quantitative parameters [slope of Hounsfield unit (HU) curves (Slope), CT values on virtual monoenergetic images (VMIs) ranging from 40 to 100 keV (HU(40 keV)-HU(100 keV)), normalized iodine concentration (NIC), and normalized effective atomic number (NEffZ)] on the BHA-induced low-density area of the right upper pulmonary artery (artifact), embolism, and the corresponding normal area of the left upper or right middle pulmonary artery (normal) were measured and calculated. The parameters among the three groups were compared, and the performances of the parameters in differentiating the three conditions were evaluated. RESULTS: Quantitative parameters, including HU(90 keV), HU(100 keV), NIC, NeffZ, and Slope were all highest for normal arteries, followed by artifacts, and lowest for embolism (all P<0.05). To differentiate artifact and embolism, all parameters had areas under the curve (AUCs) higher than 0.80 (0.808-0.963), and HU(100 keV) had the highest AUC of 0.963. The multi-variables, combining HU(100 keV), NIC, and NEffZ, had an AUC of 0.968, comparable to HU(100 keV) (P>0.05). Between normal and artifact, NIC showed the highest AUC (0.865), whereas multi-variables combining HU(100 keV), NIC, and Slope improved the AUC to 0.937 and the model quality increased to 0.90 (P<0.05). For differentiation between normal and embolism, all parameters had AUCs higher than 0.80 (0.849-0.991); HU(90 keV) and HU(100 keV) showed the highest AUCs of 0.99 and 0.991, respectively. After a multivariable analysis combining HU(90 keV), NIC, and Slope, the AUC was increased slightly to 0.995, which was comparable to that of HU(90 keV) and HU(100 keV) (P>0.05). CONCLUSIONS: Quantitative parameters derived from DECT could recognize BHA in the superior vena cava; therein, 90 keV and 100 keV VMIs and their HU measurements would be particularly valuable.