Feasibility of electron density map derived from unenhanced dual-energy CT for detecting peritonsillar abscess

Purpose To evaluate the feasibility of ED map derived from unenhanced DECT for detecting peritonsillar abscess. Methods This retrospective study included patients with clinically suspected peritonsillar abscess who underwent DECT. Two independent radiologists evaluated the presence of peritonsillar abscess on unenhanced conventional 120-kVp images (CT _conv ), 70- and 40-keV virtual monochromatic images (VMIs), ED map, and contrast-enhanced CT (CECT), using a 4-point diagnostic confidence, scored as follows: 4, highly likely; 3, likely; 2, unlikely; and 1, highly unlikely. Scores of 3 and 4 were considered positive for abscess. The final diagnosis of peritonsillar abscess was based on pus drainage. The sensitivity, specificity, and diagnostic accuracy were calculated for each image, and diagnostic accuracies were compared among image types by McNemar test. In confirmed abscess cases, the contrast-to-noise ratio (CNR) of abscess lesions was calculated. The diagnostic confidence score and CNR were then compared between unenhanced CT _conv and DECT images using the Wilcoxon signed-rank test and paired t-test, respectively. Results Totally, 89 patients (mean age, 39.3 ± 19.4 years) were evaluated. The accuracy of ED maps (75.3%) and 40-keV VMI (73.0%) was significantly higher than that of CT _conv (65.2%; p = .035 and .039, respectively). Both diagnostic confidence score and CNR were significantly higher on ED maps than CT _conv (diagnostic confidence score: 3.49 ± 0.83 vs 2.96 ± 0.93; CNR: 3.42 ± 3.19 vs 0.96 ± 0.78; both p < .001). Conclusion Unenhanced DECT, especially ED map, may enable efficient detection of peritonsillar abscess without contrast agents.