Appendicolith detection in dual-energy CT of adult acute appendicitis: comparing portovenous phase and virtual noncontrast with true noncontrast images

OBJECTIVES: Appendicoliths are associated with failed nonoperative management in acute appendicitis and are used to exclude patients from this treatment. This study evaluated whether portovenous phase (PVP) and virtual noncontrast (VNC) images from rapid-kVP-switching dual-energy CT (rsDECT), alone or combined, can reliably detect appendicoliths using true noncontrast (TNC) images as the reference. Additional aims included identifying CT features of overlooked appendicoliths and those linked to complicated appendicitis. METHODS: Consecutive adults with pathologically confirmed appendicitis who underwent preoperative rsDECT and appendectomy were retrospectively included. Two radiologists independently assessed PVP, VNC, PVP + VNC, and TNC images for appendicolith presence and number, with a third resolving discrepancies. Presence was classified as present or absent, and number into 1, 2 and > 2. Agreement with TNC was assessed using kappa statistics. Logistic regression identified predictors of overlooked appendicoliths and features associated with complicated appendicitis. RESULTS: Among 203 patients; 71 (35%) had appendicoliths. PVP, VNC, and PVP + VNC showed substantial-to-almost-perfect agreement with TNC for detection (kappa = 0.805, 0.793, and 0.817, respectively; all p < 0.001) and substantial agreement for numbering (weighted kappa = 0.734, 0.706, and 0.734, respectively; all p < 0.001). Overlooked appendicoliths had smaller perimeters (OR 12.303; p = 0.003) and lower attenuation (OR 10.456; p = 0.004). In cases with appendicoliths, larger minimum length predicted complicated appendicitis (OR 16.756; p = 0.013). CONCLUSIONS: PVP and VNC from rsDECT sufficiently detected and numbered appendicoliths. Small, low attenuation appendicoliths were easily overlooked, while larger ones were linked to complications.