Added value of color-coded virtual non-calcium dual-energy CT in the detection of acute knee fractures in non-radiology inexpert readers

Purpose: To evaluated the added value of dual-energy CT (DECT) virtual non-calcium (VNCa) protocol on conventional CT in the detection of acute knee fractures in non-radiology inexpert readers. Method: One hundred fifty-six patients (mean age, 51.97 years; age range, 17-86 years) with knee trauma, who underwent DECT and MRI within 3 days between April 2017 and October 2018, were retrospectively analyzed. Three readers (intern, 1st-year general surgery resident, 1st-year emergency medicine resident) independently analyzed CT alone and then with the additional color-coded DECT VNCa for fractures. A board-certified radiologist, analyzed CT and MRI series to define the reference standard. Sensitivity, specificity, and AUC were compared between the two reading sessions. Results: Fifty-seven patients had acute fractures and 99 had no fractures. Thirteen of 57 fractures were nondisplaced. The additional use of VNCa images significantly increased the mean AUC (reader 1: 0.813 vs. 0.919; reader 2: 0.842 vs. 0.930; reader 3: 0.837 vs. 0.921; P < 0.05). When only nondisplaced fractures included, the mean AUC was more increased in the combined analysis of CT and DECT VNCa (reader 1: 0.521 vs. 0.916; reader 2: 0.542 vs. 0.926; reader 3: 0.575 vs. 0.926; P < .01). Sensitivity increased by 15 %-20 % in total fracture group and by 69 %-77 % in nondisplaced fracture group over that with CT alone when both CT and DECT VNCa were used. Specificity did not differ significantly. Conclusions: The additional use of color-coded DECT VNCa protocol to conventional CT improved diagnostic performance in detecting acute knee fractures for inexperienced non-radiology readers.