Accuracy of liver metastasis detection and characterization: Dual-energy CT versus single-energy CT with deep learning reconstruction

•In this prospective study of 30 participants (141 lesions [124 metastases, 17 benign]) with colorectal liver metastases on abdominal CT using deep learning image reconstruction, perceived image quality of dual-energy CT (DECT)(71/90 total scores were rated as high-quality) was not significantly different from single-energy CT (SECT)(62/90 total scores were rated as high-quality) (OR, 2.01; 95% CI:0.89, 4.57; P=0.093).•Liver lesion observer detection was equivalent between DECT and SECT (140/141 lesions detected) with the same accuracy of 92.9% (95% CI:87.3%, 96.5%; 131/141).•Subsequent use of DECT spectral HU lesion curves for liver lesions initially scored as indeterminate resulted in 34 correct characterizations (31 metastases, 3 benign) and no mischaracterizations.