Lesion conspicuity assessment of prone 40-keV virtual monoenergetic imaging in arterial-phase photon-counting CT for breast cancer: a retrospective comparison with DCE-MRI

Purpose This study investigated whether contrast enhanced whole-body photon-counting detector (PCD)-CT performed for staging in the prone position can provide breast lesion conspicuity comparable to dynamic contrast-enhanced T1-weighted imaging (DCE-T1WI). Methods This single-center, retrospective diagnostic feasibility study included 22 women with surgically confirmed primary breast cancer who underwent preoperative prone PCD-CT and DCE-MRI. Then, 40- and 70-keV virtual monoenergetic image (VMI) from a single arterial-phase PCD-CT acquisition were reconstructed. Three readers independently evaluated 40- and 70-keV VMI and DCE-T1WI. The primary endpoint was lesion conspicuity. The secondary endpoints included, size error, detection rate, background parenchymal enhancement (BPE), subjective noise, and reading time. Results Lesion conspicuity was highest on 40-keV VMI, which outperformed 70-keV VMI and DCE-T1WI. Absolute size error was numerically larger for 40-keV VMI than for DCE-T1WI, but this difference was not statistically significant. Directional error indicated that 40-keV VMI significantly underestimated lesion size relative to DCE-T1WI. Detection rates showed an overall difference among modalities, but pairwise differences were not statistically significant. Further, 40-keV VMI had a higher BPE and a greater perceived noise than 70-keV VMI, and it required a longer reading time than DCE-T1WI. The radiation dose did not differ between VMI energy levels as both were reconstructed from the same acquisition. Conclusion Arterial-phase 40-keV VMIs from prone PCD-CT demonstrated the highest lesion conspicuity and showed better detection capability than 70-keV VMI. In addition, their performance was comparable to or superior to DCE-T1WI. This finding was achieved despite trade-offs in increased noise, BPE and size underestimation. Thus, 40-keV VMI may serve as a useful complementary modality for local breast assessment and may be considered a potential alternative for lesion localization when MRI is not feasible, despite potential underestimation of lesion extent.