Quantitative spectral CT parameters for predicting prostate cancer

Introduction:The diagnostic performance of spectral computed tomography (CT) for predicting prostate cancer (PC) has not been sufficiently addressed.This study compared the spectral CT parameters of malignant and normal prostate tissue in men with biopsy-proven PC and defined the optimal cut-off values for predicting malignancy.Material and methods: This cross-sectional study included men with biopsy-proven clinically significant prostate cancer (csPC) who underwent prostate magnetic resonance imaging (MRI) and contrast-enhanced abdominopelvic spectral CT.Images were postprocessed.The diagnostic performance of the spectral CT parameters was assessed using receiver operating characteristic (ROC) analysis to determine thresholds for predicting malignant prostate lesions.The optimal threshold was defined using Youden's index.Sensitivity, specificity, accuracy, and area under the curve (AUC) were also determined.Results: We included 105 men with csPC.The mean age was 65.5 6.1 years.Most patients (n = 97; 92.4%) showed PI-RADS lesions 3 on MRI.The optimal cut-off values for differentiating malignant lesions from normal prostate tissue were 137.8 Hounsfield units (HU) for lowenergy virtual monoenergetic images (VMI) at 45 keV, 1.5 mg/mL for iodine concentration density (ICD), and 8.2 for Z-effective value.Accuracy was higher for post-processed spectral maps, including VMI at 45 keV (94.3%),ICD (94.3%), and Z-effective (94.3%), with greater discriminatory power than conventional CT reconstruction (91.7%).Conclusion: Our study showed that quantitative spectral CT parameters have high diagnostic performance with optimal cut-off values for accurately distinguishing malignant prostate lesions from normal prostate tissue in patients with csPC.