Diagnostic performance of the proposed DECT NI-RADS criteria for imaging surveillance of post-treatment head and neck squamous cell carcinoma: A preliminary study.

Abstract

BackgroundThe value of dual-energy CT (DECT) in enhancing the diagnostic performance of the NI-RADS criteria has not been investigated.PurposeTo compare the diagnostic performance of the proposed DECT NI-RADS with conventional contrast-enhanced (CECT) NI-RADS.Materials and methodsThis retrospective study included 47 patients with head and neck squamous cell carcinoma (HNSCC) who underwent surgery, chemotherapy, or radiation therapy, followed by post-treatment DECT neck scans. Three circular regions of interest (ROIs) were placed in the primary tumor bed and contralateral normal sternocleidomastoid muscle (SCM) on 120-kVp conventional images, 40-keV images, and iodine concentration (IC) maps. Mean density (HU), spectral HU (SHU) at 40 keV, IC values, and their normalized tumor-to-SCM ratios were calculated. Differences between benign and malignant groups were assessed. Receiver operating characteristic (ROC) analysis identified the parameter with the highest area under the curve (AUC), which was used to modify the CECT NI-RADS into a DECT-based NI-RADS system. Lesions in category 2 exceeding the cutoff were upgraded to category 3, and vice versa. Diagnostic performance was compared between conventional NI-RADS and DECT-based NI-RADS.ResultsMalignant lesions showed significantly higher attenuation, IC, and normalized IC than benign post-treatment changes. Normalized IC had the highest AUC, with a cutoff of 2.769 yielding 86.7% sensitivity, 80.0% specificity, 65.0% PPV, 93.3% NPV, and 82.0% accuracy. The proposed DECT NI-RADS criteria may improve both diagnostic performance and interreader agreement compared with conventional CECT NI-RADS.ConclusionsDECT NI-RADS system, integrating normalized IC ratio, may enhance diagnostic performance and interreader agreement compared with the conventional CECT NI-RADS criteria.