DECT allows for the creation of virtual monochromatic images (VMIs) between 40 and 140 keV. By plotting HU values from a region-of-interest (ROI) in a spectral attenuation curve, one can calculate the slope (k) between selected energy levels. Additionally, DECT facilitates the calculation of virtual non-contrast (VNC) images and iodine concentration (IC), and utilized material decomposition to estimate effective anatomic number (Z_eff) and electon density (ρ_e)(1).

The uptake of iodine within a tumor reflects tumor hemodynamics and perfusion (specific to the scan phase). DECT-derived iodine maps reflect a combination of intravascular blood volume and extravascular iodine concentration due to capillary permeability(2). High-grade tumors tend to have more neovascularization resulting in more iodine uptake and a steeper slope (k). High-grade tumors also tend to be highly cellular leading to a higher density and thus a higher effective number (Z_eff)(3). It has been postulated that local LN metastases in HNSCC have similar characteristics as the primary tumor. If metastatic LN can be differentiated from non-metastatic LN, then suspicious LN could potentially be categorized as well, negating the need for additional FNA.

Previous studies have indicated significant differences in slope (k) ratio between the primary tumor and metastatic versus non-metastatic LNs. Additionally, IC measured in metastatic LN (excluding necrotic areas) was found to be significantly lower compared to normal and inflammatory nodes  (4–8).

We included DECT scans from patients with primary HNSCC (hypopharyngeal and laryngeal carcinoma) collected from November 2014 to March 2016. Patients underwent scanning with Siemens DECT systems (Somatom Force 80/150kVp or Somatom Flash 80/140kVp). Manual ROIs were drawn using Siemens Syngo.Via CT Dual Energy module in the most enhancing part of the primary tumour and LN, and in the common carotid artery. Normal LNs were defined by normal morphology and size ≤7mm (short axis) in neck level 2 and ≤4mm in other levels, while pathological LNs contained central necrosis. A maximum of 2 pathological and 2 normal LN were measured per patient with a minimum size of 0.1 cm² to guarantee reliable ROI measurements without partial-volume problems from surrounding fat tissue.

As the data distribution was considered significantly different from a normal distribution and because of the small groups, a non-parametric Mann-Whitney U test was used in IBM SPSS Statistics 27. The following DECT-derived parameters between normal and pathological LNs were compared: normalized HU-values at 40-50-60keV), IC, Iodine Density (mg/mL) and Rho/Z (density/effective atomic number). Additionally, a comparison between LN/tumour slope (k^40-100keV) ratio was made.