Reduction of CT number location dependency using photon-counting detector CT and virtual monoenergetic imaging

Objective.Computed tomography (CT) number accuracy is critical in radiotherapy planning. This study aimed to investigate the potential of photon-counting-detector (PCD) CT to reduce location dependency of CT numbers compared to energy-integrated-detector (EID) CT across imaging modes, tissue types, and patient sizes.Approach.Four inserts of typical tissue types: cortical bone, liver, lung, and adipose were placed inside a body size (40 × 30 cm²) multi-energy CT phantom (Gammex). To simulate a larger patient, tissue-equivalent material (Superflab) was added to increase phantom lateral width to 50 cm. Both phantoms were scanned on a PCD-CT (NAEOTOM Alpha, Siemens), a dual-source EID-CT (Force, Siemens; EID-A), and a dual-layer EID-CT (7500 Spectral, Philips; EID-B). All scans were performed with 120 kV, except dual-energy mode of EID-A (100/Sn150 kV, Sn: Tin filter). Volume CT dose index (CTDI_vol) was matched across scanners. Each insert was scanned in seven locations: iso-center and six peripheral locations. In addition to single-energy images, 70 keV virtual-monoenergetic images (VMIs) were reconstructed. CT numbers were measured using circular regions of interest. Mean CT number, standard deviation (std), coefficient of variation (CV), max-to-min and central-to-peripheral differences were quantified. CT number variation across seven locations was assessed using the Friedman test and Bonferroni corrected pairwise Wilcoxon signed-rank tests.Main results.PCD-VMI demonstrated lower CT number locational variation compared to EID-CTs among most tissue inserts and phantom sizes. For cortical bone in the standard phantom, PCD-VMI achieved std of 3.31 HU, showing 70.3% reduction compared to single-energy PCD (11.15 HU) and lower than single-energy EID-A (9.75 HU), single-energy EID-B (10.93 HU), EID-A VMI (12.65 HU), and EID-B VMI (4.10 HU). PCD-VMI CV remained below 1% and 3% for small and large phantoms, respectively. Increasing phantom size increased CT number variations, with lowest differences for PCD-VMI. PCD-VMI showed significantly lower CT number locational variation than all other scans across most inserts and phantom sizes.Significance.PCD-VMI reduces CT number locational dependency compared to most imaging modes. The greatest improvements were observed for high-Zmaterials and larger patients. PCD-VMIs are promising for radiation therapy applications.