Purpose: The purpose of this study was to define objective and reproducible standards for the quality of CT images as a function of radiation doses and therapeutic validity.
Materials and methods: CT images of the paranasal sinuses of 145 patients (77 female, 68 male; 5-83 years old; mean age, 39.9 years) were classified both subjectively (with a view toward their validity for the planning of functional endoscopic sinus surgery, FESS) and objectively by defining the pixel noise (the standard deviation, STD, of the CT number) in a homogeneous region of interest (ROI), centered on the M. masseter and on the frontal lobe. These measurements were then compared to measurements obtained from scan images of a water-filled Perspex phantom.
Results: The pixel noise measured in the phantom images was nearly identical to the respective values on the M. masseter on the patient images. The use of an edge-enhancing reconstruction algorithm and low-dose protocols, with a pixel noise amounting to 70-90 Hounsfield Units (HU), are indicated for children, chronic sinusitis, and septum deviation, while standard protocols, with a pixel noise of 50-70 HU, are recommended for the preoperative planning and postoperative control of FESS. The pixel noise for high-dose protocols is less than 50 HU; nonetheless, such protocols should generally be avoided.
Conclusion: The pixel noise measured in a water-filled Perspex phantom is indicative of the clinical potential and image quality of paranasal sinus CT scans. Alternatively, the M. masseter can be chosen as an ROI to measure the pixel noise in order to obtain a rough estimate of the image quality or radiation dose class.