Noise power spectrum (NPS) characteristics on CT imaging: Slice thickness and filter types variations.

Utilization of high-quality computed tomography (CT) imaging holds great potential in assisting clinical radiologists in diagnosing patient's illnesses with great precision and accuracy. One principal parameter of CT imaging quality is noise level, quantified as noise power spectrum (NPS). The study utilized a head water phantom, consisting of homogeneous material with a diameter of 16 cm. In this work, we performed CT imaging quality assessment by analyzing noise power spectrum (NPS) while varying slice thickness and filter types. The NPS was calculated by performing Fourier Transform on five regions of interest (ROI) along the x- and y- axes using IndoQCT. The slice thickness was varied: 0.625, 1.25, 2.5, 5, and 10 mm. It shows that slice thickness 0.625 mm has the finest sharpness but also has highest noise level with NPS peak value of 255.63 HU²mm² at frequency of 0.23 mm^-1. Slice thickness 10 mm has the lowest sharpness but also has lowest noise level of with NPS peak value of 19.79 HU²mm² at frequency of 0.2 mm^-1. Similarly, we varied the filter types, which included smooth, standard, sharp, detail, standard edge, sharp edge, brain smooth, brain standard, and brain sharp. It shows that filter detail has the finest sharpness but also has highest noise level with NPS peak value of 74.6 HU²mm² at frequency of 0.54 mm^-1. Filter brain smooth has the lowest sharpness but also has lowest noise level of with NPS peak value of 15.9 HU²mm² at frequency of 0.26 mm^-1. Results show that there is an inverse proportionality between slice thickness and NPS value at all frequency levels; thicker slice thickness leads to lower NPS value. NPS increases with filters that provide finer image detail. The choice of slice thickness and filter types depends on the object being diagnosed. [ABSTRACT FROM AUTHOR]