Impact of Iterative Bilateral Filtering on the Noise Power Spectrum of Computed Tomography Images

A bilateral filter is a non-linear denoising algorithm that can reduce noise while preserving the edges. This study explores characteristics of in changing and texture within computed tomography (CT) images an iterative implementation. We collected homogeneous Neusoft phantom scanned with tube currents 77, 154, 231 mAs. The for each current were filtered five times configuration sigma space (σd) = 2 pixels, intensity (σr) level, kernel 5 × pixels. To observe iteration, power spectrum (NPS) was obtained slices dataset averaged to generate stable curve. modulation-transfer function (MTF) also measured from original images. Tests on anthropomorphic image carried out their impact clinical scenarios. Noise measurements visual observations edge sharpness performed this image. Our results showed effective suppressing at high frequencies, which confirmed by sloping NPS curve different currents. peak frequency shifted about 0.2 0.1 mm−1 all currents, magnitude reduced more than 50% compared spatial resolution does not change number iterations filter, constant values MTF50 MTF10. test show similar pattern, up 60% object edges remaining sharp.