Influence of image resolution property on aliasing error of digital wiener spectrum

Abstract

The noise properties of a radiography system are commonly described by its wiener spectrum (WS). The two-dimensional discrete Fourier transform (2D-DFT) methods is the most commonly used and accepted techniques for measuring the digital WS. The 2D-DFT method has been adopted by the International Electrotechnical Commission (IEC) as a noise-power spectrum. However, the digital WS contains the effects of aliasing error, and this error depends on the presampled modulation transfer function (MTF) of the digital radiography (DR) system. The aim of this work was to show the influence of the aliasing error of the digital WS when the image resolution property was changed. We examined the influence of the aliasing error using simulated noise images. Two types of noise images with same pixel size and different presampled MTFs were generated by using ImageJ (National Institutes of Health: NIH). These images were used to simulate the image resolutions of an indirect/direct flat panel detector (FPD). The theoretical WS of the simulation noise image can be derived from a standard deviation σ of the Gaussian filter and added noise. Simulated noise images were analysed using the 2D-DFT method. The WS values calculated from those simulation images were compared with the theoretical WS values. The WS values in the indirect and direct FPD increased, compared to the theoretical WS values. The average relative differences for frequencies up to the Nyquist frequency were 27.9% and 85.2%, respectively. The results showed that the degree of the influence of the aliasing error of the digital WS depends largely on the presampled MTF of the DR system. Therefore, we should take into account the impact of the aliasing error of the digital WS, in the comparison between DR systems with different presampled MTFs.