Image-quality assessment method for digital phase-contrast imaging based on two-dimensional power spectral analysis.

With use of the phase shift of X-rays that occurs when they pass through an object, phase-contrast imaging (herein referred to as "phase imaging") can produce images different from those of conventional contact imaging (herein referred to as "conventional imaging"). For this reason, assessment of the image quality based on noise-equivalent quanta (NEQ) and detective quantum efficiency (DQE) which does not include object-based information may not be appropriate for comparison of image quality between phase and conventional images. As an alternative method, we conceived a new image-quality assessment method with images that contain information about an object. First, we constructed images with an object and without an object under the same imaging parameters; then, we obtained two-dimensional power spectra by Fourier transform of those images. Second, we calculated the radial direction distribution function with the power spectra, and the distribution of signal intensity, which we defined as a signal intensity distribution function (SIDF). In this way, differences in image quality were evaluated relatively based on the SIDF of the imaged object. In our study, we first confirmed that phase-imaging evaluation was not appropriate by comparing NEQ and DQE of conventional, magnification, and phase imaging. Further, comparing the image quality of projected plant seeds by employing conventional, magnification, and phase imaging, we found that the phase-imaging method provided a higher image quality regarding edge sharpness than did conventional and magnification imaging. Therefore, based on these results, our image assessment method is considered useful for evaluation of images which include object-based information.