Boundary-enhanced region-of-interest image reconstruction in propagation-based x-ray phase-contrast tomography

Abstract

Propagation-based phase-contrast tomography is an imaging method that can reconstruct the distribution of the three-dimensional complex-valued refractive index distribution of an object. In many applications, a boundary-enhanced image is sought that reveals the locations of boundaries in the refractive index distribution. In this letter, we demonstrate that within certain regions-of-interest, the three-dimensional Laplacian of the real-valued component of the refractive index distribution can be reconstructed from knowledge of truncated phase-contrast projection data. The proposed reconstruction algorithm is demonstrated by use of experimental measurement data. © 2009 American Institute of Physics. ͓doi:10.1063/1.3254235͔ Propagation-based, or in-line, x-ray phase-contrast imaging methods 1-4 exploit a contrast mechanism based on differences in an object's x-ray refractive index distribution, and therefore permits visualization of object features that possess very similar x-ray absorption properties. Phasecontrast imaging methods can also operate effectively at relatively high x-ray energies, thereby facilitating low-dose imaging. 5 Moreover, phase-contrast tomography ͑PCT͒ methods 6-11 are being actively developed for reconstructing information regarding the three-dimensional ͑3D͒ complexvalued refractive index distribution of an object, which can facilitate a wide range of imaging applications.In certain applications of PCT, only information regarding the locations of interfaces in the refractive index distribution, rather than an accurate estimate of the refractive index distribution itself, is sought after and sufficient to accomplish the diagnostic task at hand. We refer to this variant of PCT as boundary-enhanced PCT, 7,13-15 which is conceptually similar to Lambda tomography methods that have been developed for conventional CT. 16 Because the need for phase-retrieval is avoided, image reconstruction algorithms for boundary-enhanced PCT 12,17 do not possess Fourier space singularities and can be relatively robust to experimental errors and even the effects of wavefield polychromaticity. 15 Data truncation is another important experimental factor that is not accommodated by most conventional tomographic image reconstruction algorithms. In applications of boundary-enhanced PCT, the object will often not fit entirely in the field-of-view of the imaging system and/or the probing x-ray beam is intentionally collimated to minimize radiation exposure to the object, resulting in a set of truncated projection measurements. It has been demonstrated 12 that blurred estimates of 3D Laplacian of the refractive index distribution within a region of interest ͑ROI͒ can be reconstructed from truncated projection data. However, the image blurring is undesirable in applications where high spatial resolution is required.There remains an important need, which we address in this letter, to develop accurate ROI image reconstruction algorithms for boundary-enhanced PCT involving truncated measurement data. In recent years, image ...