Coded aperture phase contrast tomosynthesis

• Main Author: Szafraniec, M. B.
• Published: University College London (University of London) 2013
• Subjects: 610
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.626054

X-ray Phase Contrast (PC) Imaging has the potential to greatly improve radiography. Conventional radiography is based on the imaginary part of the refractive index, whereas PC imaging is based on the unit decrement of the real part, which can be up to 1000 times larger. This is particularly important in soft tissue imaging (e.g. mammography), where the differences in the imaginary part of the refractive index are small. Combining PC with tomosynthesis (TS) would also allow for the differentiation of overlapping tissues, thus leading to improved detection of breast lesions. Coded Aperture X-ray Phase Contrast Imaging (CAXPCi) is a technique which enables PC imaging with conventional lab sources, and which could thus be translated into a clinical environment. In this work the feasibility of a combined CAXPCi/Tomosynthesis approach is demonstrated, both on simulated and experimental data. The results of the simulations of the phase contrast signal in TS geometry are in good agreement with the experimental data. The simulator can be used in the future to understand and predict the phase contrast signal for other 3D implementations of phase contrast imaging. The implementation of the new method in the laboratory conditions is presented and four tomosynthesis reconstruction algorithms are proposed and compared: filtered backprojection (FBP), separable paraboidal surrogates (SPS), conjugate gradient (CG) and penalized conjugate gradient (penalized-CG). The analytical reconstruction (FBP) of the phase contrast data results in noisy images. This can be greatly improved by using iterative statistical methods (SPS, CG and penalized CG), at a cost of a decreased image contrast but with an overall benefit in terms of contrast to noise ratio. CAXPCi Tomosynthesis scans of various samples, including a sample containing two superimposed TORMAM phantoms as well as breast biopsy samples, were carried out to evaluate the method’s performance under realistic mammographic conditions. Strong phase contrast (PC) signal was observed in all reconstructions, and the overlapping structures were effectively separated.