Spectrography Using a Marchesini Algorithm
H Yu1*, G Wang2, (1) Wake Forest University Health Sciences, Winston-Salem, NC, (2) Virginia Tech, BLACKSBURG, VATH-A-213CD-8 Thursday 8:00:00 AM - 9:55:00 AM Room: 213CD
Purpose: Recently, 'ankylography' was proposed to perform a 3D reconstruction from a single exposure of a monochromatic radiation beam. This approach was very attractive to reveal structures such as complex proteins that can be glimpsed only once before they are destroyed by the high-intensity beam. However, the solution is neither unique nor stable in general. The purpose of this work is to solve that problem from one or two spectral views, which is referred to as 'spectrography'.
Methods: In contrast to 'ankylography', we acquire polychromatic diffractive views with energy-sensitive detectors. For example, a spectrographic system may consist of two spectral imaging chains perpendicular to each other. As a result, sufficiently many Ewald spheres can be simultaneously measured in the Fourier space for 3D reconstruction. To evaluate the performance of spectrography, the hybrid input-output algorithm is implemented for phase retrieval and improved with a subspace saddle-point optimization technique proposed by Marchesini.
Results: A 32X32X32 Shepp-Logan phantom (symmetric with respect to the z axis) is employed to demonstrate the feasibility of the proposed spectrography scheme. Extensive numerical tests are conducted in different configurations and have produced promising results. When the radiation waves span an infinite spectrum, exact results can be reconstructed without any additional constraint. When the spectrum becomes narrower, the solution becomes gradually less stable.
Conclusions: Spectrography overcomes the non-uniqueness and instability of 'ankylography'. This approach has an ultrafast imaging speed for potential applications such as drug designs.