Coded aperture design for X-ray tomosynthesis
University of Delaware
This study focuses on the coded aperture design and its optimization for X-ray tomosynthesis. In contrast to traditional cone-beam X-ray tomography, coded aperture X-ray tomosynthesis places a coded aperture between the X-ray source and the 3D object, by which certain X-rays will pass or be blocked. By using coded apertures, multiplexed x-ray projections can be made simultaneously instead of sequentially as in conventional computed tomography systems. As a result, it leads to the reduction of radiation exposure to a patient since fewer measurements are taken. In this work, we extend the situation to multiple sources. Each source produces a certain projection pattern. By placing coded aperture in front of each source, multiplexing of coded measurements is achieved. The coded aperture acts as a reference structure which disambiguates these superposed projections. The projections are measured by a 2D detector plane located under the 3D object. By applying compressive sensing theory, the Gradient Projection for Sparse Reconstruction (GPSR) algorithm is then used to recover the original 3D data cube. The thesis investigates the function of coded aperture in X-ray transform tomosynthesis systems. Three constraints are introduced, and conditions for the coded apertures are identified in order to achieve uniform sensing of the object and the detector. Simulations show higher quality of X-ray tomosynthesis image reconstruction when optimized codes are applied. A performance increase of up to 3 dB in the PSNR of the reconstructions is obtained by using optimized coded apertures compared to the use of random codes.