Limited angle geometries applied to small animal X-ray imaging

Abstract

Increasing need of integrated multimodality systems for small animal in vivo imaging motivates the conception of compact system geometries combined to low-dose acquisitions, anatomical information being typically obtained by means of micro-Computerized Tomography (CT). These requirements could be fulfilled by reducing the number of projections acquired with the micro-CT system, typically by reducing the angle between two successive projections on a full angular range. Another possibility is to reduce the angular range of the system, which result in a lack of information and degrade the reconstruction quality with classical FBP techniques. Several works have been done to adress the limited angle problem from an algorithmic point of view. Indeed, this problem occurs quite often, in techniques such as Digital Breast Tomosynthesis. In this work, a micro-CT system acquiring projections on a limited angular range is investigated, using numerical phantoms as well as preclinical mouse projections. As a first attempt to perform reconstruction from the limited angle data sets, we implemented the Adaptative Steepest Descent-Projection Onto Convex Sets (ASD-POCS) algorithm. Reconstructions from preclinical data acquired with an existing micro-CT system as well as from simulated projections allowed us to assess the performances of the algorithm visually and quantitatively. The important number of parameters lead us to conduct a systematic study in order to determine the optimal set of parameters. We present the results obtained and point out the limitations of this approach, along with the formulation of further improvement perspectives.