Tomographic Reconstruction Using Global Statistical Priors

Abstract

Recent research in tomographic reconstruction is motivated by the need to efficiently recover detailed anatomy from limited measurements. One of the ways to compensate for the increasingly sparse sets of measurements is to exploit the information from templates, i.e., prior data available in the form of already reconstructed, structurally similar images. Towards this, previous work has exploited using a set of global and patch based dictionary priors. In this paper, we propose a global prior to improve both the speed and quality of tomographic reconstruction within a Compressive Sensing framework.We choose a set of potential representative 2D images referred to as templates, to build an eigenspace; this is subsequently used to guide the iterative reconstruction of a similar slice from sparse acquisition data. Our experiments across a diverse range of datasets show that reconstruction using an appropriate global prior, apart from being faster, gives a much lower reconstruction error when compared to the state of the art.

Tomographic reconstruction to detect evolving structures

“…Having presented the application, we first review the algorithm [20] for a global (unweighted) prior-based reconstruction in Sec. 3.1.…”

“…The latter limitation was relaxed in [18,19] by building dictionary-based priors from multiple templates. However, as reported in [20], dictionary priors are not as accurate and fast as global eigenspace priors. Global eigenspace priors are better able to exploit the similarity of a test volume to a set of templates, by assuming that the new test volume lies within the space spanned by the eigenvectors of the multiple representative templates.…”

“…We first present the eigenspace-cum-CS prior-based reconstruction, which was shown [20] to be better when compared to dictionary-based priors.…”

Learning from past scans: Tomographic reconstruction to detect new structures

“…In all of these methods, it is critical and challenging to choose an optimal representative template. Recent work [9][10][11] relaxed the above limitation by building dictionary based and eigenspace based priors. [11] showed that the eigenspace based prior is better able to accommodate the variation in the match of a test volume to a set of templates.…”

“…Recent work [9][10][11] relaxed the above limitation by building dictionary based and eigenspace based priors. [11] showed that the eigenspace based prior is better able to accommodate the variation in the match of a test volume to a set of templates. The technique assumed the new test volume lies within the space spanned by the eigenvectors of multiple representative templates, effectively capturing the global properties of this set of templates.…”