Abstract | ||
---|---|---|
Maximum a posteriori (MAP) method for image reconstruction is subjected to an appropriate selection of the prior distribution. In this paper, we introduce a new approach to estimate the prior distribution using a machine learning scheme based on Relevance Vector Machine (RVM). The RVM prior is applied to the Analyzer -based Imaging (ABI) reconstruction problem. ABI is a technique capable of measuring very subtle X-ray deflection and scatter phenomena when passing through an imaged object producing three parametric images (Absorption, Refraction and ultra -small angle scatter USAXS). The need of a quasi monochromatic and highly collimated beam causes an extremely low photon count in the ABI systems detector, which leads to noisy reconstructions. Here we demonstrate the use of RVM priors to improve the resulting ABI images. |
Year | Venue | Keywords |
---|---|---|
2015 | 2015 IEEE 12th International Symposium on Biomedical Imaging (ISBI) | Analyzer-based phase contrast imaging, phase-sensitive imaging, multiple image radiography, prior estimation, Bayesian reconstruction, machine-learning, relevance vector machine, Gaussian process |
Field | DocType | ISSN |
Iterative reconstruction,Computer vision,Pattern recognition,Computer science,Support vector machine,Parametric statistics,Artificial intelligence,Maximum a posteriori estimation,Relevance vector machine,Prior probability,Detector,Collimated light | Conference | 1945-7928 |
Citations | PageRank | References |
1 | 0.38 | 3 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Oriol Caudevilla | 1 | 1 | 0.38 |
Jovan G. Brankov | 2 | 82 | 12.09 |