Bayesian Estimation of MR Images from Incomplete Raw Data
- 285 Downloads
This work concerns reduction of the MRI scan time through optimal sampling. We derive optimal sample positions from Cramér-Rao theory. These positions are nonuniformly distributed, which hampers Fourier transformation to the image domain. With the aid of Bayesian formalism we estimate an image that satisfies prior knowledge while its inverse Fourier transform is compatible with the acquired samples. The new technique is applied successfully to a real-world MRI scan of a human brain.
KeywordsMagnetic Resonance Scan Time Reduction Optimal Non-Uniform Sampling Bayesian Estimation Image Reconstruction
Unable to display preview. Download preview PDF.
- F. Wehrli, The Origins and Future of Nuclear Magnetic Resonance Imaging, Physics Today, 34–42 (1992).Google Scholar
- G.J. Marseille, M. Fuderer, R. de Beer, A.F. Mehlkopf, D. van Ormondt, Reduction of MRI Scan Time through Non-Uniform Sampling and Edge Distribution Modelling, J. Magn. Reson., B103, 192–196 (1994).Google Scholar
- M. Fuderer, Ringing Artefact Reduction by an Efficient Likelihood Improvement Method, Proc. SPIE, 1137, 84–90 (1989).Google Scholar
- A. van den Bos, Parameter Estimation, Chapter 8 in: Handbook of Measurement Science, Vol. 1, P.H. Sydenham Ed., Wiley, London (1982).Google Scholar
- G.J. Marseille, R. de Beer, A.F. Mehlkopf, D. van Ormondt, Optimization of Sample Positions in Magnetic Resonance Spectroscopy, Proc. ProRISC/IEEE Workshop on Circuits, Systems, and Signal Processing, pp. 233–238, J.P. Veen and M.J. de Ket, Eds., Utrecht, STW (1993).Google Scholar