Title :
A non-uniformly sampled Markov random field model for MAP reconstruction of magnetoencephalogram images
Author :
Gardiner, Alan H. ; Jeffs, Brian D.
Author_Institution :
Lockheed Martin Red. Syst., USA
Abstract :
The maximum a posteriori (MAP) image reconstruction of magnetoencephalograms (MEG) is investigated. A mathematical framework for vector Markov random field models (MRF) suitable for MEG modeling of brain neuron current dipole activity is developed. A new method for simulating an MRF over a non-uniformly spaced sample grid while approximating an arbitrary desired covariance structure at these samples is also presented. Simulation results validate the effectiveness of this random sampled field model, and clinical MEG evoked response data is processed to demonstrate algorithm performance
Keywords :
Gaussian processes; Markov processes; biomedical measurement; brain models; digital simulation; image reconstruction; image sampling; magnetoencephalography; medical image processing; neurophysiology; random processes; Gauss-Markov random fields; MAP reconstruction; MRF; algorithm performance; brain neuron current dipole activity; clinical MEG evoked response data; covariance structure; data processing; magnetoencephalogram images; maximum a posteriori image reconstruction; nonuniformly sampled Markov random field model; nonuniformly spaced sample grid; random sampled field model; simulation results; Biosensors; Brain modeling; Computational modeling; Image reconstruction; Magnetic fields; Magnetic sensors; Markov random fields; Mathematical model; Neurons; Sensor arrays;
Conference_Titel :
Image Processing, 1998. ICIP 98. Proceedings. 1998 International Conference on
Conference_Location :
Chicago, IL
Print_ISBN :
0-8186-8821-1
DOI :
10.1109/ICIP.1998.727219
