Browse Prior Art Database

OPTIMIZATION OF COIL DATA COMBINATION/ELIMINATION STRATEGY USING MEAN-FIELD ANNEALING FOR PARALLEL MRI

IP.com Disclosure Number: IPCOM000199444D
Publication Date: 2010-Sep-04
Document File: 6 page(s) / 208K

Publishing Venue

The IP.com Prior Art Database

Abstract

The present invention discloses a computationally efficient optimization algorithm that is derived by an optimization method of “mean-field annealing” (MFA). The Mean-field annealing is based on an analytical approximation to the simulated annealing (SA) algorithm that computes the thermal average of the optimization variables directly for a given temperature. Thus the present invention provides a computationally efficient optimization algorithm to select the best coil combination or elimination strategy with reduced computation time.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 44% of the total text.

Page 1 of 6

RP13491

OPTIMIZATION OF COIL DATA COMBINATION/ELIMINATION STRATEGY USING MEAN-FIELD ANNEALING FOR PARALLEL MRI

BRIEF ABSTRACT

    The present invention discloses a computationally efficient optimization algorithm that is derived by an optimization method of "mean-field annealing" (MFA). The Mean-field annealing is based on an analytical approximation to the simulated annealing (SA) algorithm that computes the thermal average of the optimization variables directly for a given temperature. Thus the present invention provides a computationally efficient optimization algorithm to select the best coil combination or elimination strategy with reduced computation time.

KEYWORDS

    Mean field annealing, parallel MRI, simulated annealing, coil array, re- construction

DETAILED DESCRIPTION

    Parallel MRI (pMRI) consists of several receiver coils to receive MRI signal simultaneously with varying spatial sensitivity. Coil arrays with large number of receive elements allow improved imaging performance and higher signal-to- noise ratio (SNR), but also allow more flexibility in the choice of field of view and/or scan planes. Consequently pMRI has to handle an increased amount of data and higher reconstruction burden which slows down data acquisition without significant benefits as quantified by image signal-to-noise (SNR) ratio. Therefore, it is desirable to introduce data reduction technique. Typically, a class of coil combination and elimination algorithm is used for this purpose. The algorithm selects only linear combinations of receiver channel

1

Page 2 of 6

RP13491

information. The algorithm thereby contributes the most to image SNR and quantify the SNR loss for the reduced receiver channel set. However the algorithm is not very efficient and requires large computation time.

    A conventional method of simulated annealing (SA) addresses the problem of selecting an optimal set of coils for a given region of interest (ROI) that amounts to an acceptable SNR degradation. SA is inspired by the statistical mechanics of systems with a large number of frustrated degrees of freedom like spin glasses. In such systems, the energy landscape is very uneven with many local minima and an algorithm to find the global ground state of the system based on steepest-descent methods invariably get stuck in a local minimum. SA overcomes the problem by allowing moves that increase the energy of the system with a probability that depends on temperature. An annealing schedule is set up to gradually reduce the temperature and allow convergence towards the global minimum of the energy landscape. Convergence properties of the algorithm are studied under fairly general assumptions regarding the nature of the stochastic process describing the sequential evolution of the system if the temperature is lowered. The SA algorithm finds the global minimum with probability one but in an infinitely slow annealing schedule. For more realistic annealing schedules no convergence result proves al...