Accession Number : ADP007219

Title :   Automatic Magnetic Resonance Imaging,

Corporate Author : RICE UNIV HOUSTON TX DEPT OF STATISTICS

Personal Author(s) : Ensor, Katherine B. ; Ensor, Joe E. ; Misra, Lalith

Report Date : 1992

Pagination or Media Count : 4

Abstract : Magnetic resonance imaging (MRI) is currently the most sensitive modality for detecting and differentiating pathophysiologic events, Transverse relaxation times (T2) provide quantitative information useful for evaluating a number of diseases (Dumitresco et al, (1986)). In MRI the observed T2 signal is modeled by m(t) = Lambda (Sigma k j =1 Sigma j e - Alpha jt) where the reciprocal of Alpha j the corresponding expected relaxation time. We consider maximum likelihood estimation of the parameters Lambda, Sigma j, Alpha j, = 1,..., k under the assumption that the number of excited protons measured follows a Poisson distribution. A computationally simple method for selecting k, the number of exponential components in the model, is proposed.

Descriptors :   *MAGNETIC RESONANCE, *MAXIMUM LIKELIHOOD ESTIMATION, ABSTRACTS, DISTRIBUTION, MODELS, NUMBERS, PARAMETERS, PROTONS, RELAXATION, RELAXATION TIME, RESONANCE, SIGNALS, TIME, TRANSVERSE.

Subject Categories : Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE