Premium
Automatic plaque characterization employing quantitative and multicontrast MRI
Author(s) -
Sun Binjian,
Giddens Don P.,
Long Robert,
Taylor W. Robert,
Weiss Diana,
Joseph Giji,
Vega David,
Oshinski John N.
Publication year - 2008
Publication title -
magnetic resonance in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.696
H-Index - 225
eISSN - 1522-2594
pISSN - 0740-3194
DOI - 10.1002/mrm.21279
Subject(s) - computer science , characterization (materials science) , artificial intelligence , pattern recognition (psychology) , medicine , nuclear magnetic resonance , materials science , physics , nanotechnology
Multicontrast magnetic resonance imaging (MRI) has shown promise in identifying and characterizing atherosclerotic plaques. One of the limitations of this technique is the lack of a practical automated plaque characterization scheme. In the current study, a prior‐information‐enhanced clustering (PIEC) technique that utilizes both multicontrast MR images and quantitative T 2 maps is proposed to characterize atherosclerotic plaque components automatically. The PIEC algorithm was assessed on computationally simulated images and multicontrast MRI data of coronary arteries. Multicontrast ( T 1 ‐, T 2 ‐, partial T 2 ‐, and proton density‐weighted) MR images were acquired from freshly excised human coronary arteries using a 4.7T small‐animal scanner. The T 2 distribution for each plaque constituent was measured by exponentially fitting the signal from multiple MR images with different TEs and the same TR. The calculated T 2 distributions were used as the a priori information and combined with the Fuzzy C‐Means (FCM)‐based clustering algorithm to characterize plaque constituents. The proposed PIEC technique appears to be a promising algorithm for accurate automated plaque characterization. Magn Reson Med, 2007. © 2007 Wiley‐Liss, Inc.