Title :
Optimal linear transformation for MRI feature extraction
Author :
Soltanian-Zadeh, Hamid ; Windham, Joe P. ; Peck, Donald J.
Author_Institution :
Dept. of Diagnostic Radiol. & Med. Imaging, Henry Ford Hospital, Detroit, MI, USA
fDate :
12/1/1996 12:00:00 AM
Abstract :
This paper presents development and application of a feature extraction method for magnetic resonance imaging (MRI), without explicit calculation of tissue parameters. A three-dimensional (3-D) feature space representation of the data is generated in which normal tissues are clustered around prespecified target positions and abnormalities are clustered elsewhere. This is accomplished by a linear minimum mean square error transformation of categorical data to target positions. From the 3-D histogram (cluster plot) of the transformed data, clusters are identified and regions of interest (ROI´s) for normal and abnormal tissues are defined. These ROI´s are used to estimate signature (prototype) vectors for each tissue type which in turn are used to segment the MRI scene. The proposed feature space is compared to those generated by tissue-parameter-weighted images, principal component images, and angle images, demonstrating its superiority for feature extraction and scene segmentation. Its relationship with discriminant analysis is discussed. The method and its performance are illustrated using a computer simulation and MRI images of an egg phantom and a human brain
Keywords :
biomedical NMR; brain; feature extraction; image segmentation; medical image processing; 3-D histogram; MRI images; abnormal tissues; angle images; cluster plot; egg phantom; feature extraction method; feature space; human brain; linear minimum mean square error transformation; magnetic resonance imaging; medical diagnostic imaging; normal tissues; principal component images; scene segmentation; tissue-parameter-weighted images; Computer simulation; Feature extraction; Histograms; Image generation; Image segmentation; Layout; Magnetic resonance imaging; Mean square error methods; Prototypes; Vectors;
Journal_Title :
Medical Imaging, IEEE Transactions on