Premium
Texture analysis of human liver
Author(s) -
Jirák Daniel,
Dezortová Monika,
Taimr Pavel,
Hájek Milan
Publication year - 2002
Publication title -
journal of magnetic resonance imaging
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.563
H-Index - 160
eISSN - 1522-2586
pISSN - 1053-1807
DOI - 10.1002/jmri.10042
Subject(s) - texture (cosmology) , pattern recognition (psychology) , magnetic resonance imaging , artificial intelligence , medicine , statistical analysis , radiology , nuclear medicine , computer science , mathematics , statistics , image (mathematics)
Purpose To classify healthy and diseased livers by texture analysis (TA). Materials and Methods We studied 43 patients divided into four groups according to their clinical stage and 10 controls on a 1.5‐T magnetic resonance (MR) imager, using a T2‐weighted breath‐hold sequence. For the TA, features of the first and second order were used, and several classification procedures were applied for the classification of patients and controls. The choice of features was performed manually and by use of the Fischer coefficient, average correlation coefficients between features and multidimensional discrimination measure. Results All the statistical methods employed were able to differentiate between controls and patients in each group. The classification error varied around 8%. Conclusion We have shown that texture analysis can be successfully used for separating cirrhotic patients and healthy volunteers. Different sets of TA features can be used for a similar classification of patients. J. Magn. Reson. Imaging 2002;15:68–74. © 2002 Wiley‐Liss, Inc.