Premium
Prototype system for enhancement of frontal chest radiographs using eigenimage processing
Author(s) -
Butler A,
Bones P,
Hurrell M
Publication year - 2008
Publication title -
journal of medical imaging and radiation oncology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.31
H-Index - 43
eISSN - 1754-9485
pISSN - 1754-9477
DOI - 10.1111/j.1440-1673.2008.01954.x
Subject(s) - medicine , radiography , artificial intelligence , principal component analysis , variance (accounting) , radiology , pattern recognition (psychology) , computer vision , computer science , accounting , business
Summary A prototype system is described for enhancement of radiographic images in the eigen domain. The images chosen to enhance are frontal chest radiographs. This class of images has been chosen because it is both a clinically important examination and an example of the high‐resolution images used within radiology. The enhancement method is based on principal components analysis, a multivariate statistical technique first used within image processing for face recognition. The method requires a training set of normal images to identify normal patterns of variance. The enhancement process then removes these normal patterns of variance, often increasing the relative intensity of pathologies. Enhanced images presented in this paper include a range of common pathologies found on chest radiographs. Details of implementation, computing expense and possible applications within radiology are discussed.