We propose a non-rigid registration algorithm for temporal subtraction of thorax CR-images. The images are deformed using a statistically trained B-spline deformation mesh based on principal component analysis of a training set. Optimization proceeds along the transformation components rather then along the individual spline coefficients, using pattern intensity as the criterion. The algorithm is trained on a set of 30 lung pairs and verified on a set of 46 lung pairs. In 96% of the cases the achieved registration is subjectively rated to be adequate for clinical use. © Springer-Verlag Berlin Heidelberg 2003.
CITATION STYLE
Loeckx, D., Maes, F., Vandermeulen, D., & Suetens, P. (2003). Temporal subtraction of thorax CR images. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2878, 738–745. https://doi.org/10.1007/978-3-540-39899-8_90
Mendeley helps you to discover research relevant for your work.