Share Email Print

Proceedings Paper

Point-based warping with optimized weighting factors of displacement vectors
Author(s): Ranier Pielot; Michael Scholz; Klaus Obermayer; Eckart D. Gundelfinger; Andreas Hess
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

The accurate comparison of inter-individual 3D image brain datasets requires non-affine transformation techniques (warping) to reduce geometric variations. Constrained by the biological prerequisites we use in this study a landmark-based warping method with weighted sums of displacement vectors, which is enhanced by an optimization process. Furthermore, we investigate fast automatic procedures for determining landmarks to improve the practicability of 3D warping. This combined approach was tested on 3D autoradiographs of Gerbil brains. The autoradiographs were obtained after injecting a non-metabolized radioactive glucose derivative into the Gerbil thereby visualizing neuronal activity in the brain. Afterwards the brain was processed with standard autoradiographical methods. The landmark-generator computes corresponding reference points simultaneously within a given number of datasets by Monte-Carlo-techniques. The warping function is a distance weighted exponential function with a landmark- specific weighting factor. These weighting factors are optimized by a computational evolution strategy. The warping quality is quantified by several coefficients (correlation coefficient, overlap-index, and registration error). The described approach combines a highly suitable procedure to automatically detect landmarks in autoradiographical brain images and an enhanced point-based warping technique, optimizing the local weighting factors. This optimization process significantly improves the similarity between the warped and the target dataset.

Paper Details

Date Published: 6 June 2000
PDF: 9 pages
Proc. SPIE 3979, Medical Imaging 2000: Image Processing, (6 June 2000); doi: 10.1117/12.387649
Show Author Affiliations
Ranier Pielot, Leibniz Institute for Neurobiology (Germany)
Michael Scholz, Technische Univ. Berlin (Germany)
Klaus Obermayer, Technische Univ. Berlin (Germany)
Eckart D. Gundelfinger, Leibniz Institute for Neurobiology (Germany)
Andreas Hess, Leibniz Institute for Neurobiology (Germany)

Published in SPIE Proceedings Vol. 3979:
Medical Imaging 2000: Image Processing
Kenneth M. Hanson, Editor(s)

© SPIE. Terms of Use
Back to Top