Comparative evaluation of regression methods for 3d-2d image registration

Authors:
Ana Isabel Rodrigues Gouveia;Coert Metz;Luís Freire;Stefan Klein
Affiliations:
CICS-UBI – Health Sciences Research Centre, University of Beira Interior, Covilhã, Portugal,Institute of Biophysics and Biomedical Engineering, University of Lisbon, Lisbon, Portugal;Depts. of Medical Informatics & Radiology, Erasmus MC, Biomedical Imaging Group Rotterdam, Rotterdam, The Netherlands;Escola Superior de Tecnologia da Saúde de Lisboa, Instituto Politécnico de Lisboa, Lisbon, Portugal;Depts. of Medical Informatics & Radiology, Erasmus MC, Biomedical Imaging Group Rotterdam, Rotterdam, The Netherlands
Venue:
ICANN'12 Proceedings of the 22nd international conference on Artificial Neural Networks and Machine Learning - Volume Part II
Year:
2012

Citing 10
Cited 0

The nature of statistical learning theory

The nature of statistical learning theory
Neural Networks: A Comprehensive Foundation

Neural Networks: A Comprehensive Foundation
Practical selection of SVM parameters and noise estimation for SVM regression

Neural Networks
A tutorial on support vector regression

Statistics and Computing
Rapid surface registration of 3D volumes using a neural network approach

Image and Vision Computing
GPU accelerated alignment of 3-D CTA with 2-D X-ray data for improved guidance in coronary interventions

ISBI'09 Proceedings of the Sixth IEEE international conference on Symposium on Biomedical Imaging: From Nano to Macro
A neural network-based method for affine 3D registration of FMRI time series using Fourier space subsets

ICANN'10 Proceedings of the 20th international conference on Artificial neural networks: Part I
Orthogonal least squares learning algorithm for radial basis function networks

IEEE Transactions on Neural Networks
Training feedforward networks with the Marquardt algorithm

IEEE Transactions on Neural Networks
Alignment of 4d coronary CTA with monoplane x-ray angiography

AE-CAI'11 Proceedings of the 6th international conference on Augmented Environments for Computer-Assisted Interventions

Quantified Score

Hi-index	0.00

Visualization

Abstract

We perform a comparative evaluation of different regression techniques for 3D-2D registration-by-regression. In registration-by-regression, image registration is treated as a nonlinear regression problem that relates image features of 2D projection images to the transformation parameters of the 3D image. In this work, we evaluate seven regression methods: Multiple Linear and Polynomial Regression (LR and PR), k-Nearest Neighbour (k-NN), Multiple Layer Perceptron with conjugate gradient optimization (MLP-CG) and with Levenberg-Marquardt optimization (MLP-LM), Radial Basis Function network (RBF) and Support Vector Regression (SVR). The experiments are performed using simulated X-ray images (DRRs) of nine coronary vessel trees, allowing us to compute the mean target registration error (mTRE) to the ground truth. All methods were robust to large initial misalignment and the highest accuracy was achieved using MLP-LM and RBF.