Assessment of separation of functional components with ICA from dynamic cardiac perfusion PET phantom images for volume extraction with deformable surface models

Authors:
Anu Juslin;Anthonin Reilhac;Margarita Magadán-Méndez;Edisson Albán;Jussi Tohka;Ulla Ruotsalainen
Affiliations:
Institute of Signal Processing, Tampere University of Technology, Tampere, Finland;McConnell Brain Imaging Centre, Montreal Neurological Institute, Montreal, Canada;Institute of Signal Processing, Tampere University of Technology, Tampere, Finland;Institute of Signal Processing, Tampere University of Technology, Tampere, Finland;Institute of Signal Processing, Tampere University of Technology, Tampere, Finland;Institute of Signal Processing, Tampere University of Technology, Tampere, Finland
Venue:
FIMH'05 Proceedings of the Third international conference on Functional Imaging and Modeling of the Heart
Year:
2005

Citing 1
Cited 1

A fast fixed-point algorithm for independent component analysis

Neural Computation

ICA based automatic segmentation of dynamic H215 O cardiac PET images

IEEE Transactions on Information Technology in Biomedicine - Special section on new and emerging technologies in bioinformatics and bioengineering

Quantified Score

Hi-index	0.00

Visualization

Abstract

We evaluated applicability of ICA (Independent Component Analysis) for the separation of functional components from H$_{\rm 2}^{\rm 15}$O PET (Positron Emission Tomography) cardiac images. The effects of varying myocardial perfusion to the separation results were investigated using a dynamic 2D numerical phantom. The effects of motion in cardiac region were studied using a dynamic 3D phantom. In this 3D phantom, the anatomy and the motion of the heart were simulated based on the MCAT (Mathematical Cardiac Torso) phantom and the image acquisition process was simulated with the PET SORTEO Monte Carlo simulator. With ICA, it was possible to separate the right and left ventricles in the all tests, even with large motion of the heart. In addition, we extracted the ventricle volumes from the ICA component images using the Deformable Surface Model based on Dual Surface Minimization (DM-DSM). In the future our aim is to use the extracted volumes for movement correction.