ODF’s maxima extraction using Particle Swarm Optimization. Biomedical
Résumé
Previous works show that brain Diffusion MRI (dMRI) allow to give with more or less great precision the white
matter fiber structure and its possible intersection fibers. In this paper, we show that a Particle Swarm Optimization (PSO) based approach developed for brain dMRI allows to accurately recover the orientations of complex crossing fibers characterized as Orientation Distribution Function (ODF) maxima. Diffusion was
first modeled using the Diffusion Tensor Imaging (DTI), but this model has problems detecting crossing fibers and this has led to develop many other approaches to extract crossing fibers. Models have been proposed such as the High Order Tensor techniques and High Angular Resolution Diffusion Imaging (HARDI) from which we can reconstruct functions like ODF, whose maxima do correspond to the orientations of the multiple fibers. In this paper, we parametrize PSO to extract all the crossing fibers characterized as the maxima of the Orientation Distribution Function (PSO-ODF). Promising experimental results obtained with synthetic data illustrate the potential of the technique.