EPFL
 Biomedical Imaging Group
EPFL   Student Projects: Vincent Ducry
English only    BIG > Teaching > Student Projects > Vincent Ducry
 CONTENTS
Home page
Events
Members
Publications
Tutorials and Reviews
Research
Demos
Download Algorithms
Teaching
bulletStudent projects

Multi-target tracking of neurons in a sequence of images


Vincent Ducry

Microtechnique Section, EPFL


Semester Project

February 2003


Motivation

Cellular migration plays a key role in the development of the neural system. In order to get a better insight of all the phenomena that take place in such development, a team of neurophysicians from the university of Turin is studying in-vitro the movements of young neurons. The observations are made by means of sequences of images taken by a phase contrast microscope. Paolo Ariano and his team postulate that the neurons are not moving randomly but according the Langevin equation. They are thus interested in an automatic way to get the soma (the cellular body of neuron) trajectories. Image processing is therfore relevant in order to analyze the big quantity of information.

Tracking techniques

Two methods are implemented and tested in order to process the multi-target tracking of neurons. The first method is very intuitive and uses the Watershed by flooding 2D as segmentation algorithm whereas the second technique is more innovative. 3D (2D + time) segmentation is processed on subvolumes by means of the Watershed by flooding 3D algorithm. By such method, the 3D tube-shaped basins correspond directly to the wanted trajectories.

Conclusion

The innovative tracking algorithm is working well and relatively fast whereas the intuitive method is not well adapted to the case of soma junctions or separations. Unfortunately, the conditions to use 3D segmentation are limited which make the method not appropriate to other types of images.

Figure 1:
3D tube-shaped basin showing the junction of two neurons
Figure 2:
Trajectories resulting of the innovative method (Watershed by flooding 3D)