|
Construction of dendritic trees for neuroscience |
Debasree Banerjee
Section Informatique, EPFL |
Semester project June 2006 |
Project Context |
Certain characteristics of neurons such as the number of branches, the number of bifurcations, the length of the branches etc. are important for the understanding of the signaling and learning mechanisms among neurons. Presently, most of the tracing of neurons is done manually which can be very tedious and time consuming. This project aims at the automated tracing of neurons in two dimensions.
In this project, we propose using Dijkstra's single-source shortest path algorithm for tracing the branches of neurons. Methods are suggested in order to detect cross over points where two branches seem to cross each other. A method based on using the directionality output of the steerable filters has been introduced. Further work is required in order to improve this approach.
Fig 1. Original image of neuron
|
Fig 2. Final tracing of neuron |
Figure 1 shows the original noisy image of the neuron. On applying the tracing method. We get a clear tracing of the main branches of the neuron as shown in Figure 2. Some branches are not traced in the final image and this is due to the complexity of the image itself as well as the difficulty of detecting cross over points in two dimensions. |
|
|