As part of the understanding effort of the neural system working and growing mechanism, IBCM biologists are interested in identifying the various proteins that may be related to the microtubules and showing the relation that may exist between different proteins.
Having developed dedicated staining techniques - proteins specific markers - the biologists have been able to make those proteins appear in conventional electronic microscopy images.
In order to have meaningful statistical results, a very large amount of images has to be processed. The LTS has thus been asked to develop an automatic recognition algorithm, able to 'extract' the microtubules and the markers from the rest of the image.
|

|
|
The program output: check image and computed statistics
|
Once binary masks representing the microtubules and the markers have been extracted, it is an easy task to compute the various features the biologists are asking for, such as: the overall microtubules length and surface, the overall quantity of markers, the percentage of markers that are close enough to the microtubules to be considered related to theim, etc
This application has been built using MatlabŪ and external C++ routines, allowing full 'background' automatic processing.