Cytoseg

An open source project for automatic segmentation of neuropil. Modules under development include mitochondrion, vesicle, and inner-cell detectors. Cytoseg is part of the Cell Centered Database tool set.

The goal of the Cytoseg project is to produce a tool for automatic segmentation of 3D biological datasets, with emphasis on 3D electron microscopy. The project is written in Python and uses the pythonxy platform (which includes scipy and ITK image processing tools). Cytoseg is currently in a pre-alpha stage.

This project contributes to the the larger problem of circuit reconstruction and connectomics by producing tools to segment components of the nervous system from the bottom up.

Cytoseg project source code at google code

Saturday, April 13, 2013

FIBSEM mitochondria segmentation with Cytoseg


  

Data from http://cvlab.epfl.ch/data/em
This dataset was acquired by Graham Knott and Marco Cantoni at EPFL
 
 
Cytoseg command line for this test:
python run_pipeline_test.py /home/rgiuly/images/fua/testing output --trainingImage=/home/rgiuly/images/fua/training --trainingSeg=/home/rgiuly/images/fua/training_groundtruth --voxelTrainingLowerBound=*,*,* --voxelTrainingUpperBound=*,*,* --voxelProcessingLowerBound=*,*,* --voxelProcessingUpperBound=*,*,* --contourTrainingLowerBound=*,*,* --contourTrainingUpperBound=*,*,* --contourProcessingLowerBound=*,*,* --contourProcessingUpperBound=*,*,* --accuracyCalcLowerBound=*,*,* --accuracyCalcUpperBound=*,*,* --labelConfigFile=settings2.py --voxelWeights=0.0026,0.000128 --contourListWeights=7,1 --contourListThreshold=0.5 --step1 --step2 --step3

Posted by at No comments:
Subscribe to: Posts (Atom)

Followers