Fast Mixture Model Segmentation

Fast mixture model segmentation used in Boone and Andrews labs

Python requirements

Numpy
Cython
scikit-image

Installation

Create a virtual environment (optional)

$ virtualenv -ppython3 segmentation-env
$ source segmentation-env/bin/activate

Install python requirements (needed to build the package)

$ pip install numpy cython

Install our library (pulls in all other dependencies)

$ pip install segmentation

Usage example

Programmatically:

from segmentation import segmentation
from segmentation import watershed
from skimage.io import imread

image = imread('./image.tiff', plugin='tifffile')[1]  # Read channel 1 of a tiff/flex 
im = segmentation.blur_frame(image) # gaussian blur
segmented, _ = segmentation.mixture_model(im, debug=True) # second return argument is currently unused
labels = watershed(im, segmented)

Command line

$ segment -h # for usage information

$ segment -o segmented.data image.tiff

Output is a memmaped labels array. You can read it like this: