SPADE: Small PArticle DEtection

An algorithm primarily design to detect objects whose sizes aren’t larger a few pixels (particles) on fluorescence microscopy images.

It is an simplified version of marked point process.

Requirements

SPADE has only been tested with Python 3.5 but should work with older versions. The only strict requirement is the package numpy. In order to use SPADE to its full extent, we recommend installing the scikit-image package and the full scipy ecosystem.

Usage

from skimage.io import imread, imshow
from skimage.filters import threshold_otsu
from skimage.color import label2rgb
from numpy import percentile
import matplotlib.pyplot as plt

from spade.detection_2d import spade2d
from spade.shapes.examples import potatoids5x5_smallest4px


# Load the example image.
image = imread("example_2d.png")

# Separate cell image from background, using by Otsu's thresholding method.
cell = image > threshold_otsu(image)

# Focus on brightest pixels only
potential_centers = image > percentile(image[cell], 99)

# Detect particles.
particles = spade2d(image=image,
                    shapes_library=potatoids5x5_smallest4px,
                    threshold=20,
                    potential_centers=potential_centers,
                    mask=cell)

# Show detected particles as overlay on our original image.
imshow(label2rgb(particles, image, bg_label=0))
plt.show()

python -m spade.cli INPUT_IMAGE THRESHOLD -o /path/to/output_image.png

Authors & Acknowledgments

SPADE was developed by Nicolas Cedilnik, Éric Debreuve and Xavier Descombes at the MORPHEME team.

If you use SPADE for your scientific work, please include this bibtex (or equivalent) entry:

@misc{cedilnik2016spade,
  title={SPADE: Small particle detection}
  author={Cedilnik, Nicolas and Debreuve, Éric and Descombes, Xavier}
  url={https://pypi.python.org/pypi/small-particle-detection}
  year={2016}
}

License

SPADE is released under the CeCILL-2.1 licence.