Skip to content

Commit

Permalink
Merge pull request #7 from pelkmanslab/cellprofiler_segmentation
Browse files Browse the repository at this point in the history
Adding modules for segmentation as performed in cell profiler
  • Loading branch information
scottberry committed Jul 7, 2020
2 parents 4fc9aa7 + c5059c6 commit e889aed
Show file tree
Hide file tree
Showing 2 changed files with 418 additions and 0 deletions.
@@ -0,0 +1,81 @@
---
version: 0.0.1

input:

- name: primary_label_image
type: LabelImage
key:
help: >
Labeled image encoding primary objects that should be used as seeds for
the detection of secondary objects.
- name: intensity_image
type: IntensityImage
key:
help: >
Grayscale image that should be used to find the outlines of secondary
objects.
- name: threshold
type: Numeric
value: 125
help: >
Maximum value of background.
- name: method
type: Scalar
value: propagation
options:
- propagation
- watershed_gradient
- watershed_image
- distance_n
- distance_b
help: >
Segmentation method.
- name: regularization_factor
type: Numeric
value: 0.01
help: >
Used only for 'propagation' method. Larger values cause the distance
between objects to be more important than the intensity image in
determining cut lines. Smaller values cause the intensity image to
be more important than the distance between objects.
- name: distance_to_dilate
type: Numeric
value: 3
help: >
Used only for 'distance_n', 'distance_b' methods. The number of
pixels by which the primary objects will be expanded.
- name: fill_holes
type: Boolean
value: true
help: Should holes in the secondary objects be filled?

- name: plot
type: Plot
value: false
help: Should a figure be created?

output:

- name: secondary_label_image
type: LabelImage
key: segment_secondary.secondary_label_image
help: Labeled image encoding secondary objects.

- name: figure
type: Figure
help: >
Comparison of "primary_label_image" and "secondary_label_image". Also
shows outlines of objects in "secondary_label_image" on "intensity_image".

0 comments on commit e889aed

Please sign in to comment.