Skip to content
/ Segmentation_prep Public

Image preprocessing for different segmentation approaches.

License

MIT license
2 stars 0 forks Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

SchapiroLabor/Segmentation_prep

BranchesTags

Repository files navigation

Segmentation_prep

Image preprocessing for different segmentation approaches

Docker container

The Docker container for this repository is currently at ghcr.io/schapirolabor/segmentation_prep:main

docker pull kbestak/seg_prep:0.1.6

List of scripts and functionalities

background_sub.py --> Backsub (v0.3.4)

Refers to https://github.com/SchapiroLabor/Background_subtraction which also works within this container.

clahe_segmentation_prep.py --> implementation of Skimage CLAHE

https://scikit-image.org/docs/stable/api/skimage.exposure.html#skimage.exposure.equalize_adapthist

Reshapes the image to only include selected channels. Possibility to apply Contrast-limited adaptive histogram equalization (CLAHE) on selected channels.

Arguments:

  • --input (required), path to input image
  • --output (required), path to output image
  • --keep-channel (required), channel indices which should be included in output image
  • --clahe-channel (required), channel indices on which CLAHE should be applied.
  • --pyramid (default=True), should the output be pyramidal?
  • --tile-size (default=1024), tile size for pyramid generation)
  • --pixel-size (required), physical pixel size for metadata in output
  • CLAHE options:
    • --cliplimit (required), clip limit for CLAHE
    • --nbins (default=256), number of bins for CLAHE

Example command with Docker:

docker run \
  -v /path/to/folder/with/input:/input \
  -v /path/to/folder/with/output:/output \
  kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/clahe_segmentation_prep.py \
  --input "/input/image.ome.tif" \
  --output "/output/image_out.ome.tif" \
  --keep-channel 0 1 2 19 \
  --clahe-channel 0 19 \
  --pixel-size 0.23 \
  --cliplimit 0.01 \
  --nbins 256 \
  --pyramid True \
  --tile-size 1024

Example command with Singularity on Helix:

#!/bin/sh
#SBATCH --job-name="demo"
#SBATCH --partition=single
#SBATCH --ntasks=1
#SBATCH --time=01:00:00
#SBATCH --mem=16gb

module load devel/java_jdk/1.18
module load system/singularity/3.9.2

singularity run \
  docker://kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/clahe_segmentation_prep.py \
  --input "/path/to/input/image.ome.tif" \
  --output "/path/to/output/image_out.ome.tif" \
  --keep-channel 0 1 2 19 \
  --clahe-channel 0 19 \
  --pixel-size 0.23 \
  --cliplimit 0.01 \
  --nbins 256 \
  --pyramid True \
  --tile-size 1024

crop_image_by_roi.py

Region of interest extraction based on input csv file. Not intended for large ROIs which would require pyramidal output (this functionality could be implemented if required).

Arguments:

  • --input (required), path to input image file
  • --output (required), path to output folder
  • --pixel-size (default=1.0), physical pixel size in microns
  • --roi (required), path to the csv file containing ROI information

ROI csv file columns:

Region of interest boundaries (only rectangles currently possible) are given as y(min), y(max), x(min), x(max) values for the ROI.

  • y_min specifies the minimum y coordinate of the ROI
  • x_min specifies the minimum x coordinate of the ROI
  • y_max specifies the maximum y coordinate of the ROI
  • x_max specifies the maximum x coordinate of the ROI
  • roi_name specifies the name with which the ROI gets saved (should not have the suffix .tif)

Example command with Docker:

docker run \
  -v /path/to/folder/with/input:/input \
  -v /path/to/folder/with/roi:/roi \
  -v /path/to/output/folder:/output \
  kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/crop_image_by_roi.py \
  --input "/input/image.ome.tif" \
  --output "/output" \
  --roi "/roi/roi.csv" \
  --pixel-size 0.23

Example command with Singularity on Helix:

#!/bin/sh
#SBATCH --job-name="demo"
#SBATCH --partition=single
#SBATCH --ntasks=1
#SBATCH --time=01:00:00
#SBATCH --mem=16gb

module load devel/java_jdk/1.18
module load system/singularity/3.9.2

singularity run \
  docker://kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/crop_image_by_roi.py \
  --input "/path/to/input/image.ome.tif" \
  --output "/path/to/output/folder" \
  --roi "/path/to/roi.csv" \
  --pixel-size 0.23

unstack_image.py

Saves specified channels as separate tif˙files. Channel names are saved based on the index. Channel naming based on channel name metadata will be implemented and toggleable.

Arguments:

  • --input (required), path to input image file
  • --output (required), path to output folder
  • --channels (required), channel indices that should be saved as separate images
  • --pixel-size (required), physical pixel size for metadata in output

Example command with Docker:

docker run \
  -v /path/to/folder/with/input:/input \
  -v /path/to/output/folder:/output \
  kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/unstack_image.py \
  --input "/input/image.ome.tif" \
  --output "/output" \
  --channels 0 1 2 19 \
  --pixel-size 0.23

Example command with Singularity on Helix:

#!/bin/sh
#SBATCH --job-name="demo"
#SBATCH --partition=single
#SBATCH --ntasks=1
#SBATCH --time=01:00:00
#SBATCH --mem=16gb

module load devel/java_jdk/1.18
module load system/singularity/3.9.2

singularity run \
  docker://kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/unstack_image.py \
  --input "/path/to/input/image.ome.tif" \
  --output "/path/to/output/folder" \
  --channels 0 1 2 19 \
  --pixel-size 0.23

stack_channels.py

Takes a list of separate images (single-, or multi-channel tif files) to stack to one ome.tif image. A pyramidal writer will be included in the future for larger images. Functionality to stack all images within a folder could be added if needed.

Arguments:

  • --input (required), paths to input images
  • --output (required), path to output image
  • --num-channels, total number of channels in output image (saves computing time, could be removed)

Example command with Docker:

In this example image1.ome.tif is a 2-channel image, and image2-ome.tif is a single-channel image.

docker run \
  -v /path/to/folder/with/inputs:/input \
  -v /path/to/output:/output \
  kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/stack_channels.py \
  --input "/input/image1.ome.tif" "/input/image2.ome.tif" \
  --output "/output/image_out.ome.tif" \
  --num-channels 3

Example command with Singularity on Helix:

#!/bin/sh
#SBATCH --job-name="demo"
#SBATCH --partition=single
#SBATCH --ntasks=1
#SBATCH --time=01:00:00
#SBATCH --mem=16gb

module load devel/java_jdk/1.18
module load system/singularity/3.9.2

singularity run \
  docker://kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/stack_channels.py \
  --input "/path/to/folder/with/inputs/image1.ome.tif" "/path/to/folder/with/inputs/image2.ome.tif" \
  --output "/output/image_out.ome.tif" \
  --num-channels 3

mask_out.py

Takes a single-, or multi-channel tif image and multiplies it with a binary mask to produce a pyramidal ome tif image.

Arguments:

  • --input (required), paths to input image
  • --output (required), path to output image
  • --mask (required), path to input binary mask file
  • --pixel-size, add pixel size to metadata
  • --pyramid (default True), should the output be pyramidal
  • --tile-size (default 1024), tile size for pyramidal outputs

Example command with Docker:

In this example image1.ome.tif is a 2-channel image, and image2-ome.tif is a single-channel image.

docker run \
  -v /path/to/input:/input \
  -v /path/to/mask:/input \
  -v /path/to/output:/output \
  kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/mask_out.py \
  --input "/input/image.ome.tif" \
  --mask "/input/mask.tif" \
  --pixel-size 0.23 \
  --output "/output/output.ome.tif"

Example command with Singularity on Helix:

#!/bin/sh
#SBATCH --job-name="demo"
#SBATCH --partition=single
#SBATCH --ntasks=1
#SBATCH --time=01:00:00
#SBATCH --mem=16gb

module load devel/java_jdk/1.18
module load system/singularity/3.9.2

singularity run \
  docker://kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/mask_out.py \
  --input "/path/to/folder/with/inputs/image.ome.tif" \
  --mask "/path/to/folder/with/mask/mask.tif" \
  --output "/output/output.ome.tif" \
  --pixel-size 0.23

max_projection.py

Takes a multi-channel tif image and applies max projection on nuclear and membrane channels. Outputs a 2-channel image with the nuclear max projections, and membrane max projections with adjustable order.

Arguments:

  • --input (required), paths to input image
  • --output (required), path to output image
  • --nuclear-channels (required), channel indices specifying the nuclear channels on which max projection should be applied
  • --membrane-channels (required), channel indices specifying the membrane channels on which max projection should be applied
  • --first (default nuclear), write --first membrane if the membrane channel should be the first channel in the output image
  • --pixel-size, add pixel size to metadata
  • --pyramid (default True), should the output be pyramidal
  • --tile-size (default 1024), tile size for pyramidal outputs

Example command with Docker:

In this example image.ome.tif is a 20-channel image

docker run \
  -v /path/to/input:/input \
  -v /path/to/output:/output \
  kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/max_projection.py \
  --input "/input/image.ome.tif" \
  --nuclear-channels 0 \
  --membrane-channels 16 17 19 \
  --pixel-size 0.23 \
  --output "/output/output.ome.tif"

Example command with Singularity on Helix:

#!/bin/sh
#SBATCH --job-name="demo"
#SBATCH --partition=single
#SBATCH --ntasks=1
#SBATCH --time=01:00:00
#SBATCH --mem=16gb

module load devel/java_jdk/1.18
module load system/singularity/3.9.2

singularity run \
  docker://kbestak/seg_prep:0.1.6 \
  python3 /seg_prep/max_projection.py \
  --input "/path/to/folder/with/inputs/image.ome.tif" \
  --output "/output/output.ome.tif" \
  --pixel-size 0.23 \
  --nuclear-channels 0 \
  --membrane-channels 16 17 19

About

Image preprocessing for different segmentation approaches.

Resources

Readme

License

MIT license
Activity
Custom properties

Stars

2 stars

Watchers

2 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Languages