A small python repo to analyse SEM images of fibrous materials. The repo can also be used to annotate SEM images that have been taken with the Zeiss Auriga.

The diameter analysis requires the MATLAB® Engine API for Python. Since this repository is written in Python 3.9, this requires MATLAB® 2021b or newer.

Install the python fibresem module in a virtual environment env:

$ py -m venv env
$ pip install fibresem

The module can now be run within the environment env as follows:

$ ./env/Scripts/activate
(env) $ py -m fibresem

To use the fibre diameter analysis, install the MATLAB® Engine API for Python as follows:

$ cd {matlabroot}/extern/engines/python
$ python setup.py build --build-base=$HOME/tmp/build install --user

Remove --user flag for installation within environment.

python -m fibresem [OPTIONS] INPUT_PATH COMMAND1 [ARGS] [COMMAND2 [ARGS]] ...

The following options can be used for [OPTIONS]:

• -v, --verbose
• --help

The commands can be chained. The following commands can be used for [COMMAND1], [COMMAND2], ....

• crop Crop and annotate .tif files and export to png. See Annotating.
• diam Perform diameter analysis. See Diameter analysis.
• rename Auto rename files. See Auto-renaming.

E.g.:

python -m fibresem C:\testfiles\ rename overview.txt crop diam --thick-opt

To get help for a specific command, e.g. diam, use:

fibresem diam --help

Cropping and annotating can be done with the crop command.

py -m fibresem INPUT_PATH crop

• Crops .tif file, removes SEM Bar
• Adds scalebar based on Pixel Size
• Adds sample name
• Saves the image as a .png image in a separate /output/ folder.

Diameter analysis can be done with diam command based on the Simpoly algorithm developed by Murphy et al.[1] Requires MATLAB® 2021b or newer for the integration with Python through the MATLAB® Engine API for Python.

py -m fibresem INPUT_PATH diam [options]

Additional options:

• --thick-opt/--no-thick-opt default: false

By default, the tiff image is loaded by MATLAB®, since this is faster than converting an uint from Python through the API.

The diameter analysis algorithm performs a number of morphological operations to acquire a segmented binary image. The --thick-opt flag will thicken the skeleton and remove branchpoints in an additional cleaning step. This option is recommended, when fibres have diameters over 20 px and display significantly contrasting shading (i.e. when using a secondary electron detector). Otherwise, leave the flag out for the default option. The defaults work best for fibres with diameters between 5 and ~30 px.

Pixel sizes are extracted from the ZEISS tiff tags.

py -m fibresem INPUT_PATH rename OVERVIEW_FILE

The OVERVIEW_FILE should have a structure similar to the following:

img	dish	pos	width	remarks	sample
1	1	2	20	-	PU.088
2	1	2	100	-	PU.088
3	1	2	200	-	PU.088

Two sample images are provided in /sampledata. To run FibreSem with the sample data, activate the virtual environment. Navigate to the folder containing the sample data or copy the sample data path. (Remember to use double quotes, when the sample data path contains spaces.)

$ ./env/Scripts/activate
(env) $ cd [path to sample data]
(env) $ py -m fibresem -v ./ crop diam --no-thick-opt

The verbose output should be as follows:

[14:56:56] Adding images.
- Input Path: C:\dev\python\sem\fibresem\sampledata
-- sample.01_img08.tif
-- sample.02_img20.tif
[14:56:56] Running command 1 of 2
[14:56:56] Running annotate script on 2 images.
[14:56:56] Annotating sample.01_img08.tif
[14:56:56] -- Cropping
[14:56:56] -- Adding scalebar, pixelsize = 9.766 nm
[14:56:56] -- Adding sample name
[14:56:56] -- Writing output to: sample.01_img08.png
[14:56:57] -- Output written
...
[14:56:58] Running command 2 of 2
[14:56:58] Running script: diameter_analysis
[14:56:58] Engine Handler not defined.
[14:56:58] Matlab module loaded.
[14:56:58] Starting Matlab Engine ...
[14:57:01] Starting diameter analysis.
[14:57:01] Diameter analysis parameter 'optimise_for_thin_fibres' = True
[14:57:01] Analyzing 01 of 2: sample.01_img08.tif
Loading image in MATLAB
Enhance contrast
Erode Grayscale
Morphological Reconstruction
...
sample.01_img08.tif:    avgp: 8.570 px     sdevp: 1.786 px   avg: 0.084 µm   sdev: 0.017 µm 
sample.02_img20.tif:    avgp: 11.949 px    sdevp: 3.125 px   avg: 0.583 µm   sdev: 0.153 µm 
[14:57:16] Exporting analysis to C:\dev\python\sem\fibresem\sampledata\export.xlsx
[14:57:17] Successfully exported.
[14:57:17] Saved .mat file
[14:57:17] Process exited.

• (30%) Port SIMPoly to pure Python code
• ( 0%) Add automatic quality of analysis indicator.

[1] Murphy, R., Turcott, A., Banuelos, L., Dowey, E., Goodwin, B., & Cardinal, K. O. (2020). SIMPoly: A Matlab-Based Image Analysis Tool to Measure Electrospun Polymer Scaffold Fiber Diameter. Tissue engineering. Part C, Methods, 26(12), 628–636. https://doi.org/10.1089/ten.TEC.2020.0304