Segmentation via Layered Surface Detection

1. Description of repository content
2. Set-up to run tutorials
3. Resources and inspiration
4. Contributions

1. Description of repository content

The layered surface tool uses a sparse layered graphs in order detect layers in 2D and 3D images. The primary benefit of using this tool over standard edge detection is that it can detect layers even when there are edge breaks or gaps in the layer.

This repository contains tutorials for using the layered surface detection tool. Included is a few Jupyter notebook tutorials that are designed to give you an idea on how to use the layered surface tool. You should start with the LayeredSurfaceDetection_tutorial.ipynb notebook which discuss the basics of the layered surface tool and applies it on synthetic data as well as on simple real-world data. More complex examples are provided in the NerveSegmentation2D_example.ipynb and NerveSegmentation3D_example.ipynb notebooks, which describe how to apply the layered surface tool to circular regions via radial unwrapping. A Python file containing helper functions used in the tutorials is included in utilsLS.py, and visualization functions are included in the utilsVisualizationLS.py.

These tutorials will give you an understanding of how to apply the Layered Surface tool to image data so that you can use the tool on your own data. You can open the tutorials at .

2. Set-up to run tutorials

These instructions will help you to set up the tool on your local computer. If you need additional help, please see installing Jupyter Notebooks.

For running the tool locally on your computer, you need to install the slgbuilder Python package using pip install slgbuilder.

If you are using Anaconda, you can install slgbuilder and the packages it depends on via the GUI (check package names in environment.yml), or alternatively by runing the environment file on the terminal:

Mac/Linux:

1. Navigate to the folder containing these tutorials and the environment.yml file.
2. To open the terminal, right click on the folder and navigate to
   • Mac: Services/New Terminal at Folder
   • Linux: Open in Terminal
3. Type conda env create -f environment.yml and press enter.
4. Type conda activate qim-LS and press enter.

Windows:

1. Open Anaconda Prompt.
2. cd <tutorials_path>, where tutorials_path is the absolute (full) path to the folder containing these tutorials.
3. Type conda env create -f environment.yml and press enter.
4. Type conda activate qim-LS and press enter.

3. Resources and inspiration

For inspiration, check out the following papers on user cases we have solved with the layered surfaces tool:

• Dahl, Dahl, Larsen. Surface Detection using Round Cut. International Conference on 3D Vision, 2014.
• Dahl, Andersson, Trinderup, Gundlach. Layered Surface Detection in Micro-CT Tetra Pak Data. International Conference on Neutrons and Food 2016.
• Dahl, Einarsson, Darvann, Hermann, Hove, Kakimoto, Kreiborg, Dahl. Automatic measurement of orbital volume in unilateral coronal synostosis. International Symposium on Biomedical Imaging 2016.
• Lindberg, Larsen, Dahl, Jørgensen, Hamann. Quantitative measure of optic disc drusen location in enhanced depth imaging optical coherence tomography scans. Investigative Ophthalmology & Visual Science, 2017.
• Malmqvist, Lindberg, Dahl, Jørgensen, Hamann. Quantitatively measured anatomic location and volume of optic disc drusen: an enhanced depth imaging optical coherence tomography study. Investigative Ophthalmology & Visual Science, 2017.
• Christensen, Larsen, Jensen, Petersen, Larsen, Conradsen, Dahl. Automatic segmentation of abdominal fat in MRI-Scans, using graph-cuts and image derived energies. Image Analysis, Lecture Notes in Computer Science, 2017.
• Dahl, Dahl, Trinderup, Gundlach. Layered surface detection for virtual unrolling. International Conference on Pattern Recognition, 2018.
• Bodner, Bentzen, Dahl, Alfaro, Steenberg, Hjuler, Simonsen. Structural characterization of membrane-electrode-assemblies in high temperature polymer electrolyte membrane fuel cells. Journal of the Electrochemical Society, 2019.
• Borg, Sporring, Dam, Dahl, Dyrby, Feidenhans’l, Dahl, Pingel. Muscle fibre morphology and microarchitecture in cerebral palsy patients obtained by 3D synchrotron X-ray computed tomography. Computers in Biology and Medicine, 2019.
• Dahlin, Rix, Dahl, Dahl, Jensen, Cloetens, Pacureanu, Mohseni, Thomsen, Bech. Three-dimensional architecture of human diabetic peripheral nerves revealed by X-ray phase contrast holographic nanotomography. Scientific Reports, 2020.
• Jeppesen, Christensen, Dahl, Dahl. Sparse layered graphs for multi-object segmentation. Conference on Computer Vision and Pattern Recognition, 2020.
• Jensen, Dahl, Dahl. Multi-object graph-based segmentation with non-overlapping surfaces. Computer Vision for Microscopy Image Analysis, 2020.
• Linndberg, Dahl, Karlesand, Rueløkke, Malmquist, Hamann. Determination of peripapillary vessel density in optic disc drusen using EDI-OCT and OCT angiography. Experimental Eye Research, 2020.
• Andersson, Kjer, Rafael-Pation, Pacureanu, Pakkenberg, Thiran, Ptito, Bech, Dahl, Dahl, Dyrby. Axon morphology is modulated by the local environment and impacts the noninvasive investigation of its structure-function relationship. Proceedings of the National Academy of Sciences of the United States of America, 2020.
• Engberg, Amini, Willerslev, Larsen, Sander, Kessel, Dahl, Dahl. Automated quantification of macular vasculature changes from OCTA images of hematologic patients. International Symposium on Biomedical Imaging, 2020.

4. Contributions

The development of the tutorials is a combined effort from several researchers in the QIM team. The collection of scripts and exercises is in constant development, and actively used to demonstrate the tool and teach at workshops. We would therefore very much appreciate to hear about your experience.

Please contact William Laprade wl@di.ku.dk with issues and feedback.

If you use the tool for research, please cite the original paper: Sparse layered graphs for multi-object segmentation..

@inproceedings{jeppesen2020sparse,
  title={Sparse Layered Graphs for Multi-Object Segmentation},
  author={Jeppesen, Niels and Christensen, Anders N and Dahl, Vedrana A and Dahl, Anders B},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={12777--12785},
  year={2020}
}