Skip to content

Latest commit

 

History

History
84 lines (52 loc) · 5.02 KB

README.md

File metadata and controls

84 lines (52 loc) · 5.02 KB
Raw

DOI

README - oct-cbort

This library aims to provide entire processing library (in Python) that is used by the OCT research groups within the Center for Biomedical OCT Research (CBORT) at The Wellman Center for Photomedicine. We aim to provide processing framework for structure, angiography, polarization-senstive OCT that we use for our own and collaborative research projects.

This library is developed by the following members of the CBORT directed by Professor Brett E. Bouma:

  • Damon DePaoli (depaoli9; Lead Developer)
  • A. Stephanie Nam (sweetzinc; Co-developer, provided MATLAB module for OCTA, current maintainer)
  • Martin Villiger (Primary advisor, provided MATLAB reconstruction scripts for PS-OCT)
  • Benjamin J. Vakoc (Co-advisor)

With the support from P41EB015903, awarded by the National Institute of Biomedical Imaging and Bioengineering of the National Institutes of Health.

Background:

Unlike traditional imaging modalities, OCT contrasts requires "reconstruction" of the acquired interferometric fringes. With this in mind, the oct-cbort Python package serves the following purposes:

  1. When coupled with acquisition frameworks, the package allows for automated processing of raw OCT data into analyzable endpoint images with no/minimal user input. This is desirable for users of OCT technology, whose primary goal is to view OCT images.

  2. The package has been designed in a modular format to allow for "scripting", namely for OCT contrast developpers. The ability to write Python scripts in Jupyter Notebooks alongside core reconstruction algorithms and validated priors allows for the rapid debugging and creation of novel algorithms.

  3. Finally, the package is a general tool for dissemination, education and rapid community uptake of novel contrast algorithms developped at CBORT.

Steps to begin processing on a new machine:

  1. Install Anaconda (In Windows, choose to install with conda exec in PATH). Install GitHub Desktop.

  2. Clone or download oct-cbort code from GitHub

  3. Open terminal (MacOS/Linux) or command prompt (Windows) and type:

    1. conda create --name oct38 python=3.8
    2. conda activate oct38

  4. Install requirements:

    1. For CPU processing :
      pip install -r requirements.txt
    2. For GPU processing (only perform if you have an NVIDIA GPU:
      1. Install NVIDIA Cuda toolkit (V11.0) (for nvcc)
      2. conda install -c anaconda cudatoolkit=11 (for cuda and cudnn)
      3. pip install -r requirements-gpu.txt (for rest of dep. including cupy)

  5. Within the terminal navigate to oct-cbort directory (ie. cd Documents\Github\oct-cbort) and run a first test by processing example data by typing:
    python examples/3.py

  6. Run processing on a directory with desired settings:

    1. For all frames in file:
      python -m oct E:\OFDIData\user.Damon\[p.SPARC_Phantom][s.BirefPhant_0125_7][01-25-2019_15-06-24] tomo+struct+angio+ps mgh all
    2. For a range of frames:
      python -m oct C:\Users\ofdi_process_six\Desktop\[p.SPARC][s.CalibTest_holder_ChkNerve_NoChange_2][11-24-2018_11-41-52] tomo+angio+struct mgh+tif 1-10
    3. For a very simply gui:
      python -m oct

  7. For Jupyter Notebook tutorials, it is best to perform the following as well to assure your the IPkernel can see your new environment (these are not requird for post-processing) : To install these type in terminal:

    1. conda install -c anaconda ipykernel
    2. python -m ipykernel install --user --name=oct38

Usage tutorials:

Jupyter notebook tutorials are found in /tutorials. These notebooks show how to use the library for more technical scripting purposes. Best practice is to run all tutorials right after installing oct-cbort repository to get familiar with the extended funcitonality.

Contributing:

To contribute, please create your own branch, develop on it, and once your addition is tested, create a "pull request" for administrators to look over before adding to the main branch.

Issues:

If there are issues or bugs, please submit as much information as possible to the issue channel of the repository for future debugging and tracking.

Documentation

The documentation is under active development at this point. However, the library follows Numpy Docstring format and the API information can be found directly on the source code.