AirFlow-ImaLife

Pipeline for automated calculation of bronchial parameters

Introduction

This repo combines a number of tools into an automated process for the extraction and measurement of bronchial parameters on a low-dose CT scan.

It combines the 3D-Unet method bronchinet to obtain the initial airway lumen segmentation. This is followed by the Opfront method which uses optimal-surface graph-cut to separate the inner surface of the airway from the outer surface of the airway. From this, various bronchial parameters can be derived.

Repository Structure

    .
    ├── AirMorph     -> Lobar lung segmentation and lobar airway branch labelling.  
    ├── airflow_legacy      -> Legacy resources for compiling pre/post processing tools.  
    ├── airway_analysis     -> Package processing opfront output and calculating bronchial parameters.
    ├── bronchinet          -> 3D-Unet developed for airway lumen segmentations.  
    ├── opfront          -> Opfront tools for segmenting airway lumen and wall surfaces.  
    ├── phantom_trainer     -> Set of tools for automatically determining parameters for the opfront tool  
    ├── run_scripts         -> Bash scripts used to automate the docker image.  
    ├─────── Dockerfile          -> Dockerfile for the pipeline  
    ├─────── airflow_libs.tar.gz -> Package containing compiled legacy runtime libraries for opfront tools.  
    ├─────── README.md           -> This file.  
    ├─────── requirements.txt    -> List of required packages for airflow docker. Install with pip install -r requirements.txt  
    Submodules in italics.