Skip to content

vansure/3d_qifp

 
 

Repository files navigation

Quantitative Image Feature Engine

Installation and how to run it.

The QIFE can be run through DOCKER or natively if Matlab is present

Matlab

Matlab Instalation

The QIFE has been tested with versions of MATLAB 2015b or later. To install download the source code from this repository to any folder and navigate to the folder from your MATLAB GUI

Matlab Run Instructions

There are two entry functions in the QIFE.

  • runPipeline.m - That takes the configuration as a Matlab structure and runs the QIFE with that configuration
  • runDSOPipelineCommandLine.m - Loads the configuration from a system file (see below for format and examples), it also sets some defaults for the pipeline.
runPipeline.m

To execute:

runPipeline(config);
runPipeline.m

This function defaults to use the DSOLoader to load files in and assumes that DSO and DICOM images are present in the same directory. To execute:

runDSOPipelineCommandLine(dsoDirPath, outputDirPath, configFilePath, qifpFormat);

where

  • dsoDirPath: directory where one can find the DSO and Dicom files
  • outputDirPath: directory where the output should be written
  • configFilePath configuration file
  • qifpFormat a flag to let it know if your file is using version 0 or 1 of the config file

Docker

Docker Instalation

You need to have Docker installed There are two versions of the dockerized QIFE:

  • Latest: This version contains the latest bugpatches and features but has not been 100% tested
  • Stable: This version contains all the tested components.

To download the latest image use:

$ docker pull riipl/3d_qifp:latest

To download the stable image use:

$ docker pull riipl/3d_qifp:stable

Docker Run Instructions

$ docker run -v DIR_TO_MOUNT:/riipl/data riipl/3d_qifp data/dicoms data/output data/config.ini 1
  • DIR_TO_MOUNT: directory you want to mount inside the docker container
  • /riipl/data/: recommended directory to mount data directory inside the docker container
  • data/dicoms: directory where one can find the DSO and Dicom files
  • data/output: directory where the output should be written
  • data/config.ini configuration file
  • 1: a flag to let it know that you are using the new config file format

Configuration File

There are two formats of configuration files

Configuration File Version 0

Version 0 follows the ini formatting using brackets to define sections and uses key=value to define settings under each section. Spaces and semicolons are ignored. To define multiple values for the same configuration use a comma-separated list:

Example Version 0 Configuration File:

[global]
parallelMode="none"
numberOfProcessors="max"
uidToProcess="all"

[input]
component="dsoLoader"

[preprocessing]
components="maximumConnected,holeFilling"

[featureComputation]
components="information,size,intensity,sphericity,roughness,edgeSigmoidFitting,lvii,glcm,connectedRegions"

[output]
components="csvOutput,maxAreaImage,references"

[edgeSigmoidFitting]
numberOfNormals=1200

[csvOutput]
final=true
each=true

[maxAreaImage]
each=true
windowLevelPreset="ctLung"

Configuration File Version 1

Version 1 fmakes a modification on the ini standard by prepending sections to each key using a pipe, i.e. section|key=value to define settings under each section. Spaces and semicolons are ignored. To define multiple values for the same configuration use a comma-separated list:

Example Version 1 Configuration File:

global|parallelMode="none"
global|numberOfProcessors="max"
global|uidToProcess="all"

input|component="dsoLoader"
preprocessing|components="maximumConnected,holeFilling"
featureComputation|components="information,size,intensity,sphericity,roughness,edgeSigmoidFitting,lvii,glcm,connectedRegions"
output|components="csvOutput,maxAreaImage,references"

edgeSigmoidFitting|numberOfNormals=1200

csvOutput|final=true
csvOutput|each=true

maxAreaImage|each=true
maxAreaImage|windowLevelPreset="ctLung"

About

3d pipeline

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages

  • MATLAB 98.5%
  • HTML 1.5%