Public Repository

Last pushed: 17 days ago
Short Description
Stanford 3d image feature pipeline for the QIFP
Full Description

Quantitative Image Feature Engine

Computes 3D computational features (intensity, shape, edge sharpness, texture, size) given one or more volumetric datasets in DICOM format and nodule segmentations in DICOM Segmentation Object (DSO) Format.

Reference: (your submitted paper once published)

Versions

  • Stable: Fuily-tested version of the engine, use this version for production.
  • Latest: This version contains the latest bug-fixes and features. Might contain untested patches.

Installation

You need to have Docker installed

$ docker pull riipl/3d_qifp

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

Example Configuration File

```ini
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"

Docker Pull Command
Owner
riipl

Comments (0)