NIST-MNI / patch_morphology

ITK patch morphology tools: denoising, segmentation, grading
GNU General Public License v3.0
3 stars 0 forks source link

ITK patch morphology tools: denoising, segmentation, grading

implements methods from:

Building:

Usage (concise):

These programs were originally designed to be used with MINC files, but should work with any file format supported by ITK version 4 Most tools have an optional paramter called search radius - which specifies how non-local the search should be (in voxels) and patch radius - the radius of the local patch used to extract features.

Denoising

Denoising requires specifying noise level ($sigma), and optionall search radius $search_radius and patch radius $patch_radius

itk_minc_nonlocal_filter input.mnc output.mnc --noise $sigma --search $search_radius --patch $patch_radius

Adaptative denoising

Adaptative denoising have optional parameters: search radius $search_radius and patch radius $patch_radius

itk_minc_nonlocal_filter input.mnc output.mnc --search $search_radius --patch $patch_radius --anlm

Segmentation

All segmentation tools and scripts require library of labelled samples.

Segmetnation tool require library of presegmented examples $train, number of classes including background $classes and optionally search radius and patch radius Training examples are referenced in a comma separated file in the format: <image.mnc>,<labels.mnc>

itk_patch_morphology input.mnc output_labels.mnc --discrete $classes --search $search_radius --patch $patch_radius --train $train

Several high level scripts are included in scripts directory:

Grading

All grading tools and scripts require library of labelled samples.

Similar to segmetnation tool, grading require library of presegmented examples $train and optionally search radius and patch radius Training examples are referenced in a comma separated file in the format: <image.mnc>,<labels.mnc>,<grading> Two training libraries can be provided , which are both loaded (and optionally each is used independently for pre-selection)

itk_patch_morphology input.mnc --grading output_grading.mnc  --search $search_radius --patch $patch_radius --train $train --train2 $train2

High level script for simultaneous grading and segmentation: scripts/snipe_grading_pipeline.pl