RUBY NIfTI

Ruby NIfTI is a pure-ruby library for handling NIfTI data in Ruby. NIfTI (Neuroimaging Informatics Technology Initiative) is an image format designed primarily for the storage and analysis of MRI & PET imaging data.

Ruby NIfTI currently supports basic access to NIfTI files, including basic and extended header information, as well as image information. It doesn't attempt to touch the image data (rotate it, etc.), but it does provide access to qform and sform orientation matrices. Nonetheless it does provide a nice interface to get at NIfTI info from within ruby. Since Ruby isn't as widely seen as a quantitative scripting language yet (Python is more well known, and PyNifti is more mature than NumericalRuby / NArray) I don't know how widely this will be used, but hopefully it will be useful for somebody.

gem install nifti

Initialize the library:

require "nifti"

Read file:

obj = NIFTI::NObject.new("some_file.nii")

Display some key information about the file:

  puts obj.header['sform_code_descr']
=> "NIFTI_XFORM_SCANNER_ANAT"

Retrieve the pixel data in a Ruby Array:

image = obj.get_image

Load the pixel data to an NArray image object and display it on the screen:

image = obj.get_image_narray

Or load the pixel data into an NImage (for easier indexing):

image = obj.get_nimage

There are several good NIfTI libraries around (the canonical nifticlib, which includes c, python and matlab interfaces, and Matlab interfaces in the Mathworks image processing toolbox (IPT) and Toolbox for NIfTI by Jimmy Shen.

This library does not intend to replace them, but rather to provide a few ruby convenience methods for quickly accessing imaging data from ruby without having to call out to external programs, as well as write custom extensions to the NIfTI header.

Feel free to use the github issue tracker to post any questions, or email me directly.

Ruby NIfTI is highly derivative of the very good library Ruby DICOM, from which all of the design and most of the code has been cold-heartedly stolen. Many thanks to Chris Lervag - this wouldn't exist without his examples.

• Development / Source code
• Documentation

Each NObject that is successfully read has an array of extended_header hashes. Each extended_header hash has 3 keys, :esize, :ecode, and :data corresponding to the available fields in the nifti header extension.

Here's a silly example (since you'd probably want to read do this using something that's designed for it, like 3dinfo). Suppose you wanted to collect the history of an image from inside the AFNI extended header. Since the data of the AFNI extended header is just xml, you could easily parse it with an xml parser like nokogiri (using the AFNI ecode of 4 to select it from the list of extended headers, and assuming that there's only 1 AFNI header per file):

require 'nifti'; require 'nokogiri'
obj = NIFTI::NObject.new('./T1.nii')
afni_extended_header = obj.extended_header.select{|ext| ext[:ecode] == 4 }[:data]
afni_xml = Nokogiri::XML(xml)
history = afni_xml.xpath("//AFNI_atr[@atr_name='HISTORY_NOTE']").first.children.first.text

Again, this is a trivial example, but shows one usage of the library. Say you want to take find the mean and std dev of the entire image.

obj = NIFTI::NObject.new('./T1.nii', :narray => true)
mean = obj.image.mean
stddev = obj.image.stddev

If you don't have narray installed, you could still use obj.image as a ruby array, but you'd have to collect the summary stats yourself.