NIfTI (Neuroimaging Informatics Technology Initiative) Reader/Writer

Licensing options for this format begin with FME Desktop Professional Edition.

The NIfTI Reader provides FME with access to data in the NIfTI format.

Overview

NIfTI (Neuroimaging Informatics Technology Initiative) is a data format for the storage of Functional Magnetic Resonance Imaging (fMRI) and other medical images.

The NIfTI format is adapted from Analyze™ 7.5, developed by Biomedical Imaging Resource (BIR) at Mayo Clinic. NIfTI and Analyze 7.5 are still compatible – NIfTI simply adds additional fields.

There are two variants of the NIfTI format: NIfTI-1 and NIfTI-2. NIfTI-2 improves the data types supported by NIfTI-1, as well as precision and voxel size; however, the two formats are not binary compatible. Currently only NIfTI-1 is supported by the NIfTI reader.

NIfTI Files can be extended to include additional information, such as attribution (although NIfTI extensions are often program-specific and are not yet supported by FME).

FME interprets NIfTI format files as follows:

  • NIfTI Reader – Reads one or more raster slices from one or more volumes in a single NIfTI file.
  • NIfTI Writer - Writes a series of raster slices into one or more volumes into a single NIfTI file.

Data Information

NIfTI images are registered in a local coordinate system. There is currently no support for spatial referencing information.

Each NIfTI file contains metadata and a voxel in up to 7 dimensions and supports a variety of data types.

Usually NIfTI files have a .nii or .nii.gz extension containing both the header and the data. For backwards compatibility with Analyze 7.5, NIfTI files can be split into a binary header (.hdr) and an image data (.img/.img.gz).

NIfTI metadata provides additional information about the coordinate system and how to interpret the data of the image. This may include parameters such as an intent, a description, or fMRI-specific metadata.

Reader Overview

The NIfTI reader reads one NIfTI file as one or a series of rasters:

  • 2D NIfTI files with a single slice will be read as one raster feature.
  • 3D NIfTI files with a single volume of multiple slices will be read as a series of slices, each one resulting in a raster feature.
  • 4D NIfTI files with multiple volumes of multiple slices will be read as a series of slices, each one resulting in a raster feature with the nifti_volume_number attribute identifying the originating volume.
  • The convention for the axis names is x,y,z, and t, as the 4th dimension is usually used to locate multiple volumes in time.

Feature attribution may include (depending on what is in the header):

  • NIfTI intent (data meaning)
  • Description
  • Units
  • Functional MRI (fMRI) slice information

Writer Overview

The NIfTI writer accepts many slices in the form of FME rasters that will form one or more volumes into a single NIfTI file. The slices are composed to form the volume in the writer so that external applications will be able to read the NIfTI file as a single 3D or 4D file.

Usually these slices will have originated in a NIfTI reader or a DICOM reader as slices of one or more volumes.

The information written to the NIfTI header is taken from the first feature.

The dataset is an output folder into which the writer will write the NIfTI file with the name of the feature type, suffixed by the .nii extension.

Note that the NIfTI writer does not currently write compressed datasets.

Terminology

fMRI

Functional Magnetic Resonance Imaging, or Functional MRI

NIfTI

Neuroimaging Informatics Technology Initiative

voxel

Volume pixel, the smallest distinguishable box-shaped part of a three-dimensional image

NIfTI Quick Facts

About Quick Facts Tables

Format Type Identifier

NIfTI

Reader/Writer

Both

Licensing Level

Professional Edition and above

Dataset Type

File

Feature Type

NIfTI file

Typical File Extensions

  • .nii
  • .nii.gz
  • .hdr .img pair
  • .hdr .img.gz pair

Automated Translation Support

No

User-Defined Attributes

Yes, from a fixed set

Coordinate System Support

No

Generic Color Support

No

Spatial Index

No

Schema Required

No

Transaction Support

No

Encoding Support

Yes

Geometry Type Attribute

nifti_type

Geometry Support

Geometry

Supported?

aggregate

no

circles

no

circular arc

no

donut polygon

no

elliptical arc

no

ellipses

no

line

no

none

no

point

no

polygon

no

raster

yes

solid

no

surface

no

text

no

z values

no

Raster Support

Band Interpretations

Int8, UInt8, Int16, UInt16, Int32, UInt32, Int64, UInt64, Real32, Real64, Red8, Green8, Blue8, Alpha8

Palette Key Interpretations

not applicable

Palette Value Interpretations

not applicable

Nodata Value

not applicable

Cell Origin (x, y)

0.5, 0.5

Rotation Support

No

GCP Support

No

World File Support

No

TAB File Support

No

FME Raster Features

FME raster features represent raster data and use several concepts that are unlike those used in the handling of vector data. The topics below describe how FME processes raster data.

About FME Rasters Tiling and Mosaicking
Raster Properties Band Combining and Separating
Band Properties Band and Palette Selection
Palette Properties Raster Processing
Compression Raster versus Vector Features
Pyramiding Raster File Naming
Interleaving World Files
Interpretation and Data Type TAB Files
Palette Resolution  

NIfTI files can contain rasters with a variety of interpretations. Commonly, they will contain a single numeric band of signed/unsigned integers (8-, 16-, 32- or 64-bit), or floating-point values (32- or 64-bit). Other times, they may contain RGB or RGBA multi-band rasters (8 bits per band).