Home

NRRD

NRRD, short for Nearly Raw Raster Data, is a file format and accompanying software library designed for storing and exchanging multi-dimensional raster data along with metadata. It is widely used in scientific imaging, particularly medical imaging and neuroscience, due to its flexibility and simplicity.

A NRRD file consists of a human-readable header that is followed by binary or ASCII data. The

NRRD is designed to be simple to parse yet extensible, supporting n-dimensional datasets and rich metadata.

In typical use cases, NRRD stores volumes such as 2D images or 3D medical scans, potentially with

header
starts
with
a
version
line
and
contains
key-value
pairs
describing
the
data,
including
the
data
type
(type),
the
number
of
dimensions
(dimension),
the
size
along
each
dimension
(sizes),
and
the
encoding
of
the
data
(encoding).
Other
common
fields
specify
the
coordinate
system
(space),
the
origin
and
directions
of
the
axes
(space
origin,
space
directions),
and
how
the
data
is
stored
(for
example,
whether
the
data
is
in
the
same
file
or
in
a
separate
data
file).
Encodings
supported
in
practice
include
raw,
ascii,
gzip,
and
bzip2.
The
header’s
metadata
enables
precise
interpretation
of
voxel
data,
spacing,
orientation,
and
coordinate
framing.
It
is
commonly
used
with
open-source
image
processing
and
visualization
tools
and
libraries,
such
as
Teem
(the
NRRD
library),
ITK,
VTK,
and
3D
Slicer,
often
with
Python
bindings
in
related
tooling.
The
format’s
clear
separation
of
metadata
and
data,
along
with
its
human-readable
header,
facilitates
data
sharing
and
reproducibility.
nontrivial
voxel
spacing
and
orientation.
Its
combination
of
straightforward
structure,
broad
support,
and
flexible
metadata
has
made
it
a
durable
choice
for
scientific
imaging
workflows.