nd2

.nd2 (Nikon NIS Elements) file reader.

This reader provides a pure python implementation of the Nikon ND2 SDK.

It used to wrap the official SDK with Cython, but has since been completely rewritten to be pure python (for performance, ease of distribution, and maintenance) while retaining complete API parity with the official SDK.

Note: This library is not affiliated with Nikon in any way, but we are grateful for assistance from the SDK developers at Laboratory Imaging.

Features good metadata retrieval, direct to_dask and to_xarray options for lazy and/or annotated arrays, and output to OME-TIFF.

This library is tested against many nd2 files with the goal of maximizing compatibility and data extraction. (If you find an nd2 file that fails in some way, please open an issue with the file!)

:book: Documentation

install

pip install nd2

or from conda:

conda install -c conda-forge nd2

Legacy nd2 file support

Legacy nd2 (JPEG2000) files are also supported, but require imagecodecs. To install with support for these files use the legacy extra:

pip install nd2[legacy]

Faster XML parsing

Much of the metadata in the file stored as XML. If found in the environment, nd2 will use lxml which is much faster than the built-in xml module. To install with support for lxml use:

pip install nd2 lxml

Usage and API

Full API documentation is available at https://tlambert03.github.io/nd2

Quick summary below:

import nd2
import numpy as np

my_array = nd2.imread('some_file.nd2')                          # read to numpy array
my_array = nd2.imread('some_file.nd2', dask=True)               # read to dask array
my_array = nd2.imread('some_file.nd2', xarray=True)             # read to xarray
my_array = nd2.imread('some_file.nd2', xarray=True, dask=True)  # read to dask-xarray

# or open a file with nd2.ND2File
f = nd2.ND2File('some_file.nd2')

# (you can also use nd2.ND2File() as a context manager)
with nd2.ND2File('some_file.nd2') as ndfile:
    print(ndfile.metadata)
    ...


# ATTRIBUTES:   # example output
f.path          # 'some_file.nd2'
f.shape         # (10, 2, 256, 256)
f.ndim          # 4
f.dtype         # np.dtype('uint16')
f.size          # 1310720  (total voxel elements)
f.sizes         # {'T': 10, 'C': 2, 'Y': 256, 'X': 256}
f.is_rgb        # False (whether the file is rgb)
                # if the file is RGB, `f.sizes` will have
                # an additional {'S': 3} component

# ARRAY OUTPUTS
f.asarray()         # in-memory np.ndarray - or use np.asarray(f)
f.to_dask()         # delayed dask.array.Array
f.to_xarray()       # in-memory xarray.DataArray, with labeled axes/coords
f.to_xarray(delayed=True)   # delayed xarray.DataArray

# OME-TIFF OUTPUT (new in v0.10.0)
f.write_tiff('output.ome.tif')  # write to ome-tiff file

                    # see below for examples of these structures
# METADATA          # returns instance of ...
f.attributes        # nd2.structures.Attributes
f.metadata          # nd2.structures.Metadata
f.frame_metadata(0) # nd2.structures.FrameMetadata (frame-specific meta)
f.experiment        # List[nd2.structures.ExpLoop]
f.text_info         # dict of misc info
f.voxel_size()      # VoxelSize(x=0.65, y=0.65, z=1.0)

f.rois              # Dict[int, nd2.structures.ROI]
f.binary_data       # any binary masks stored in the file.  See below.
f.events()          # returns tabular "Recorded Data" view from in NIS Elements/Viewer
                    # with info for each frame in the experiment.
                    # output is passabled to pandas.DataFrame

f.ome_metadata()    # returns metadata as an ome_types.OME object
                    # (requires ome-types package)

# allll the metadata we can find...
# no attempt made to standardize or parse it
# look in here if you're searching for metadata that isn't exposed in the above
# but try not to rely on it, as it's not guaranteed to be stable
f.unstructured_metadata()

f.close()           # don't forget to close when not using a context manager!
f.closed            # boolean, whether the file is closed

Metadata structures

These follow the structure of the nikon SDK outputs (where relevant). Here are some example outputs

<details> <summary><code>attributes</code></summary>

Attributes(
    bitsPerComponentInMemory=16,
    bitsPerComponentSignificant=16,
    componentCount=2,
    heightPx=32,
    pixelDataType='unsigned',
    sequenceCount=60,
    widthBytes=128,
    widthPx=32,
    compressionLevel=None,
    compressionType=None,
    tileHeightPx=None,
    tileWidthPx=None,
    channelCount=2
)

</details> <details> <summary><code>metadata</code></summary>

Note: the metadata for legacy (JPEG2000) files will be a plain unstructured dict.

Metadata(
    contents=Contents(channelCount=2, frameCount=60),
    channels=[
        Channel(
            channel=ChannelMeta(
                name='Widefield Green',
                index=0,
                color=Color(r=91, g=255, b=0, a=1.0),
                emissionLambdaNm=535.0,
                excitationLambdaNm=None
            ),
            loops=LoopIndices(NETimeLoop=None, TimeLoop=0, XYPosLoop=1, ZStackLoop=2),
            microscope=Microscope(
                objectiveMagnification=10.0,
                objectiveName='Plan Fluor 10x Ph1 DLL',
                objectiveNumericalAperture=0.3,
                zoomMagnification=1.0,
                immersionRefractiveIndex=1.0,
                projectiveMagnification=None,
                pinholeDiameterUm=None,
                modalityFlags=['fluorescence']
            ),
            volume=Volume(
                axesCalibrated=[True, True, True],
                axesCalibration=[0.652452890023035, 0.652452890023035, 1.0],
                axesInterpretation=(
                    <AxisInterpretation.distance: 'distance'>,
                    <AxisInterpretation.distance: 'distance'>,
                    <AxisInterpretation.distance: 'distance'>
                ),
                bitsPerComponentInMemory=16,
                bitsPerComponentSignificant=16,
                cameraTransformationMatrix=[-0.9998932296054086, -0.014612644841559427, 0.014612644841559427, -0.9998932296054086],
                componentCount=1,
                componentDataType='unsigned',
                voxelCount=[32, 32, 5],
                componentMaxima=[0.0],
                componentMinima=[0.0],
                pixelToStageTransformationMatrix=None
            )
        ),
        Channel(
            channel=ChannelMeta(
                name='Widefield Red',
                index=1,
                color=Color(r=255, g=85, b=0, a=1.0),
                emissionLambdaNm=620.0,
                excitationLambdaNm=None
            ),
            loops=LoopIndices(NETimeLoop=None, TimeLoop=0, XYPosLoop=1, ZStackLoop=2),
            microscope=Microscope(
                objectiveMagnification=10.0,
                objectiveName='Plan Fluor 10x Ph1 DLL',
                objectiveNumericalAperture=0.3,
                zoomMagnification=1.0,
                immersionRefractiveIndex=1.0,
                projectiveMagnification=None,
                pinholeDiameterUm=None,
                modalityFlags=['fluorescence']
            ),
            volume=Volume(
                axesCalibrated=[True, True, True],
                axesCalibration=[0.652452890023035, 0.652452890023035, 1.0],
                axesInterpretation=(
                    <AxisInterpretation.distance: 'distance'>,
                    <AxisInterpretation.distance: 'distance'>,
                    <AxisInterpretation.distance: 'distance'>
                ),
                bitsPerComponentInMemory=16,
                bitsPerComponentSignificant=16,
                cameraTransformationMatrix=[-0.9998932296054086, -0.014612644841559427, 0.014612644841559427, -0.9998932296054086],
                componentCount=1,
                componentDataType='unsigned',
                voxelCount=[32, 32, 5],
                componentMaxima=[0.0],
                componentMinima=[0.0],
                pixelToStageTransformationMatrix=None
            )
        )
    ]
)

</details> <details> <summary><code>experiment</code></summary>

[
    TimeLoop(
        count=3,
        nestingLevel=0,
        parameters=TimeLoopParams(
            startMs=0.0,
            periodMs=1.0,
            durationMs=0.0,
            periodDiff=PeriodDiff(avg=16278.339965820312, max=16411.849853515625, min=16144.830078125)
        ),
        type='TimeLoop'
    ),
    XYPosLoop(
        count=4,
        nestingLevel=1,
        parameters=XYPosLoopParams(
            isSettingZ=True,
            points=[
                Position(stagePositionUm=[26950.2, -1801.6000000000001, 498.46000000000004], pfsOffset=None, name=None),
                Position(stagePositionUm=[31452.2, -1801.6000000000001, 670.7], pfsOffset=None, name=None),
                Position(stagePositionUm=[35234.3, 2116.4, 664.08], pfsOffset=None, name=None),
                Position(stagePositionUm=