You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
Jim Rybarski 8aaa453bbd attempt to fix red highlighting 9 years ago
functional_tests Bump version to 2.1.0, updated documentation, removed skips from some unit tests 9 years ago
nd2reader Bump version to 2.1.0, updated documentation, removed skips from some unit tests 9 years ago
tests revert versions to the way they were since it broke publishing, added unit test to ensure they're in sync 9 years ago
.dockerignore added build files to dockerignore 9 years ago
.gitignore added python2.7 virtualenv to gitignore 9 years ago
CHANGELOG.md Bump version to 2.1.0, updated documentation, removed skips from some unit tests 9 years ago
CONTRIBUTING.md Update CONTRIBUTING.md 9 years ago
CONTRIBUTORS.txt realized unit testing isn't even worth it here. yeah, I said that. quote me on it. 10 years ago
Dockerfile resolves #129: renamed filter() to select() 9 years ago
LICENSE.txt license and attribution 10 years ago
Makefile #114: wrote fast filter that works, but we need way more testing for corner cases 9 years ago
README.md attempt to fix red highlighting 9 years ago
ftest.py resolves #140 9 years ago
requirements.txt updated docs, fixed typos 9 years ago
setup.cfg resolves #26 10 years ago
setup.py Bump version to 2.1.0, updated documentation, removed skips from some unit tests 9 years ago
test.py resolves #135. we're ready to publish version 2.0.0 9 years ago

README.md

nd2reader

About

nd2reader is a pure-Python package that reads images produced by NIS Elements 4.0+. It has only been definitively tested on NIS Elements 4.30.02 Build 1053. Support for older versions is being actively worked on.

.nd2 files contain images and metadata, which can be split along multiple dimensions: time, fields of view (xy-plane), focus (z-plane), and filter channel.

nd2reader loads images as Numpy arrays, which makes it trivial to use with the image analysis packages such as scikit-image and OpenCV.

Installation

If you don't already have the packages numpy, six and xmltodict, they will be installed automatically:

pip3 install nd2reader for Python 3.x

pip install nd2reader for Python 2.x

nd2reader is an order of magnitude faster in Python 3. I recommend using it unless you have no other choice.

ND2s

A quick summary of ND2 metadata can be obtained as shown below.

>>> import nd2reader
>>> nd2 = nd2reader.Nd2("/path/to/my_images.nd2")
>>> nd2
<ND2 /path/to/my_images.nd2>
Created: 2014-11-11 15:59:19
Image size: 1280x800 (HxW)
Image cycles: 636
Channels: 'brightfield', 'GFP'
Fields of View: 8
Z-Levels: 3

You can iterate over each image in the order they were acquired:

import nd2reader
nd2 = nd2reader.Nd2("/path/to/my_images.nd2")
for image in nd2:
    do_something(image)

Image objects are just Numpy arrays with some extra metadata bolted on:

>>> image = nd2[20]
>>> image
array([[1894, 1949, 1941, ..., 2104, 2135, 2114],
       [1825, 1846, 1848, ..., 1994, 2149, 2064],
       [1909, 1820, 1821, ..., 1995, 1952, 2062],
       ...,
       [3487, 3512, 3594, ..., 3603, 3643, 3492],
       [3642, 3475, 3525, ..., 3712, 3682, 3609],
       [3687, 3777, 3738, ..., 3784, 3870, 4008]], dtype=uint16)

>>> image.timestamp
10.1241241248
>>> image.frame_number
11
>>> image.field_of_view
6
>>> image.channel
'GFP'
>>> image.z_level
0

If you only want to view images that meet certain criteria, you can use select(). It's much faster than iterating and checking attributes of images manually. You can specify scalars or lists of values. Criteria that aren't specified default to every possible value. Currently, slicing and selecting can't be done at the same time, but you can set a range with the start and stop arguments:

for image in nd2.select(channels="GFP", fields_of_view=(1, 2, 7)):
    # gets all GFP images in fields of view 1, 2 and 7, regardless of z-level or frame
    do_something(image)

for image in nd2.select(z_levels=(0, 1), start=12, stop=3000):
    # gets images of any channel or field of view, with z-level 0 or 1, between images 12 and 3000
    do_something(image)

Slicing is also supported and is extremely memory efficient, as images are only read when directly accessed:

for image in nd2[50:433]:
    do_something(image)

# get every other image in the first 100 images
for image in nd2[:100:2]:
    do_something(image)

# iterate backwards over every image
for image in nd2[::-1]:
    do_something(image)

You can also just index a single image:

# gets the 18th image
my_important_image = nd2[17] 

The Nd2 object has some programmatically-accessible metadata:

>>> nd2.height  # in pixels
1280
>>> nd2.width  # in pixels
800
>>> len(nd2)  # the number of images
30528
>>> nd2.pixel_microns  # the width of a pixel in microns
0.22

Each camera has its own settings. If you image multiple wavelengths with one camera, each channel will appear as its own camera:

>>> nd2.camera_settings
{'GFP': <Camera Settings: GFP>
Camera: Andor Zyla VSC-00461
Camera ID: VSC-00461
Exposure Time (ms): 100.0
Binning: 2x2, 'BF': <Camera Settings: BF>
Camera: Andor Zyla VSC-00461
Camera ID: VSC-00461
Exposure Time (ms): 100.0
Binning: 2x2}

Camera information can be accessed programmatically:

>>> nd2.camera_settings['GFP'].id
'VSC-00461'
>>> nd2.camera_settings['GFP'].name
'Andor Zyla VSC-00461'
>>> nd2.camera_settings['GFP'].exposure
100.0
>>> nd2.camera_settings['GFP'].x_binning
2
>>> nd2.camera_settings['GFP'].y_binning
2

Contributing

If you'd like to help with the development of nd2reader or just have an idea for improvement, please see the contributing page for more information.

Bug Reports and Features

If this fails to work exactly as expected, please open an issue. If you get an unhandled exception, please paste the entire stack trace into the issue as well.

Citation

You can cite nd2reader in your research if you want:

Rybarski, Jim (2015): nd2reader. figshare.
http://dx.doi.org/10.6084/m9.figshare.1619960

Acknowledgments

Support for the development of this package was provided by the Finkelstein Laboratory.