Reading OME XML from ome.tif file

[hadim]

Hi,

I woud like to know if you could add a method to easly read OME XML file from .ome.tif file. Because I really don't know what it the best way to implement it so I prefer to ask you directly. OME XML can be read when opening the tiff stack with TiffStack_tifffile.

I wrote a small piece of code to show you how xml can be accessed for TiffFile object.

import matplotlib as mpl
import matplotlib.pyplot as plt

import numpy as np
import pandas as pd

import pims
import trackpy as tp

import xmltodict

frames = pims.TiffStack_tifffile("my_file.ome.tif")
if frames._tiff.is_ome:
    t = frames._tiff[0]
    xmlstr = t.tags['image_description'].value
    print(xmlstr[:100])
    print()

    namespaces = {'http://www.openmicroscopy.org/Schemas/OME/2013-06': None,
                              'http://www.openmicroscopy.org/Schemas/SA/2013-06': None}
    md = xmltodict.parse( xmlstr, process_namespaces=True, namespaces=namespaces)

    p = md['OME']['Image']['Pixels']
    print(p['@TimeIncrement'])
    print(p['@PhysicalSizeX'])
    print(p['@PhysicalSizeY'])
    print(p['@PhysicalSizeZ'])

b'<?xml version=\'1.0\' encoding=\'UTF-8\'?>\n<ns0:OME xmlns="http://www.openmicroscopy.org/Schemas/OME/201'

10.076799392700195
0.065
0.065
0.3

Two more things:

• I use a small python module called xmltodict to convert xml file to a dict which is really more convenient to deal with. Since OME XML are rarely huge file even if you don't want to use an external module maybe we/you can add a small method doing the conversion into pims codebase.
• I saw a PR adding bioformats support to pims and I think it's great. However in my case I really don't know to load java because I only deal with OME XML files and I am afraid bioformats will only slowing me down when processing a large number of files.

Thank you

[caspervdw]

I believe you may be able to parse the xml file using the omexml functionality of python-bioformats only (https://github.com/CellProfiler/python-bioformats/blob/master/bioformats/omexml.py). It looks like there is no javabridge involved.

This code parses the xml as an elementtree and defines python classes to conveniently readout the tree. The getPhysicalSize are missing, however.

If you're able to import the OMEXML class from bioformats.omexml, what you would have to do then is:

ome = OMEXML(xmlstr)
print ome.image(0).pixels.channel_count

But your approach does seem more direct. I am curious about the differences in processing time between the elementtree and the dict.

[hadim]

Of course OMEXML doesn't need javabridge. I'm stupid :-)

So I bet elementtree approach is faster than dict conversion but it should be tested to confirm that.

Thanks. I close.

[hadim]

I forgot to ask if you think it's possible to add a class inside tiff_stack.py which add an easy method to retrieve an OMEXML instance ? I saw you did the same thing for Metamorph files (MM_TiffStack).

I can do the PR.

[danielballan]

Please do.
On Mon, Dec 22, 2014 at 12:58 PM Hadrien Mary notifications@github.com
wrote:

I forgot to ask if you think it's possible to add a class inside
tiff_stack.py which add an easy method to retrieve an OMEXML instance ? I
saw you did the same thing for Metamorph files (MM_TiffStack).

I can do the PR.

—
Reply to this email directly or view it on GitHub
#125 (comment).

[hadim]

I will but I need to wait python-{javabridge, bioformats} are compatible with python 3 since all my projects require python 3 !

[caspervdw]

It may be better to write our own OMEXML class OR to make the dependency on python-bioformats optional. We don't want all users that use Tiffstack to be required to install bioformats. Omexml in python-bioformats is under BSD license so it can be copied over.

Besides that, a field in tiffstack that produces some metadata object would be a nice addition!

edit: sorry about the mess with the closing - reopening and the double messages, I guess my phone isn't the best way to send messages

[hadim]

I agree that python-bioformats is an heavy dependency but I don't like the idea to copy OMEXML class from bioformats to pims (for maintenance reason).

What do you think if we make this class optional so it is loaded only if the lib exists otherwise it fails silently or with a warning ?

[caspervdw]

Sounds good

[caspervdw]

@hadim can you try the new implementation of the Bioformats reader that makes use of JPype? (pip install jpype1). It's now supporting Py3.