Image-funcut

Overview

The image-funcut project is kind of a sandbox or testbed for utilities to view, analyse and transform two-photon microscopy data or any other series of images.

At the moment, the project includes the following Python modules:

• imfun.atrous: comprises functions for à trous wavelet transform and related utilities. (Synonims: starlet transform, stationary wavelet transform, non-decimated wavelet transform). Besides transform, there are utility functions to smooth input data with B-splines, remove trends in data or enhance data by noise suppression.
• imfun.bwmorph: helper functions for black-white morphology and binary masks
• imfun.cluster: naive implementations of a few clustering algorithms and distance functions
• imfun.emd: a stub for empirical mode decomposition functions
• imfun.filt: various filters
• imfun.fnmap: collection of functions which project XYT data to 2D images in various non-trivial ways
• imfun.fnutils: a few functional programming-inspired utils
• imfun.fseq: a keystone module: Class definitions and functions to read from files of several formats and represent sequences of images (both lazy and not) and operations on them.
• imfun.lib: miscellaneous helper functions
• imfun.leica: parsing XML files produced by Leica Software during export
• imfun.mes: reading MES files, as created by a Femtonics microscope
• imfun.mmt: multiscale median transform and hybrid median/starlet transform
• imfun.multisale: working with multiscale supports for starlet and median/starlet transforms, including iterative reconstruction from significant coefficients
• imfun.mvm: an implementation of the Multiscale Vision Model object detection algorithm
• imfun.opt: a stub for optimization functions
• imfun.pca: unused, example PCA
• imfun.pica: PCA and ICA implementations
• imfun.som: implementation of Self-organizing Kohonen maps clustering algorithm
• imfun.synthdata: functions to create simple synthetic data sets should be collected here
• imfun.tiffile: Tiffile library by Christoph Gohlke
• imfun.track: functions to track objects in a changing environment or align frames should be collected here
• imfun.ui: Picker class -- a matplotlib-based backend-independent user interface to operate on fseq instances
• imfun.MLFImage: interface to load MLF files produced by Moor laser speckle imaging device.
• frame_viewer.py: a Traits-based GUI wrapper over imfun.ui and other modules with additional features

One of the motivations to start this project was a functional programming approach to image data analysis, hence the name. Also, it's like a final-cut, but with some (geeky) fun.

Example usage

The following will load a series of TIFF files with all color channels and start and interface to pick up ROIs, etc.

    import imfun
    fs = imfun.fseq.open_seq("/path/to/many/tiff/files/*.tif",ch=None)
    p = imfun.ui.Picker(fs)
    p.start()

Documenting all the features is a work in progress...

Dependencies

The project of course relies on the usual core numeric Python packages: Numpy, SciPy and Matplotlib. It draws some ideas from scikit-learn and scikit-image, and may in future use these two more. The package also keeps a copy of tiffile.py by Christoph Gohlke (version 2013.01.18) to load multi-frame TIFF files.

The script frame_viewer.py, a simple GUI wrapper for imfun, also uses Traits and TraitsUI.

License

Except for files, adopted from external sources (such as tiffile.py) the code is GPL. Other open licensing (e.g. MIT LGPL) can be leased on demand.

Publications

The software has been used in production of the following journal articles:

• PMID: 23219568
• PMID: 23211964
• PMID: 24218625
• PMID: 24692513