Changelog

0.1.0 (2017-02-27)

  • First release on PyPI.

0.1.1 (2017-02-27)

  • Bug fixes during installation of Numba.
  • Added the run() method for main_multiscaler_readout.

0.1.2 (2017-02-27)

  • Includes tifffile and minor improvements.

0.1.3 (2017-02-28)

  • Changed IO from .read() to .readlines() for better Linux compatibility.
  • .tif is now saved frame-by-frame to save memory, and the method was renamed to create_tif().

0.1.4 (2017-02-28)

  • Frames are now generated with a generator.
  • Fix to installation problems of previous version.

0.1.5 (2017-02-28)

  • Single-lined frames are now supported.

0.1.6 (2017-02-28)

  • More tests coverage.
  • Enforced a few types checks.

0.1.7 (2017-03-01)

  • Potential fix to pip install issues.
  • Start of TAG lens interpolation support.

0.2.0 (2017-03-05)

  • Support for TAG lens added - phase interpolation and image display. Note: The algorithm currently assumes that the pulse is triggered at the zero-phase of the TAG lens.
  • pip installation fixed by requiring Numba as a prerequisite.
  • Number of pixels in the “Frame” direction (x) supersedes the number of frames as listed by the user.
  • Due to massive changes, one test is currently broken.

0.3.0 (2017-03-07)

  • Added method create_array to Movie() that returns a deque containing the raw data generated by the np.histogram function, for visualization and analysis purposes.
  • Added method create_single_volume to Movie() that sums all stacks into a single array.
  • Fixed bugs in tag_tools, mainly in verify_periodicity().
  • Allows for more elaborate user inputs, requiring to choose which type of output you wish for.
  • Basic FLIM support.

0.3.1 (2017-03-07)

  • Tiffs are now saved untiled. Depth axis is x-axis.
  • Installation should run smoothly if following the instructions.

0.3.2 (2017-03-07)

  • Added verifications on the FLIM input.
  • Bug fixes in FLIM implementation.

0.3.3 (2017-03-08)

  • Code can take care of the the infamous [-1, ..., -1] index list.
  • Added debug mode in which the algorithm reads only a limited amount of lines from a file.
  • Fixed minor bug in __create_hist.
  • Decreased size of package by removing excess lines of data for tests.

0.3.4 (2017-03-09)

  • More fixes to the [-1] vector problem.
  • Added a sort function before handling the data, because of irregularities.

0.3.5 (2017-03-11)

  • Added sinusoidal interpolation to TAG phase.
  • Sorting is now only done for TAG lens input.
  • Added fileIO_tools.py module for increased simplicity.
  • Added more verifications to user inputs from GUI that pop up sooner, before heavy computation is made.
  • Increased file IO speed with a new np.fromfile method.

0.3.6 (2017-03-14)

  • Basic support for TAG bits - no actual interpolation yet.
  • GUI additions and changes.
  • Minor performance upgrades.

0.4.0 (2017-03-16)

  • Changes file IO completely. Performance should be higher.
  • TAG lens bug fixes.
  • Updated docs.
  • Updated tests.

0.4.1 (2017-03-30)

  • Updates to setup.py to allow docs to build successfully.
  • Small updates to docs.
  • GUI improvements.

0.4.2 (2017-03-30)

  • Added Dask delayed interface.

0.4.3 (2017-04-02)

  • Removed Dask.
  • Refactored class structure, remove the Frame class.
  • Refactored GUI code.

0.4.4 (2017-04-08)

  • Changes to file IO.
  • Number of requested frames should actually matter now.
  • GUI improvements.

0.4.5 (2017-04-17)

  • Bug fixes and improvements to TAG lens interpolation.

0.4.6 (2017-05-16)

  • Use Debug? to read a small portion of an .lst file.
  • Changed defaults in GUI.
  • Allows acquisition in bi-directional scanning mode. This is enabled with the Mirror phase and Flyback parameters in the GUI.
  • Backend changes for possible future support of binary files.
  • The code allows to dismiss unwanted input channels by specifying them as “Empty”.
    • If you mark a channel as containing data while it’s inactive, an error will terminate execution.
  • Massive refactoring of pipeline.

0.4.7 (2017-05-25)

  • Fixed some of the tests.
  • Added option to save or discard photons arriving during the returning phase of a unidirectional scan. This is the default option now.
  • Introduced Fill Fraction parameter that determines the amount of time the mirrors spend “inside” the image.
  • Some tests are working again.
  • Many other bugfixes.

0.4.8 (2017-05-31)

  • Added type hinting. As a result, disabled support for Python version 3.5. Code is now entirely 3.6-dependent.
  • Added .json configuration files to the GUI. It also automatically loads the last modified configuration file.
  • Updated docs.

0.5.0 (2017-06-04)

  • Added the CensorCorrection class for processing generated data using the censor correction method. Current available methods are:
    • censored.gen_bincount_deque(): Bin the photons into their relative laser pulses, and count how many photons arrived due to each pulse.
    • censored.find_temp_structure_deque(): Generate a summed histogram of the temporal structure of detected photons.
  • Fixed linux bug with Deque import.
  • Added tests.

0.5.1 (2017-06-04)

  • Another go at Linux namespace conflicts.

0.5.2 (2017-06-06)

  • Added basic support for “Censor Correction”.

0.5.3 (2017-06-06)

  • Bug fixes, including support for single-pixel images.
  • Script should require less memory while running.

0.5.4 (2017-06-06)

  • Fixed untested typo.

0.5.5 (2017-06-07)

  • Bug fixes for single-pixel frames
  • Bug fixes for defining amount of frames manually in script.
  • Censor correction shouldn’t require as much memory as it did. It’s still not as fast as it can be.
  • Loading a configuration file will make it the “last used” file, reloading it when re-running PySight.

0.5.6 (2017-06-08)

  • ML classification is functional.

0.5.7 (2017-06-08)

  • More robust data generation.
  • Added scikit-learn to requirements.txt and setup.py.
  • CensorCorrection().learn_histograms() now receives the power label as its input - must be an integer.
  • Return of CensorCorrection().learn_histograms() is now data, labels.
  • Made __get_bincount_deque() private. To be accessed using learn_histograms() only.

0.5.8 (2017-06-08)

  • Robustness upgrades.
  • QOL changes to GUI.
  • A “power” number is needed for learn_histograms() - the percent of power given to the Qubig. It’s just for saving, labeling is done with the label keyword.
  • A foldername to which the data will be saved to has also been added.

0.5.9 (2017-06-11)

  • Much (MUCH) faster concatenation of the data.
  • Fixed a bug with the number of empty histograms that were added to the learning dataset.

0.5.10 (2017-06-12)

  • Changes and optimizations for the file IO process.
  • Fixed a bug with laser pulses verification.
  • Added offset parameter for laser input.

0.5.11 (2017-06-22)

  • Added FLIM functionality with laser clock in the Multiscaler’s clock.

0.5.12 (2017-06-22)

  • Fixed small bug with GUI.
  • Possible fix to TAG lens interpolation.

0.5.13 (2017-06-26)

  • Added SciPy dependency.
  • Added nanoFLIM histogramming.

0.5.14 (2017-06-26)

  • Bug fixes and performance improvements.

0.5.15 (2017-06-27)

  • Refactoring of output:

    • Start of censor correction is integrated into the generation of the outputs.
    • More efficient when required to output several types of data.

0.5.16 (2017-06-27)

  • Fixed small bug with censor correction.

0.5.17 (2017-06-29)

  • Fixed .tif generation.
  • Refactoring of FileIO (tests still pass).

0.5.18 (2017-06-29)

  • Added metadata from .lst file to the saved .tif file. Variables saved:
    • “fstchan”
    • “holdafter”
    • “periods”
    • “rtpreset”
    • “cycles”
    • “sequences”
    • “range”
    • “sweepmode”
    • “fdac”

0.5.19 (2017-06-29)

  • Fixed small bug with censoring.
  • Added checks to see whether we need censor correction.
  • Added tests for lst_tools - they should pass, much like file_io‘s tests.

0.5.20 (2017-07-01)

  • Refactored the output-generating script, while changing the possible outputs of PySight:
    • Summed tif.
    • Full stack as tif.
    • In memory - both stack and tif accessed through the movie object.

0.5.21 (2017-07-07)

  • Added the acquisition delay and “hold-after” times to the calculation of the the absolute time of each event.
  • Decreased package size dramatically by deleting unneeded test data.
  • All 34 tests of code pass. I’ll try to keep it that way :)
  • Added an extrapolation method to create fake lines when the line data is too corrupt to work with. This is done using the new “line frequency” and “frame delay” parmaeters in the GUI.

0.5.22 (2017-07-10)

  • Added an optional line frequency entry to the GUI.
  • Refactoring of some parts of the validation tools.
  • Small performance upgrade.
  • Added an option to treat sweeps as lines.