0.8.2
 - fix: pin scikit-image version to 0.14.0 because threshold_li has
   changed (https://github.com/scikit-image/scikit-image/issues/3605)
 - enh: implement "roi" configuration option "force" (#8)
 - ref: replace default config values of `nan` with `None` (#13)
 - ref: turn comments in definitions.py into real strings (#23)
 - docs: add automation tutorial (usage of profiles and scripts) 
0.8.1
 - fix: replaced hack for plotting units with better solution due
   to incompatibility with tight_layout in matplotilb 3.0.2
0.8.0
 - feat: implement profile management via `dm_profile` command and
   allow to select profiles via the "--profile" command-line argument
   in the command-line interface (#20)
 - fix: if time is unknown, write "nan" instead of "0" to the output
   statistics file
 - ref: explicitly use row-column coordinates in skimage's regionprops
   (bump skimage version to 0.14.1)
0.7.3
 - fix: do not create "_dm" results directory for root directory in
   recursive analysis mode (#19)
 - docs:
   - add DryMass logo
   - update software description
0.7.2
 - skipped
0.7.1
 - maintenance release
0.7.0
 - feat: support recursive analysis with the "--recursive" command-line
   argument (#15)
0.6.2
 - fix: regression in 0.6.0 ImageJ cannot open compressed hyperstack data;
   disabling compression for now (see #14) 
0.6.1
 - fix: MetaDataMissingError for some data types (qpformat 0.4.3)
0.6.0
 - fix: error when removing ROIs that overlap with two ROIs in the
   background image
 - feat: allow to exclude individual ROIs in drymass.cfg (#5)
 - enh: improve error message verbosity
 - enh: use compression for TIFF output
 - enh: use new class for handling ROIs internally (#9)
 - enh: start counting of ROIs and sensor images at 1 instead of 0 (#11)
 - ref: handle 'None' values directly in config.ConfigFile
0.5.0
 - feat: automatically extract meta data from QPImage/QPSeries data
   (qpformat 0.4.0)
 - fix: setting "dist border px" to 0 still removed objects from border
 - fix: scikit-image methods could not be used for "binary threshold"
 - fix: set "image fix phase offset" to True by default, otherwise
   objects in the background cause a false offset
 - improvement: optimize CLI plots
 - docs:
   - update tutorials with new CLI plots
   - add tutorial for background correction (#3)
0.4.1:
 - fix: extracted wrong phase image from SID4Bio data (qpformat 0.3.5) 
0.4.0
 - feat: ROI background correction supports masking of the
   imaged object via the "amplitude mask sphere" and
   "phase mask sphere" settings in the [bg] section.
 - setup: support qpimage>=0.4.0
 - tests: improve coverage
 - docs: minor update (#7)
0.3.3
 - ci: automate PyPI release with travis-ci
 - setup: bump versions of dependent packages
0.3.2
 - docs: update of introduction and minor changes
0.3.1
 - bump version because of regression fixed in qpformat 0.2.1
 - feat: flush statistics file during writing
0.3.0
 - drop support for Python 3.5
 - ref: background correction in convert.py uses background indices
   starting at 0 while the CLI keeps using indices starting at 1
 - fix: hologram keyword arguments were not used for background
   correction
 - fix: case where index-based background correction does not work
 - feat: sphere analysis is picked up where left off in previous run
 - command line interface:
   - fix: allow white spaces for input data path argument
   - fix: improve verbosity of status messages
 - docs: add troubleshooting section
0.2.0
 - feat: reuse computed ROI data from previous runs
 - ref: update [roi] configuration key names to reflect pixel units:
   "dist border px", "exclude overlap px", "pad border px"
 - fix: improve verbosity of error messages
 - fix: config value "[roi] size variation" must be in interval [0,1]
 - fix: handle ROIs with failed edge detection in "dm_analyze_sphere"
0.1.4
 - feat: add local thresholding step before segmentation
 - fix: check number of qpimages in sensor_data.h5 during dm_convert
 - fix: "exclude overlap" only used for background image
 - fix: not possible to analyze phasics data folder when raw data
   is present (qpformat 0.1.5)
0.1.3
 - automatically fix inverted ranges when loading "roi_slices.txt"
 - remove "holo" section from drymass.cfg if not relevant
 - allow user-defined ROI selection via "[roi] enabled = False"
 - added hologram analysis parameter tuning (qpimage 0.1.6)
 - added 2D model fitting to sphere analysis (qpsphere 0.1.4)
 - fixed `cfg_funcs.int_or_str`
0.1.2
 - allow to disable sensor image tif export
 - support input QPI data with different shapes
 - fixed wrong assumption about definition of dry mass in cells
   (water vs. intracellular salt solution)
 - allow to use separate file or series index for background correction
   with experimental data (#6)
 - small visualization improvements
0.1.1
 - add file name to image identifier and update plotting (#4)
 - renamed "object" to "identifier" in statistics output
 - binary-based background correction (manual or automated threshold)
 - allow to disable background correction with "[bg] enabled = False"
 - ask for missing meta data keys in CLI only when they are required
0.1.0
 - initial release
