v0.1.8

0.1.8 — 2024-08-05

Fixes

• Fix a bug in read_gdf method when the column values are stringified.

v0.1.7

0.1.7 — 2024-07-03

Chore

• Update geopandas version ^1.0.0

Features

• Add a new methods for clustering: cluster_cells, find_lisa_clusters, connected_components and label_connected_components. The cluster_cells works as a polished runner function to cluster spatially the cells of interest with local moran lisa analysis.

Fixes

• remove matplotlib.cm.get_cmap from plot.py file that caused issues

v0.1.6

Fixes

Fix a bug gdf_to_file. Failed when qupath data fields were added to gdf.

v0.1.5

0.1.5 — 2024-02-28

Features

• Add support for querying a WSI with bbox coords.
• Add support for rasterizing a gdf into a instance or semantic segmentation mask
• add support for intensity features

v0.1.4

0.1.4 — 2024-02-05

Docs

• Add cell clustering example
• Add cell regionalization example
• Add medial line extraction example

Features

• Add medial line extraction funtion cellseg_gsontools.geometry.medial_lines. Extracts the medial lines using voronoi diagram.
• Add perpendicular line extraction funtion cellseg_gsontools.geometry.perpendicular_lines. Extracts the perpendicular lines from a given line gdf.
• Add num_processes arguments in local_character, local_distances, local_diversity and shape_metric functions.

v0.1.3

0.1.3 — 2023-12-22

Breaking changes

• The SpatialContext API now takes params: parallel and num_processes in the class constructor instead of the fit()-method.

Features

• Added support for new backends (python libraries) to compute the spatial joins in the SpatialContext-classes. This allows for more efficient computation of the spatial joins especially for very large tissue areas.

• Backends:

  • dask-geopandas
  • spatialpandas

• Add h3 hexgrid support in grid fitting for spatial context classes

Performance

• sjoin operations replaced with sindex ops. This makes subsetting 10-100x faster.

Docs

New documentation website: https://okunator.github.io/cellseg_gsontools/

v0.1.2

0.1.2 — 2023-11-01

Deprecated

• the Pipeline class was removed.

Features

• Add local_distances function to compute the mean/median/sum of the distances between nodes in local neighborhoods

• Add weights2gdf function to convert libpysal.weights.W into a gpd.GeoDataFrame with geometry column consisting of shapely.geometry.LineStrings

• Add grid fitting to _SpatialContext classes. Allows a grid of patches to be overlayed on top of different context areas. The patch sizes and strides can be user defined.

• Better looking links in _SpatialContext.plot(). Different link classes now represented with different colors.

• Add grid_classify method to classify grids based on heuristics.

Performance

• Support for parallel spatial context fitting: _SpatialContext.fit(parallel=True)

Fixes

• Drop duplicates in context2gdf-class method

• clarify the apply_gdf function api.

• clarify the _SpatialContext api.

v0.1.1

0.1.1 — 2023-10-13

Features

• Add join predicate param for sjoin operation in get_objs_within function

Performance

• Optimize spatial subsetting operations by getting rid of redundant operations in spatial context classes.

Perf

• Parallelize and memory optimize AreaMerger.

Fixes

• Simplify border_network computation and return only the node-node links that go accross the border.

• Simplify weights plotting interfacee

• Fix bug in interface and roi network fitting.

v0.1.0-alpha.2

0.1.0-alpha.2 — 2023-09-15

Fixes

• Set crs in _SpatialContext to avoid warnings
• Unify the local_character and local_diversity function api.

Features

• Add option to add multiple columns to local_diversity and local_character
• add is_categorical helper func
• Add hbdscan clustering method.
• Add option to not fit graphs in .fit() method of _SpatialContext-classes

Chore

• add scikit-learn (1.3.0) dependency
• Update to latest geopandas (0.13) and shapely (> 2.0)