Skip to contents

Changelog

Source: NEWS.md

ggbrain 1.0 (7Dec2025)

  • feature: added cluster_slices() function that allows user to choose slices based on connected clusters in a given image
  • feature: added geom_brain_clusterized(), which allows users to display connected clusters instead of continuous values
  • feature: added target_resolution() function to resample slices to a different resolution for display. This allows users to upsample images to finer resolution (e.g., show 2.5mm data at 1mm) or downsample for faster rendering and smaller file sizes. Supports nearest neighbor, linear, cubic (default), and lanczos interpolation methods.
  • feature: added annotate_orientation() function to add anatomical orientation labels to plots
  • feature: added annotate_crosshairs() function to add requested cross-hairs to relevant panels
  • feature: Improve speed of find_threads by only decrementing neighbor counts for pixels identified from the initial full scan.
  • feature: hardened tests of input images to check their voxel sizes
  • feature: Text colors on plots default to a color that contrasts with the background (allowing dark-on-light or light-on-dark plots)
  • bugfix: avoid double calls to refine_image() when unifying categorical scales by releveling before data assignment
  • bugfix: ensure fill_holes uses mode for categorical/labeled data (even with fixed outlines) and cap imputation radius for small slices
  • bugfix: replace deprecated aes_string() with tidy-eval aes() mappings for rasters and labels
  • bugfix: replace deprecated rlang::flatten() with internal flatten_list_once() helper
  • bugfix: correct row/column swap in the eastward check of count_neighbors.
  • bugfix: switched flood_fill to stack-based approach to avoid overflow due to recursion.
  • bugfix: Correct invalid 1-based indexing assumption in nearest_pts used for filling holes.
  • bugfix: Use bisided scale for all slices when unify=TRUE but only some slices have positive and negative values
  • bugfix: Use square data for outlines to avoid warnings about misalignment
  • bugfix: Always apply filter expressions passed with images(), not just for labeled images
  • tests: add core vignette-style integration tests covering rendering, outlines, and labeled atlas slices; add tests for refine_image call count and hole filling

ggbrain 0.9.1 (22Aug2025)

CRAN release: 2025-08-22

  • updated ggplot_add methods to be compatible with ggplot2 4.0

ggbrain 0.9.0 (18Mar2025)

CRAN release: 2025-03-20

  • notice: annotate_panel is now called annotate_slice for consistency
  • notice: the plot() method now works with rendered and un-rendered objects, always drawing to the current graphics device
  • feature: Support use of label columns in contrast specification
  • feature: Support use of inline factor() specification in aes() for a geom_outline or geom_brain layer
  • feature: If layer definition uses := syntax, use the lefthand side as the layer name when no name is provided
  • bugfix: Allow for quantile “q” specifications for label position that can be 1-100 (percentile) or 0-1 (quantile)
  • bugfix: Do not throw error when contrast specification includes equals sign (logical tests)
  • bugfix: Only use := syntax for naming contrasts. For compound contrast expressions, use = ; =
  • bugfix: Pass through label_columns as attribute from inline contrast definitions of the form =
  • bugfix: Handle fixed fill in ggbrain_layers when alpha < 1
  • bugfix: retain the order of ordered factors in labels when unify_scales = TRUE
  • bugfix: copy through label columns for simple subset contrasts
  • bugfix: extend background color rectangle to cover entire plot when legend is large
  • bugfix: support ggplot2 3.5.0+, working around collision with ggnewscale for categorical images

ggbrain 0.8.1 (21Mar2023)

CRAN release: 2023-03-21

  • feature: ggbrain objects can be added together, allowing common parts across plots to be reused
  • feature: geom_outline now supports dilation and erosion using the dil_ero argument
  • feature: geom_region_label* and geom_region_text* now support a min_px argument that controls the minimum number of pixels on a slice that will result in a label. Helps reduce labeling of small regions
  • feature: support aes(outline=, group=) for geom_outline, allowing for subdivision of outlines
  • bugfix: add workaround for show.legend bug when there are multiple fill layers: https://github.com/eliocamp/ggnewscale/issues/32
  • bugfix: switch to linewidth instead of size for borders in ggbrain_panel object
  • bugfix: default to cyan color (with message) if geom_outline is added without any color information
  • bugfix: correction to pass forward data.frame after fill_holes algorithm is applied
  • bugfix: nn interpolation for fill_holes now takes the mean of neighbors for continuous-valued data
  • bugfix: use the correct fill column when using remove_specks for a categorical image

ggbrain 0.8 (21Oct2022)

  • first public release of the package to CRAN (hooray!)
  • feature: add define() function to support contrast definition in ggbrain pipeline outside of a given layer
  • feature: fixed fill colors are now supported in geom_brain using geom_brain(fill='color_here')
  • feature: images() supports a volumes argument so that volumes from 4-D images can be visualized
  • feature: plot() on a ggbrain object supports guides argument that passes to plot_layout() in patchwork. Also, the default is now guides="collect", which combines shared legends.
  • feature: support min_coord and max_coord in montage function for coordinate-based ranges
  • update: archive ggbrain_legacy and remove cowplot dependency
  • update: add_images() is now images() for nonredundancy
  • update: add_slices() is now slices() for nonredundancy
  • update: renamed outline_size to just size in geom_outline for consistency with ggplot2 logic (e.g., geom_contour).
  • update: removed add_slice() in favor of slices(). If attributes such as bg_color are set by slices(), these will be the same for each slice in the operation.
  • bugfix: properly display geom_outline for continuous images

ggbrain 0.7 (23Sep2022)

  • feature: abstract fill_holes, trim_threads, and remove_specks to ggbrain_layer – exposed in geom_brain and geom_outline. This allows these modifications to be made for contrast layers.
  • feature: fill_holes now supported in geom_brain and geom_outline objects. Uses flood fill algorithm to find interior holes, then uses nearest neighbor imputation, taking the mode for categorical images and mean for continuous images.
  • feature: for annotations, q syntax now allows for quantiles above 1 or below 0, where 2 = 200% of max and -1 is -100% of min. This allows for annotations to be placed outside of image boundaries.
  • feature: implement Rcpp-based flood fill algorithm for finding interior holes on slices. Used for fill_holes.
  • feature: implement Rcpp-based pixel neighbor counting algorithm to clean up frayed edges of outlines and fills.
  • feature: switch to Rcpp mat2df function instead of reshape2::melt for speed. Remove reshape2 dependency
  • bugfix: when geom_raster is provided with alpha and na.value=‘transparent’, it draws semi-transparent squares. Switched to na.omit() when alpha < 1.
  • bugfix: ggbrain_slices still returns data for layers that are all NA so that layers are not dropped from range calculations.
  • bugfix: do not add a ggbrain_layer to a plot if all of its values are NA. Adds all_na field.
  • bugfix: properly handle ggbrain_images filter when an integer vector is provided
  • bugfix: fix simple_subset for contrast_parser when there are no brackets in the expression
  • bugfix: correct contrast_parser for conjunctions so that both numeric value and label are returned. Also now use := syntax for values.
  • bugfix: fully remove assumption of “label” column in categorical layers, fixing calculation of unique values in each categorical layer
  • bugfix: convert NAs to 0s within contrast parsing so that numeric logical comparisons always yield TRUE/FALSE and not NA.

ggbrain 0.6 (9Aug2022)

  • feature: for labeled images, clean_specks now counts specks by ROI/mask value
  • feature: support filter expression in add_image() to filter (subset) the data for a given image. Useful for thresholding or for subsetting an atlas to retain only certain values.
  • feature: montage() function to make adding many slices in a plane easy
  • feature: support plot() on ggbrain() objects so that the ggbrain settings can be amended prior to the render and plot steps
  • feature: support overall plot aesthetics including title, base_size, bg_color, and text_color. These are arguments to ggbrain()
  • feature: set tiny values on each slice to NA to support transparency at image boundaries (e.g., in ‘dead space’ on a structural)
  • feature: add_slice() function adds single slice and supports panel-specific aesthetics such as title and text_color
  • feature: contrasts that are simple subsetting operation on a labeled image retain labels in the contrast data
  • bugfix: correct accidental resetting of unified scale limits when a bisided scale was empty on one side
  • bugfix: correct range_breaks() for empty data
  • bugfix: pass through breaks to layers regardless of whether scale or breaks are set first
  • bugfix: cleaning specks no longer fails when a slice is empty
  • bugfix: refactor outlines so that empty slices do not generate completely filled layer

ggbrain 0.5.1 (2Aug2022)

  • abstract geom_brain and geom_outline to different R6 classes that inherit from ggbrain_layer
  • support blur_edge to allow for blurred outlines that are less jagged
  • support clean_specks = to remove regions of this size or smaller from slices
  • support alpha transparency in geom_brain and geom_outline layers with alpha=<number> syntax
  • support control of how outlines are grouped/calculated in labeled images using geom_outline(mapping=aes(group=<label_col>))
  • bugfix: allow for empty annotations in overall plot

ggbrain 0.5 (29Jul2022)

  • Initial internal release of ggbrain package