Parcel centroids using geodesic medoids

Usage

parcel_geodesic_centroid(
  labeled_surface,
  method = c("medoid", "euclidean"),
  component_policy = c("error", "largest", "each", "merge"),
  weights = NULL,
  chunk_size = 2000,
  cache = TRUE
)

Arguments

labeled_surface: A LabeledNeuroSurface.
method: "medoid" (geodesic mean distance) or "euclidean" (closest to Euclidean centroid).
component_policy: How to handle fragmented parcels: "error", "largest", "each", "merge".
weights: Optional numeric edge weights (defaults to E(g)$dist).
chunk_size: Number of source vertices per Dijkstra batch.
cache: Logical; cache symmetric results when targets is NULL or identical to vertices.

Value

Data frame with one row per parcel unit: unit, parcel_id, component, size, centroid_vertex, and Cartesian coordinates.

Examples

# \donttest{
# Requires a LabeledNeuroSurface
# lsurf <- read_freesurfer_annot("lh.aparc.annot", geom)
# centroids <- parcel_geodesic_centroid(lsurf)
# }