struct.clustering: Clustering volumes by RoI

In espadon: Easy Study of Patient DICOM Data in Oncology

Clustering volumes by RoI

Description

The struct.clustering function creates a new volume in which voxels are clustered and labeled by region of interest defined in an rt-struct.

Usage

struct.clustering(
  vol,
  struct,
  roi.name = NULL,
  roi.sname = NULL,
  roi.idx = NULL,
  T.MAT = NULL,
  alias = "",
  description = NULL,
  verbose = TRUE
)

Arguments

vol	"volume" class object.
struct	"struct" class object.
roi.name	Vector of exact names of the RoI in the struct object. By default roi.name = NULL. See Details.
roi.sname	Vector of names or parts of names of the RoI in the struct object. By default roi.sname = NULL. See Details.
roi.idx	Vector of indices of the RoI that belong to the struct object. By default roi.idx = NULL. See Details.
T.MAT	"t.mat" class object, created by load.patient.from.Rdcm or load.T.MAT. If T.MAT = NULL, struct$ref.pseudo must be equal to vol$ref.pseudo.
alias	Character string, $alias of the created object.
description	Character string, describing the created object. If description = NULL (default value), it will be set to paste (struct$object.alias,"clustering")
verbose	Boolean. if TRUE (default), the RoI studied are listed.

Details

If roi.name, roi.sname, and roi.idx are all set to NULL, all RoI are selected.

Value

Returns a "volume" class object (see espadon.class for class definitions), of "cluster" modality. This object contains the $cluster.info field, detailing the label and volumes in cm³ of the different clusters. Note that the label NA or value 0 is used for the voxels which are not contained in any RoI (air for instance).

See Also

get.roi.connection

Examples

# loading of toy-patient objects (decrease dxyz for better result)
step <- 5
patient <- toy.load.patient(modality = c("mr", "rtstruct"),  
                            dxyz = rep(step, 3))
MR <- patient$mr[[1]]
S <- patient$rtstruct[[1]]
cluster.vol <- struct.clustering(MR, S, T.MAT = patient$T.MAT, verbose = FALSE)
head(cluster.vol$cluster.info)

# Display
n = nrow(cluster.vol$cluster.info)
col = paste0(c("#000000", rainbow(n - 1)), "af")
breaks <- seq(cluster.vol$min.pixel - 0.5, cluster.vol$max.pixel + 0.5, 
              length.out = n + 1)
par0 <- par()
par(mfrow = c(1,2), mar = c(1, 15, 1, 1))
display.palette(col, factors = cluster.vol$cluster.info$label)
par(mar = c(1, 1, 1, 1))
display.plane(MR, cluster.vol, main = "RoI clustering", view.coord = 0,
              top.col = col, top.breaks = breaks, interpolate = FALSE)
par(mfrow = par0$mfrow, mar = par0$mar)

espadon documentation built on May 8, 2026, 9:07 a.m.