A function to compute the distance from spots to borders of classes

distance2border(
  points,
  img.classes,
  x.microns,
  y.microns,
  z.microns,
  class1,
  class2 = NULL,
  mask = array(TRUE, dim(img.classes)),
  voxel = FALSE,
  hist = FALSE,
  main = "Minimal distance to border",
  xlab = "Distance in Microns",
  xlim = c(-0.3, 0.3),
  n = 20,
  stats = TRUE,
  file = NULL,
  silent = FALSE,
  parallel = FALSE
)

Arguments

points: Data frame containing the coordinates of points in microns as X-, Y-, and Z-variables.
img.classes: 3D array (or image) of classes for each voxel.
x.microns: Size of image in x-direction in microns.
y.microns: Size of image in y-direction in microns.
z.microns: Size of image in z-direction in microns.
class1: Which class is the reference class. If is.null(class2), the function computes the distance of points to the border of class (in img.classes).
class2: Which class is the second reference class. If not is.null(class2), the function computes the distance of points from the border between classes class1 and class2. Default: class2=NULL.
mask: Array of mask. Needs to have same dimension as img.classes. Only voxels with mask[i,j,k]==TRUE are used. Default: array(TRUE,dim(img.classes))
voxel: Logical. If TRUE, points coordinates are given as voxels rather than in microns.
hist: Automatically plot histogram using hist() function. Default: FALSE.
main: If (hist) title of histogram. Default: "Minimal distance to border".
xlab: If (hist) description of x axis. Default: "Distance in Microns".
xlim: If (hist) vector of range of x axis (in microns). Default: c(-.3,.3)
n: If (hist) number of bins used in hist(). Default: 20.
stats: If (hist) write statistics into plot. Default: TRUE.
file: If (hist) the file name of the produced png. If NULL, the histogram is plotted to the standard device. Default: NULL.
silent: if TRUE, function remains silent during running time
parallel: Logical. Can we use parallel computing?

Value

The function returns a vector with distances. Negative values correspond to points lying in class1.

Details

This function computes the distances from points to the border of a class or the border between two classes. For the latter, only points in these two classes are used.

Note

Warning: So far no consistency check for arguments is done. E.g., distance2border(randompoints,img.classes=array(1,c(100,100,2)),3,3,1,class1=2) will fail with some cryptic error message (because class1 > max(img.classes)).

Examples

if (FALSE) {
#simulate random data
randompoints<-data.frame("X"=runif(100,0,3),"Y"=runif(100,0,3),"Z"=runif(100,0,.5))
# coordinates in microns!
plot(randompoints$X,randompoints$Y,xlim=c(0,3),ylim=c(0,3),pch=19)

# points in a circle
circlepoints<-read.table(system.file("extdata","kreispunkte.table",
                               package="bioimagetools"),header=TRUE)
plot(circlepoints$X,circlepoints$Y,xlim=c(0,3),ylim=c(0,3),pch=19)

# a circle like image
img<-readTIF(system.file("extdata","kringel.tif",package="bioimagetools"))
img<-array(img,dim(img)) # save as array for easier handling
img(img, z=1)

#and a mask
mask<-readTIF(system.file("extdata","amask.tif",package="bioimagetools"))
img(mask, z=1, col="greyinverted")

xy.microns <- 3 # size in x and y direction (microns)
z.microns <- 0.5 # size in z direction (microns)

# distance from points to class 
d1<-distance2border(randompoints, img, xy.microns, xy.microns, z.microns, class1=1,hist=TRUE)
d2<-distance2border(circlepoints, img, xy.microns, xy.microns, z.microns, class1=1,hist=FALSE)
plot(density(d2),type="l")
lines(c(0,0),c(0,10),lty=3)
lines(density(d1),col="blue")

# use mask, should give some small changes
d3<-distance2border(circlepoints, img, xy.microns, xy.microns, z.microns, 
                                                class1=1,mask=mask,hist=FALSE)
plot(density(d2),type="l")
lines(c(0,0),c(0,10),lty=3)
lines(density(d3),col="blue")

# distance from border between classes
anotherimg<-img+mask
image(seq(0,3,length=300),seq(0,3,length=300),anotherimg[,,1])
points(circlepoints,pch=19)
d4<-distance2border(circlepoints, anotherimg, xy.microns, xy.microns, z.microns, 
                                                               class1=1,class2=2)
plot(density(d4),lwd=2)

# this should give the same answer
d5<-distance2border(circlepoints, anotherimg, xy.microns, xy.microns, z.microns, 
                                                                class1=2,class2=1)
lines(density(-d5),lty=3,col="blue",lwd=1.5)
}