Fit a single transform to a bridging registration between brains

Source: R/fit-xform.R

Fit a single transform to a bridging registration between brains

fit_xform_brain(
  sample,
  reference,
  via = NULL,
  type = c("affine", "rigid", "similarity", "tps"),
  pts = NULL,
  npts = if (is.null(pts)) 1000 else NULL,
  scale = c(1, 1),
  ...
)

Arguments

sample

Source template brain (e.g. IS2) that data is currently in. Specified either as character vector or a templatebrain object.

reference

Target template brain (e.g. IS2) that data should be transformed into.

via

(optional) intermediate template brain that the registration sequence must pass through.

type

A character string specifying the type of registration. See Morpho::computeTransform for details.

pts

Optional set of points to use for the fit. If they are not specified they will be randomly sampled (see details).

npts

Number of points to use for fit (defaults to 1000 when no pts argument is specified).

scale

a 2-vector indicating the amount to scale the sample and reference points. You can supply one vector if this is the same. If the transform expects points in microns and returns points in microns then you would need scale=c(1000,1) if you want the input to be in microns and the output to be in nm.

...

Additional arguments passed to fit_xform

Value

A homogeneous affine matrix or a nat::tpsreg object n.b. only in development nat (>= 1.10.1)

Details

If pts are supplied these will be used to construct the fit. This might be useful if you are focussing on a particular brain region and supply relevant neurons. But if you want to work with one registration to use for the whole brain then it's better to use points across the brain.

When points are not provided then fit_xform_brain will first check to see if it can find a neuropil surface model for the sample. If it can, it will sample a number of points that lie inside the surface model. If no surface model can be found, then it will look for a templatebrain object specifying a bounding box around the sample brain. If successful, it will sample npts within that bounding box. If not there will be an error and you will have to supply the points.

See also

fit_xform

Examples

if (FALSE) {
library(nat.flybrains)
t1=fit_xform_brain(sample='FCWB', reference="JFRC2", type='affine')
t1
# the same but for points in nm
fit_xform_brain(sample='FCWB', reference="JFRC2", type='affine', scale=1000)
# run the fit based on a particular group of Kenyon cells
t2=fit_xform_brain(sample='FCWB', reference="JFRC2",
  pts=nat::kcs20, npts=300, type='affine')

nclear3d()
mfrow3d(1, 2, sharedMouse = TRUE)
plot3d(JFRC2)
plot3d(xform(FCWB.surf, t1))
next3d()
plot3d(JFRC2)
plot3d(xform(FCWB.surf, t2))
# the whole brain surfaces clearly match better when you fit on the whole brain
}