voxelwise neighborhood random forest segmentation and prediction

Represents feature images as a neighborhood and uses the features to build a random forest prediction from an image population

vwnrfs(
  y,
  x,
  labelmasks,
  rad = NA,
  nsamples = 8,
  ntrees = 500,
  asFactors = TRUE,
  reduceFactor = 1,
  ...
)

Arguments

y

list of training label images, can be a factor or numeric vector this can also be a regular old vector

x

a list of lists where each list contains feature images

labelmasks

a list of masks where each mask defines the image space for the given list. that is, the nth mask indexes the nth feature set. multi-label masks will try to balance sampling for each label.

rad

vector of dimensionality d define nhood radius

nsamples

(per subject to enter training)

ntrees

(for the random forest model)

asFactors

boolean - treat the y entries as factors

reduceFactor

integer factor by which to reduce (imaging) data resolution

...

arguments to pass to randomForest

Value

list a 4-list with the rf model, training vector, feature matrix and the random mask

Author

Avants BB, Tustison NJ, Pustina D

Examples

mask <- makeImage(c(10, 10), 0)
mask[3:6, 3:6] <- 1
mask[5, 5:6] <- 2
ilist <- list()
lablist <- list()
masklist <- list()
inds <- 1:5
scl <- 0.33 # a noise parameter
for (predtype in c("label", "scalar"))
{
  for (i in inds) {
    img <- antsImageClone(mask)
    imgb <- antsImageClone(mask)
    limg <- antsImageClone(mask)
    if (predtype == "label") { # 4 class prediction
      img[3:6, 3:6] <- rnorm(16) * scl + (i %% 4) + scl * mean(rnorm(1))
      imgb[3:6, 3:6] <- rnorm(16) * scl + (i %% 4) + scl * mean(rnorm(1))
      limg[3:6, 3:6] <- (i %% 4) + 1 # the label image is constant
    }
    if (predtype == "scalar") {
      img[3:6, 3:6] <- rnorm(16, 1) * scl * (i) + scl * mean(rnorm(1))
      imgb[3:6, 3:6] <- rnorm(16, 1) * scl * (i) + scl * mean(rnorm(1))
      limg <- i^2.0 # a real outcome
    }
    ilist[[i]] <- list(img, imgb) # two features
    lablist[[i]] <- limg
    masklist[[i]] <- mask
  }
  rfm <- vwnrfs(lablist, ilist, masklist[[1]], rad = c(2, 2)) # use single mask
  rfm <- vwnrfs(lablist, ilist, masklist, rad = c(2, 2))
  if (predtype == "label") {
    print(sum(rfm$tv != predict(rfm$rfm)))
  }
  if (predtype == "scalar") {
    print(cor(as.numeric(rfm$tv), as.numeric(predict(rfm$rfm))))
  }
} # end predtype loop
#> [1] 1
#> [1] 0.7200797