Neuroimaging images are large, orientation‑sensitive rasters. The goal of these helpers is to make reasonable defaults easy to use: perceptually uniform palettes, robust scaling, fixed aspect ratios, and clean legends—without extra heavy dependencies or any JavaScript.
This vignette shows how to:
The helpers used here are:
resolve_cmap(), scale_fill_neuro(),
theme_neuro()plot_montage(), plot_ortho(),
plot_overlay()annotate_orientation()The examples below try to read a sample NIfTI included with the
package. If that is not available, they create a small synthetic 3D
volume and wrap it in NeuroVol. Either way, the rest of the
code is identical.
set.seed(1)
make_synthetic_vol <- function(dims = c(96, 96, 72), vox = c(2, 2, 2)) {
i <- array(rep(seq_len(dims[1]), times = dims[2]*dims[3]), dims)
j <- array(rep(rep(seq_len(dims[2]), each = dims[1]), times = dims[3]), dims)
k <- array(rep(seq_len(dims[3]), each = dims[1]*dims[2]), dims)
c0 <- dims / 2
g1 <- exp(-((i - c0[1])^2 + (j - c0[2])^2 + (k - c0[3])^2) / (2*(min(dims)/4)^2))
g2 <- 0.5 * exp(-((i - (c0[1] + 15))^2 + (j - (c0[2] - 10))^2 + (k - (c0[3] + 8))^2) / (2*(min(dims)/6)^2))
x <- g1 + g2 + 0.05 * array(stats::rnorm(prod(dims)), dims)
sp <- NeuroSpace(dims, spacing = vox)
NeuroVol(x, sp)
}
# Prefer an included demo file. Use a real example from inst/extdata.
demo_path <- system.file("extdata", "mni_downsampled.nii.gz", package = "neuroim2")
t1 <- if (nzchar(demo_path)) {
read_vol(demo_path)
} else {
make_synthetic_vol()
}
dims <- dim(t1)
# Build a synthetic "z-statistic" overlay matched to t1's dims
mk_blob <- function(mu, sigma = 8) {
i <- array(rep(seq_len(dims[1]), times = dims[2]*dims[3]), dims)
j <- array(rep(rep(seq_len(dims[2]), each = dims[1]), times = dims[3]), dims)
k <- array(rep(seq_len(dims[3]), each = dims[1]*dims[2]), dims)
exp(-((i - mu[1])^2 + (j - mu[2])^2 + (k - mu[3])^2) / (2*sigma^2))
}
ov_arr <- 3.5 * mk_blob(mu = round(dims * c(.60, .45, .55)), sigma = 7) -
3.2 * mk_blob(mu = round(dims * c(.35, .72, .40)), sigma = 6) +
0.3 * array(stats::rnorm(prod(dims)), dims)
overlay <- NeuroVol(ov_arr, space(t1))The montage helper facettes a single ggplot object—so you get a shared colorbar, clean panel labels, and proper aspect ratio.
# Choose a sensible set of axial slices
zlevels <- unique(round(seq( round(dims[3]*.25), round(dims[3]*.85), length.out = 12 )))
p <- plot_montage(
t1, zlevels = zlevels, along = 3,
cmap = "grays", range = "robust", probs = c(.02, .98),
ncol = 6, title = "Axial montage (robust scaling)"
)
p + theme_neuro()Notes
range = "robust" uses quantiles (default 2–98%) to
ignore outliers.coord_fixed() + reversed y are handled internally to
preserve geometry and radiological convention.downsample = 2 (or higher) when plotting huge
volumes interactively.plot_montage(
t1, zlevels = zlevels, along = 3,
cmap = "grays", range = "robust", ncol = 6, downsample = 2,
title = "Downsampled montage (for speed)"
)plot_ortho() produces aligned sagittal, coronal, and
axial slices with a shared scale, optional crosshairs, and compact
orientation glyphs.
center_voxel <- round(dim(t1) / 2)
plot_ortho(
t1, coord = center_voxel, unit = "index",
cmap = "grays", range = "robust",
crosshair = TRUE, annotate = TRUE
)Tip: If you have MNI/world coordinates, pass unit = "mm"
and a length‑3 numeric; internally it will convert using
coord_to_grid(space(vol), …) if available.
The overlay compositor colorizes each layer independently (so each can use its own limits and palette) and stacks them as rasters. No extra packages required.
plot_overlay(
bgvol = t1, overlay = overlay,
zlevels = zlevels[seq(2, length(zlevels), by = 2)], # fewer panels for the vignette
bg_cmap = "grays", ov_cmap = "inferno",
bg_range = "robust", ov_range = "robust", probs = c(.02, .98),
ov_thresh = 2.5, # make weaker signal transparent
ov_alpha = 0.65,
ncol = 3, title = "Statistical overlay (threshold 2.5, alpha 0.65)"
)All examples above use neuro‑friendly defaults:
resolve_cmap() wraps base R’s
hcl.colors() with aliases like “grays”, “viridis”,
“inferno”—and safe fallbacks.theme_neuro() keeps panels quiet and legends
slim.scale_fill_neuro().You can switch palettes easily:
plot_montage(
t1, zlevels = zlevels[1:6], along = 3,
cmap = "viridis", range = "robust", ncol = 6,
title = "Same data, Viridis palette"
)downsample for exploration;
keep downsample = 1 for final figures.sessionInfo()
## R version 4.5.1 (2025-06-13)
## Platform: aarch64-apple-darwin20
## Running under: macOS Sonoma 14.3
##
## Matrix products: default
## BLAS: /Library/Frameworks/R.framework/Versions/4.5-arm64/Resources/lib/libRblas.0.dylib
## LAPACK: /Library/Frameworks/R.framework/Versions/4.5-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.1
##
## locale:
## [1] C/en_CA.UTF-8/en_CA.UTF-8/C/en_CA.UTF-8/en_CA.UTF-8
##
## time zone: America/Toronto
## tzcode source: internal
##
## attached base packages:
## [1] stats graphics grDevices utils datasets methods base
##
## other attached packages:
## [1] ggplot2_4.0.1 assertthat_0.2.1 purrr_1.2.0 neuroim2_0.8.5
## [5] Matrix_1.7-3
##
## loaded via a namespace (and not attached):
## [1] sass_0.4.10 generics_0.1.4 mmap_0.6-22
## [4] stringi_1.8.7 lattice_0.22-7 digest_0.6.39
## [7] magrittr_2.0.4 bigstatsr_1.6.2 evaluate_1.0.5
## [10] grid_4.5.1 RColorBrewer_1.1-3 iterators_1.0.14
## [13] rmio_0.4.0 fastmap_1.2.0 foreach_1.5.2
## [16] doParallel_1.0.17 jsonlite_2.0.0 RNifti_1.8.0
## [19] deflist_0.2.0 scales_1.4.0 codetools_0.2-20
## [22] jquerylib_0.1.4 cli_3.6.5 rlang_1.1.7
## [25] crayon_1.5.3 cowplot_1.2.0 splines_4.5.1
## [28] withr_3.0.2 cachem_1.1.0 yaml_2.3.12
## [31] otel_0.2.0 flock_0.7 tools_4.5.1
## [34] parallel_4.5.1 memoise_2.0.1 bigassertr_0.1.7
## [37] dplyr_1.1.4 vctrs_0.6.5 R6_2.6.1
## [40] lifecycle_1.0.5 bigparallelr_0.3.2 stringr_1.6.0
## [43] dbscan_1.2.3 pkgconfig_2.0.3 RcppParallel_5.1.11-1
## [46] bslib_0.9.0 pillar_1.11.1 gtable_0.3.6
## [49] glue_1.8.0 Rcpp_1.1.1 xfun_0.55
## [52] tibble_3.3.0 tidyselect_1.2.1 knitr_1.51
## [55] farver_2.1.2 htmltools_0.5.9 labeling_0.4.3
## [58] RNiftyReg_2.8.4 rmarkdown_2.30 compiler_4.5.1
## [61] S7_0.2.1