spatialKMeans-methods: Spatially-aware k-means clustering

In Cardinal: A mass spectrometry imaging toolbox for statistical analysis

Description

Performs spatially-aware (SA) or spatially-aware structurally-adaptive (SASA) clustering of imaging data. The data are first projected into an embedded feature space where spatial structure is maintained using the Fastmap algorithm, and then ordinary k-means clustering is performed on the projected dataset.

Usage

## S4 method for signature 'SparseImagingExperiment'
spatialKMeans(x, r = 1, k = 3,
    method = c("gaussian", "adaptive"),
    dist = "chebyshev", tol.dist = 1e-9,
    iter.max = 10, nstart = 10,
    algorithm = c("Hartigan-Wong", "Lloyd", "Forgy", "MacQueen"),
    ncomp = 10, BPPARAM = getCardinalBPPARAM(), ...)

## S4 method for signature 'SpatialKMeans2'
summary(object, ...)

## S4 method for signature 'SImageSet'
spatialKMeans(x, r = 1, k = 3,
    method = c("gaussian", "adaptive"),
    iter.max = 10, nstart = 10,
    algorithm = c("Hartigan-Wong", "Lloyd", "Forgy",
        "MacQueen"),
    ncomp = 10, ...)

Arguments

x	The imaging dataset to cluster.
r	The spatial neighborhood radius of nearby pixels to consider. This can be a vector of multiple radii values.
k	The number of clusters. This can be a vector to try different numbers of clusters.
method	The method to use to calculate the spatial smoothing kernels for the embedding. The 'gaussian' method refers to spatially-aware (SA) clustering, and 'adaptive' refers to spatially-aware structurally-adaptive (SASA) clustering.
dist	The type of distance metric to use when calculating neighboring pixels based on r. The options are ‘radial’, ‘manhattan’, ‘minkowski’, and ‘chebyshev’ (the default).
tol.dist	The distance tolerance used for matching pixels when calculating pairwise distances between neighborhoods. This parameter should only matter when the data is not gridded. (Only considers ‘radial’ distance.)
iter.max	The maximum number of k-means iterations.
nstart	The number of restarts for the k-means algorithm.
algorithm	The k-means algorithm to use. See kmeans for details.
ncomp	The number of fastmap components to calculate.
...	Ignored.
object	A fitted model object to summarize.
BPPARAM	An optional instance of BiocParallelParam. See documentation for bplapply.

Value

An object of class SpatialKMeans2, which is a ImagingResult, or an object of class SpatialKMeans, which is a ResultSet. Each element of the resultData slot contains at least the following components:

cluster:

A vector of integers indicating the cluster for each pixel in the dataset.

centers:

A matrix of cluster centers.

correlation:

A matrix with the feature correlations with each cluster.

Author(s)

Kylie A. Bemis

References

Alexandrov, T., & Kobarg, J. H. (2011). Efficient spatial segmentation of large imaging mass spectrometry datasets with spatially aware clustering. Bioinformatics, 27(13), i230-i238. doi:10.1093/bioinformatics/btr246

Faloutsos, C., & Lin, D. (1995). FastMap: A Fast Algorithm for Indexing, Data-Mining and Visualization of Traditional and Multimedia Datasets. Presented at the Proceedings of the 1995 ACM SIGMOD international conference on Management of data.

See Also

spatialShrunkenCentroids

Examples

setCardinalBPPARAM(SerialParam())

set.seed(1)
x <- simulateImage(preset=3, dim=c(10,10), npeaks=10,
    peakheight=c(4,6,8), representation="centroid")

res <- spatialKMeans(x, r=1, k=4, method="adaptive")

summary(res)

image(res, model=1)

Cardinal documentation built on Nov. 8, 2020, 11:10 p.m.