DynClust (version 2.2)

ClusteringFct: ClusteringFct

Description

'ClusteringFct' clusters together the denoised pixels x time with statistically homogeneous signals at a level 'fp.alpha'

Usage

ClusteringFct(fp.res.denois, fp.var = 1, fp.alpha = 0.05)

Arguments

fp.res.denois

a list containing the results of the denoising procedure (cf. 'callDenoiseVoxel')

fp.var

a numeric indicating the variance of the dataset

fp.alpha

a numeric value indicating the level of the statistical multitest H0

Value

• returns a list containing the indexes of the pixels present in each cluster (as vectors)

Details

As for the denoising procedure, the classes are built such that within each class the clustered signals are statistically coherent. The clustering procedure is a recursive algorithm, and the main loop of which can be decomposed into the following steps: - the list of all the pixels, Lv, is ordered by increasing neighborhood sizes; - a class is defined from a single pixel: at step 1, the class is defined using the first pixel of Lv. The coherence between the signal in the current pixel and all other pixels in Lv is statistically tested. The pixels for which the null hypothesis was not rejected are removed from Lv and included in the current cluster. The estimated signal of the cluster is computed as the averaged signal over all the pixels of the cluster. - the estimated signal of the new class is statistically tested for coherence with the estimated signal of all other existing clusters; - statistically coherent clusters are merged. The algorithm stops when Lv is empty and when the clusters cannot be merged anymore. The number of clusters could be decreased by tuning the level alpha or the variance during the denoising procedure. However, by tuning one of these parameters, users expose themselves to the risk of obtaining under or over-smoothed estimated signals in the clusters. Details about the statistical test for coherence can be found in the references.

References

Rozenholc, Y. and Reiss, M. (2012) _Preserving time structures while denoising a dynamical image_, Mathematical Methods for Signal and Image Analysis and Representation (Chapter 12), Florack, L. and Duits, R. and Jongbloed, G. and van~Lieshout, M.-C. and Davies, L. Ed., Springer-Verlag, Berlin

Examples

# ## Not run:
# library(DynClust)
# data("adu340_4small",package="DynClust")
# #gain of the CCD camera
# #necessary in order to compute the variance of the dataset
# #estimated on calibration experiments
# G             <- 0.146
# #readout variance
# sro2          <- (16.4)^2
# #dataset's variance
# FT            <- adu340_4small
# FT_varhat     <- G*FT+G^2*sro2
# FT            <- FT/sqrt(FT_varhat)
# #launches the denoising step on the dataset with a statistical level of 5\%
# denoisres     <- callDenoiseVoxel(FT,1,fp.mask.size=NULL,fp.nproc=2)
# #launches the clustering step on the dataset with a statistical level of 5\%
# clustres      <- ClusteringFct(denoisres,1)
# x11()
# matplot(getClusterCenters(FT,denoisres,clustres),t="l",lty=1,lwd=2,bty="n")
# x11()
# par(mar=rep(0,4))
# clustmap <- getClusterMap(dim(FT),clustres)
# image(clustmap,col=rainbow(max(clustmap)))
# x11()
# par(mar=rep(0,4))
# image(apply(FT,1:2,mean),col=grey(seq(0,1,len=2^8)))
# ## End(Not run)