Last modified: 2021-11-04 21:30:18
Compiled: 2021-11-04 21:31:22

Getting started

#Install packages
install.packages(c("devtools", "BiocManager"), repos="http://cran.r-project.org")
BiocManager::install(c("EBImage", "BioImageDbs"), force = TRUE )
devtools::install_github( "kumeS/rMiW", force = TRUE )

#Load packages
library(EBImage)
#remove.packages("rMiW")
library(rMiW)

Optional: Update (ver 3.14)

BiocManager::install(version = "3.14")

Import a kidney image from rMiW

The displayed image (`Mouse01_Kid_x20_z0_RR01.png’) is an image of whole slide imaging (WSI) for observing the kidney tissue of C57BL/6J mouse (male, 10 week-old) stained by H&E.

Read image

file <- system.file("extdata", "Mouse01_Kid_x20_z0_RR01.png", package="rMiW")
file

#Read image
Img <- EBImage::readImage(files = file)
str(Img)

Here, :: in R script indicates an explicit relation between the package and the functions.

Visualization

#Visualization
EBImage::display(Img, method = "raster")

Basic image processing

In this section, we will handle the Image object of EBIimage and perform image processing.

Convert to the grey image

#Read image: delete the 4th element of 3th dimension
#Img <- EBImage::readImage(files = file)

#Convert to the gray image
GrayImg <- rMiW::toGrayScale(Img, mode = "luminance")
str(GrayImg)
str(Img)

#Visualization
EBImage::display(GrayImg, method = "raster")

Resize the image

#1% area size
Img10 <- EBImage::resize(Img, 
                         w = round(dim(Img)[1]*0.1, 0), 
                         filter = "bilinear")

#6% area size
Img25 <- EBImage::resize(Img, 
                         w = round(dim(Img)[1]*0.25, 0), 
                         filter = "bilinear")

#25% area size
Img50 <- EBImage::resize(Img, 
                         w = round(dim(Img)[1]*0.5, 0), 
                         filter = "bilinear")

#50% area size
Img70 <- EBImage::resize(Img, 
                         w = round(dim(Img)[1]*0.707, 0), 
                         filter = "bilinear")

#Visualization
par(mfrow=c(2,2))
EBImage::display(Img10, method = "raster")
text(x = dim(Img10)[1]/20, y = dim(Img10)[2]/10, label = "1% area size", adj = c(0,0), col = "black", cex = 1.5)
EBImage::display(Img25, method = "raster")
text(x = dim(Img25)[1]/20, y = dim(Img25)[2]/10, label = "6% area size", adj = c(0,0), col = "black", cex = 1.5)
EBImage::display(Img50, method = "raster")
text(x = dim(Img50)[1]/20, y = dim(Img50)[2]/10, label = "25% area size", adj = c(0,0), col = "black", cex = 1.5)
EBImage::display(Img70, method = "raster")
text(x = dim(Img70)[1]/20, y = dim(Img70)[2]/10, label = "50% area size", adj = c(0,0), col = "black", cex = 1.5)

Transpose the image

Here we will perform some rotation operations.

#Transpose the image
ImgTrans <- EBImage::transpose(Img)

#Flip or flop the image
Imgflip <- EBImage::flip(Img)
Imgflop <- EBImage::flop(Img)

#Visualization
par(mfrow=c(2,2))
EBImage::display(Img, method = "raster")
text(x = dim(Img)[1]/20, y = dim(Img)[2]/10, label = "Original", adj = c(0,0), col = "black", cex = 1.5)
EBImage::display(ImgTrans, method = "raster")
text(x = dim(ImgTrans)[1]/20, y = dim(ImgTrans)[2]/10, label = "Transpose", adj = c(0,0), col = "black", cex = 1.5)
EBImage::display(Imgflip, method = "raster")
text(x = dim(Imgflip)[1]/20, y = dim(Imgflip)[2]/10, label = "Flip", adj = c(0,0), col = "black", cex = 1.5)
EBImage::display(Imgflop, method = "raster")
text(x = dim(Imgflop)[1]/20, y = dim(Imgflop)[2]/10, label = "Flop", adj = c(0,0), col = "black", cex = 1.5)

Transpose : 90 degree rotation + Horizontal rotation (90度回転 +左右方向の回転)
Flip : Vertical rotation (上下方向の回転)
Flop : Horizontal rotation (左右方向の回転)

Save image

We can save the file in image format (jpeg, png, tiff) or R binary (.Rds).

#Save as PNG
EBImage::writeImage(Img, files = "./Img.png",
                    type = "png")

#Save as TIFF
EBImage::writeImage(Img, files = "./Img.tif",
                    type = "tiff")

#Save as R binary for single Objects (.Rda)
saveRDS(Img, "./Img.Rds")
dir()

#Read Rds format
ImgRds <- readRDS("./Img.Rds")
str(ImgRds)

When we save the object/variable in R as a Rds file and load it, the same R object will be reproduced.

Mark objects in images

Overlap two images and color them.

file <- system.file("extdata", "Cell_Img.Rds", package="rMiW")

#Read image
CellImg <- readRDS(file)
str(CellImg)

#Cell image
par(mfrow=c(1,1))
EBImage::display(CellImg$X, method = "raster")

#Display them side-by-side.
EBImage::display(EBImage::combine(CellImg$X, CellImg$Y), 
                 nx=2, all=TRUE, spacing = 0.01, margin = 70, method = "raster")

#Overlap them
EBImage::display(EBImage::paintObjects(CellImg$Y,
                                       EBImage::toRGB(CellImg$X),
                                       opac=c(0.2, 0.2),
                                       col=c("red","red"), thick=TRUE, closed=FALSE),
                 method = "raster")

#rMiW function
rMiW::ImageView2D(ImgArray_x=CellImg$X,
                  ImgArray_y=CellImg$Y,
                  ImgN=1, 
                  lab=c("Original", "Overlay", "Ground truth"))

Use `drop=FALSE` to prevent the array from being deformed

file <- system.file("extdata", "Cell_Img.Rds", package="rMiW")
file

#Read image
CellImg <- readRDS(file)
str(CellImg)

#drop=TRUE
CellImgT <- CellImg$X[1:512,1:512,,drop=TRUE]
str(CellImgT)

#drop=FALSE
CellImgF <- CellImg$X[1:512,1:512,,drop=FALSE]
str(CellImgF)

Basic clustering using k-means

k-means clustering is an unsupervised clustering technique to partition N observations into k clusters in which each observation belongs to the cluster with the nearest mean.

The clustering with a compressed image of 1024x1024px

Here, we use the k-means clustering to divide the RGB intensity of the image into three classes.

#Read image: delete the 4th element of 3th dimension
#Load from the R binary
Img <- readRDS(system.file("extdata", "Mouse01_Kid_x20_z0_RR01.Rds", package="rMiW"))
str(Img)

#Resize 1024x1024 and perform 3 clustering 
ImgClus3 <- rMiW::Img2DClustering(x=Img, Cluster = 3, XY=1024)
str(ImgClus3)

#Visualize as a color image
rMiW::rasterMiW(ImgClus3, method = "raster")

#Barplot
Calc <- table(unlist(ImgClus3$Cluster)*ImgClus3$ClusterNumber)
barplot(Calc,
        ylab="Pixel number of cluster", ylim=c(0, max(Calc)*1.25), 
        col=colorspace::rainbow_hcl(ImgClus3$ClusterNumber, c = 70))

Session information

## R version 4.1.1 (2021-08-10)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Ubuntu 20.04.3 LTS
## 
## Matrix products: default
## BLAS/LAPACK: /usr/lib/x86_64-linux-gnu/openblas-pthread/libopenblasp-r0.3.8.so
## 
## locale:
##  [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
##  [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
##  [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=C             
##  [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
##  [9] LC_ADDRESS=C               LC_TELEPHONE=C            
## [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
## [1] BiocStyle_2.20.2
## 
## loaded via a namespace (and not attached):
##  [1] bookdown_0.24       rprojroot_2.0.2     digest_0.6.28      
##  [4] crayon_1.4.2        R6_2.5.1            magrittr_2.0.1     
##  [7] evaluate_0.14       stringi_1.7.5       rlang_0.4.12       
## [10] cachem_1.0.6        fs_1.5.0            jquerylib_0.1.4    
## [13] ragg_1.2.0          rmarkdown_2.11      pkgdown_1.6.1      
## [16] textshaping_0.3.6   desc_1.4.0          tools_4.1.1        
## [19] stringr_1.4.0       yaml_2.2.1          xfun_0.27          
## [22] fastmap_1.1.0       compiler_4.1.1      systemfonts_1.0.3  
## [25] BiocManager_1.30.16 memoise_2.0.0       htmltools_0.5.2    
## [28] knitr_1.36

rMiW: 01. Providing a basic workflow for microscopy-based images

Satoshi Kume

2021-11-04

Getting started

Optional: Update (ver 3.14)

Import a kidney image from rMiW

Read image

Visualization

Basic image processing

Convert to the grey image

Resize the image

Transpose the image

Save image

Mark objects in images

Use `drop=FALSE` to prevent the array from being deformed

Basic clustering using k-means

The clustering with a compressed image of 1024x1024px

Session information

rMiW: 01. Providing a basic workflow for microscopy-based images

Satoshi Kume

2021-11-04

Getting started

Optional: Update (ver 3.14)

Import a kidney image from rMiW

Read image

Visualization

Basic image processing

Convert to the grey image

Resize the image

Transpose the image

Save image

Mark objects in images

Use drop=FALSE to prevent the array from being deformed

Basic clustering using k-means

The clustering with a compressed image of 1024x1024px

Session information

Use `drop=FALSE` to prevent the array from being deformed