rootSize: Convert a matrix of semi-processed DICOM images to root...

In troyhill/coreCT: Programmatic Analysis of Sediment Cores Using Computed Tomography Imaging

Description

Calculates the number of root/rhizome particles, volumes, and surface areas, for different size classes. This approach directly replicates Earl Davey's manual classification approach. This method is deprecated as of coreCT version 1.3.0.

Usage

rootSize(mat.list, pixelA, diameter.classes = c(1, 2, 2.5, 10),
class.names = diameter.classes,
thickness = 0.625,
airHU = -850.3233,
airSD = 77.6953,
waterHU = 63.912,
waterSD = 14.1728,
pixel.minimum = 4)

Arguments

mat.list	list of DICOM images for a sediment core (values in Hounsfield Units)
pixelA	pixel area (mm2)
diameter.classes	an integer vector of diameter cut points. Units are mm (zero is added in automatically).
class.names	not used presently
thickness	CT image thickness (mm)
airHU	mean value for air-filled calibration rod (all rod arguments are in Hounsfield Units)
airSD	standard deviation for air-filled calibration rod
waterHU	mean value for water-filled calibration rod
waterSD	standard deviation for water-filled calibration rod
pixel.minimum	minimum number of pixels needed for a clump to be identified as a root

Details

Calculates the number of root/rhizome particles, volumes, and surface areas, for different size classes. This function requires that values be Hounsfield Units (i.e., data must be semi-processed from the raw DICOM imagery).

Value

value rootSize returns a dataframe with one row per CT slice. Values returned are the number, volume (cm3), and surface area (cm2) of particles in each size class with an upper bound defined in diameter.classes.

See Also

conv

Examples

ct.slope <- unique(extractHeader(core_426$hdr, "RescaleSlope"))
ct.int   <- unique(extractHeader(core_426$hdr, "RescaleIntercept")) 
# convert raw units to Hounsfield units
HU_426 <- lapply(core_426$img, function(x) x*ct.slope + ct.int)

rootChars <- rootSize(HU_426, pixelA = 0.0596,
diameter.classes = c(2.5, 10))

## Not run: 
# plot using "ggplot" package after transforming with "reshape2" package
area.long <- reshape2::melt(rootChars, id.vars = c("depth"), 
   measure.vars = grep("Area", names(rootChars)))
ggplot2::ggplot(data = area.long, ggplot2::aes(y = -depth, x = value, 
   color = variable)) + ggplot2::geom_point() + ggplot2::theme_classic() + 
   ggplot2::xlab("root external surface area per slice (cm2)")

## End(Not run)

troyhill/coreCT documentation built on Feb. 6, 2021, 7:16 p.m.