convertDir: Convert a directory of raw DICOM images to material classes

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

Description

Calculates the area and volume of material classes for each CT slice in a directory. This version accommodates calibration curves with >4 calibrants, and uses density thresholds converted to Hounsfield Units using the calibration curve (rather than direct calibration rod values) to partition sediment components.

Usage

convertDir(directory = file.choose(), upperLim = 3045, lowerLim = -1025, 
means     = c(-850.3233, 63.912, 271.7827, 1345.0696), 
sds       = c(77.6953, 14.1728, 39.2814, 45.4129),
densities = c(0.0012, 1, 1.23, 2.2),
rootData = TRUE, 
diameter.classes = c(1, 2, 2.5, 10), 
class.names = diameter.classes,
pixel.minimum = 4)

Arguments

directory	a character string that can be a matrix of DICOM images or the address of an individual DICOM file in a folder of DICOM images. The default action is <code>file.choose()</code>; a browser menu appears so the user can select the the desired directory by identifying a single DICOM file in the folder of images.
upperLim	upper bound cutoff for pixels (Hounsfield Units)
lowerLim	lower bound cutoff for pixels (Hounsfield Units)
means	mean values (units = Hounsfield Units) for calibration rods used.
sds	standard deviations (units = Hounsfield Units) for calibration rods used. Must be in the same order as means.
densities	numeric vector of known cal rod densities. Format must be c(air, water, Si, glass)
rootData	if TRUE, rootSize is also called on the matrix
diameter.classes	if rootData is TRUE, this argument provides an integer vector of diameter cut points used by rootSize. Units are mm (zero is added in automatically).
class.names	placeholder, not used presently
pixel.minimum	minimum number of pixels needed for a clump to be identified as a root

Details

Calculates the area and volume of material classes for each CT slice in a directory. Unlike conv, convDir accepts a folder of raw values and makes the conversion to Hounsfield Units using the metadata associated with the DICOM images.

Value

value convertDir returns a dataframe with one row per CT slice. Values returned are the area and volume of seven material classes: gas, peat, roots and rhizomes, rock and shell, fine mineral particles, sand, and water. If rootData = TRUE, the output will also contain data on the abundance (number of particles), volume (cm3), and external surface area (cm2) of the root size classes specified in the diameter.classes argument.

See Also

convertDir is a wrapper for convert. getRootsDir operates similarly.

Examples

materials <- convertDir("core_426", rootData = FALSE)


## Not run:  
# plot using "ggplot" package after transforming with "reshape2" package
mass.long <- reshape2::melt(materials, id.vars = c("depth"), 
   measure.vars = grep(".g", names(materials)))
ggplot2::ggplot(data = mass.long, ggplot2::aes(y = -depth, x = value, 
   color = variable)) + ggplot2::geom_point() + ggplot2::theme_classic() + 
   ggplot2::xlab("mass per section (g)") 

## End(Not run)

coreCT documentation built on Feb. 5, 2021, 5:06 p.m.