Select: Filter to select a particular subpopulation in the structure

In kbolab/moddicom: Package for handling DICOM RT files

Description

Instatiated an object of the class geoLet, Select works on a single patient containing the extracted ROI N.B: pay attention, Select requires a geoLet object in which the ROI is already extracted. See the examples for more details. Select applies two image preprocessing steps on each study you want to analyze: 1) Normalize the values inside ROI, considering the histogram of gray levels and setting as extremes fstPerc and lst Perc NormValue = (OriginalValue-fstPerc)/(lstPerc-fstPerc) 2) Binarize the images inside ROI by applying a Threshold, and setting to normalised values, the pixels with intensities between ThresholdDown and ThresholdUp NaN all values with intensities outside the threshold values

Usage

Select(geoLetVoxelList, fstPerc = 0, lstPerc = 100, ThDown = 0,
  ThUp = 100, ThStep = 10, without.na = FALSE)

Arguments

fstPerc	is the lowest Percentile of the histogram that you use to normalize the pixel values
lstPerc	is the highest Percentile of the histogram that you use to normalize the pixel values
ThDown	is the lowest threshold value
ThUp	is the highest threshold value
ThStep	is the step threshold value
without.na	is the option to decide to if analyze the image with or without na. Set this parameter to TRUE is suggested.

Value

#' Select returns a mmButo oject with the values selected

Examples

## Not run: 
#First of all create an object obj of the class geoLet()
obj <- geoLet()

#now load the DICOM serie of the patient that you want analyze
#using the method openDICOMFolder and indicating the path
# obj$openDICOMFolder(Path = "/home/kboaria/Desktop/Davide/Retto/Validation/Patient1/")

#Before running Select you have to extract the ROI that you want analyze by using the method getROIVoxels of the geoLet object
geoLetVoxelList<-obj$getROIVoxels(Structure = "GTV")
# Now you can run Select on GTV object;#'


## End(Not run)

kbolab/moddicom documentation built on Nov. 29, 2020, 9:11 p.m.