R/medical.R

In EagerAI/fastai: Interface to 'fastai'

#' @title Dicom class
#' @return None
#' @export
Dicom = function() {
  invisible(fastai2$medical$imaging$PILDicom)
}




#' @title get_dicom_files
#'
#' @description Get dicom files in `path` recursively, only in `folders`, if specified.
#'
#'
#' @param path path to files
#' @param recurse recursive or not
#' @param folders folder names
#' @return lsit of files
#'
#'
#' @examples
#'
#' \dontrun{
#'
#' items = get_dicom_files("siim_small/train/")
#'
#'
#' }
#'
#' @export
get_dicom_files <- function(path, recurse = TRUE, folders = NULL) {

  args = list(
    path = path,
    recurse = recurse,
    folders = folders
  )

  if(is.null(args$folders))
    args$folders <- NULL


  do.call(medical()$get_dicom_files, args)

}

#' @title Read dicom
#'
#' @description Open a `DICOM` file
#'
#'
#' @param fn file name
#' @param force logical, force
#' @return dicom object
#'
#'
#' @examples
#'
#' \dontrun{
#'
#' img = dcmread('hemorrhage.dcm')
#'
#'
#' }
#'
#' @export
dcmread <- function(fn, force = FALSE) {

  medical()$Path$dcmread(
    fn = fn,
    force = force
  )

}


#' @title Dicom
#' @description prints dicom file
#' @param x dicom file
#' @param ... additional parameters to pass
#' @return None
#' @export
print.pydicom.dataset.FileDataset <- function(x, ...) {
  cat(x$`__str__`())
}


#' @title Get image matrix
#'
#' @param img dicom file
#' @param type img transformation
#' @param scan apply uniform or gaussian blur effects
#' @param size size of image
#' @param convert to R matrix or keep tensor
#' @return tensor
#'
#'
#' @examples
#'
#' \dontrun{
#'
#' img = dcmread('hemorrhage.dcm')
#' img %>% get_dcm_matrix(type = 'raw')
#'
#' }
#'
#' @export
get_dcm_matrix <- function(img, type = 'raw', scan = '', size = 50, convert = TRUE) {

  obj = medical()$dicom_windows

  fun = function() {
    if(scan=='uniform_blur2d') {
      img = medical()$uniform_blur2d(
        img$scaled_px$windowed(res[[1]], res[[2]]), as.integer(size))
    } else if(scan=='gauss_blur2d') {
      img = medical()$gauss_blur2d(
        img$scaled_px$windowed(res[[1]], res[[2]]), as.integer(size))
    } else {
      img = img$scaled_px$windowed(res[[1]], res[[2]])
    }

    if(convert) {
      img$cpu()$numpy()
    } else {
      img
    }
  }

  if (type=='raw') {
    img$pixel_array
  } else if (type=='abdomen_soft') {
    res = obj$abdomen_soft
    fun()
  } else if (type=='brain') {
    res = obj$brain
    fun()
  } else if (type=='brain_bone') {
    res = obj$brain_bone
    fun()
  } else if (type=='brain_soft') {
    res = obj$brain_soft
    fun()
  } else if (type=='liver') {
    res = obj$liver
    fun()
  } else if (type=='lungs') {
    res = obj$lungs
    fun()
  } else if (type=='mediastinum') {
    res = obj$mediastinum
    fun()
  } else if (type=='spine_bone') {
    res = obj$spine_bone
    fun()
  } else if (type=='spine_soft') {
    res = obj$spine_soft
    fun()
  } else if (type=='stroke') {
    res = obj$stroke
    fun()
  } else if (type=='subdural') {
    res = obj$subdural
    fun()
  } else if (type=='normalized') {
    if(scan=='uniform_blur2d') {
      img = medical()$uniform_blur2d(
        img$scaled_px$hist_scaled(), as.integer(size))
    } else if(scan=='gauss_blur2d') {
      img = medical()$gauss_blur2d(
        img$scaled_px$hist_scaled(), as.integer(size))
    } else {
      img = img$hist_scaled()
    }

    if (convert) {
      img$cpu()$numpy()
    } else {
      img
    }
  }

}


#' @title Mask from blur
#' @param img image
#' @param window windowing effect
#' @param sigma sigma
#' @param thresh thresholf point
#' @param remove_max remove maximum or not
#'
#' @export
mask_from_blur <- function(img, window, sigma = 0.3, thresh = 0.05, remove_max = TRUE) {

  img$mask_from_blur(
    window = list(as.integer(window[[1]]),as.integer(window[[2]])),
    sigma = sigma,
    thresh = thresh,
    remove_max = remove_max
  )

}

#' @title Zoom
#' @param img image files
#' @param ratio ratio
#' @return image
#' @export
zoom <- function(img, ratio) {

  img$zoom(
    ratio = ratio
  )

}

#' @title Mask2bbox
#' @param mask mask
#' @param convert to R matrix
#' @return tensor
#' @export
mask2bbox <- function(mask, convert = TRUE) {

  res = medical()$mask2bbox(
    mask = mask
  )

  if(convert) {
    lo = res[[0]]$cpu()$numpy()
    hi = res[[1]]$cpu()$numpy()
    list(lo,hi)
  } else {
    res
  }

}

#' @title Abdomen soft
#' @return list
#' @export
win_abdoment_soft <- function() {
  medical()$dicom_windows$abdomen_soft
}

#' @title Brain
#' @return list
#' @export
win_brain <- function() {
  medical()$dicom_windows$brain
}

#' @title Brain bone
#' @return list
#' @export
win_brain_bone <- function() {
  medical()$dicom_windows$brain_bone
}

#' @title Brain soft
#' @return list
#' @export
win_brain_soft <- function() {
  medical()$dicom_windows$brain_soft
}

#' @title Liver
#' @return list
#' @export
win_liver <- function() {
  medical()$dicom_windows$liver
}

#' @title Lungs
#' @return list
#' @export
win_lungs <- function() {
  medical()$dicom_windows$lungs
}
#' @title Mediastinum
#' @return list
#' @export
win_mediastinum <- function() {
  medical()$dicom_windows$mediastinum
}

#' @title Spine bone
#' @return list
#' @export
win_spine_bone <- function() {
  medical()$dicom_windows$spine_bone
}
#' @title Spine soft
#' @return list
#' @export
win_spine_soft <- function() {
  medical()$dicom_windows$spine_soft
}

#' @title Stroke
#' @return list
#' @export
win_stroke <- function() {
  medical()$dicom_windows$stroke
}

#' @title Subdural
#' @return list
#' @export
win_subdural <- function() {
  medical()$dicom_windows$subdural
}