Skip to contents

Tissue Classification via LIBS

Source: R/tissue.R
ls_tissue_classify.Rd

Classifies tissue types based on LIBS spectral signatures. Implements three methods:

  • "ratio" — rule-based using canonical elemental ratios (Ca/Na, Fe/Na, Zn/Na, K/Na, C/Ca) derived from tissue biochemistry. No training data required.

  • "plsda" — fits a PLS-DA model on the supplied labelled reference dataset and predicts class for x.

  • "svm" — fits an SVM on reference and predicts class.

Usage

ls_tissue_classify(
  x,
  method = "ratio",
  reference = NULL,
  group_col = "tissue",
  verbose = TRUE
)

Arguments

x

A ls_spectrum() or ls_dataset() object (unknowns).

method

Character. "ratio", "plsda", or "svm". Default "ratio".

reference

A ls_dataset() with labelled tissue reference spectra (needed for "plsda" and "svm").

group_col

Character. Column in reference sample_info with tissue labels. Default "tissue".

verbose

Logical. Default TRUE.

Value

A tibble::tibble() with columns sample_id, predicted_tissue, confidence.

Examples

ds <- ls_example_data("tissue")
ls_tissue_classify(ds[1:3])
#> # A tibble: 3 × 3
#>   sample_id predicted_tissue confidence
#>   <chr>     <chr>                 <dbl>
#> 1 bone_01   bone                  0.999
#> 2 bone_02   bone                  1.000
#> 3 bone_03   bone                  1.000