Mean Normalization for Matrix Data

Normalizes each row of a matrix by dividing its elements by the row mean, ignoring NA values. This step is commonly used to adjust for global intensity differences across subjects before applying statistical comparisons or functional data analysis.

Usage

meanNormalization(
  matrixData,
  handleInvalidRows = c("warn", "error", "omit"),
  returnDetails = FALSE,
  quiet = FALSE
)

Arguments

matrixData

A matrix where each row represents one subject’s PET data, typically generated by matrixCreator.

handleInvalidRows

character. Specifies how to handle rows with invalid means (either zero or NA). Options include "warn" (default), "error", or "omit".

returnDetails

logical. If TRUE, returns a list with the normalized matrix and additional diagnostics. If FALSE (default), returns only the normalized matrix.

quiet

logical. If TRUE, suppresses console messages. Default is FALSE.

Value

A normalized matrix, or a list if returnDetails = TRUE.

• normalizedMatrix – The normalized matrix.

• problemRows – Indices of rows that had zero or NA means.

Details

The function performs the following steps

1. Computes the row means of the input matrix, ignoring NAs.

2. Divides each row by its corresponding mean.

3. Replaces NaN values (from division by 0) with 0 if applicable.

4. Handles problematic rows according to the selected handleInvalidRows option: "warn" (default) issues a warning, "error" stops execution, and "omit" removes the affected rows from the result.

This step is often used prior to applying SCC methods to ensure comparability across subjects.

See also

matrixCreator for building the matrix input to normalize.

Examples

# Generate a minimal database and create a matrix (1 control subject)
dataDir <- system.file("extdata", package = "neuroSCC")
controlPattern <- "^syntheticControl1\\.nii\\.gz$"
databaseControls <- databaseCreator(pattern = controlPattern,
                                    control = TRUE,
                                    quiet = TRUE)
matrixControls <- matrixCreator(databaseControls, paramZ = 35, quiet = TRUE)

# Normalize the matrix (with diagnostics)
normalizationResult <- meanNormalization(matrixControls,
                                         returnDetails = TRUE,
                                         quiet = FALSE)
#> 
#>  Mean before normalization: 4.349632
#> 
#>  Normalization completed.