Calibration Parameters

Calibration process is described by

• Calibration quality stipulating approximation method to correlate data,
• Lateral discretization to account for scanner sensor to sensor variations,
• Dose prefiltering to mitigate noise and noise related artifacts.

All calibration parameters can be accessed through the context menu via the 'Libra' button at bottom center

Calibration quality stipulates the approximation method of the correlation and is described by

• Approximation order
  Polynomial order used to approximate calibration data.
  
  • Low order
    highly stable (always produces result), but more modest accuracy.
  • High order
    high accuracy for smooth data, but may be instable when data distorted (may not deliver a result).
• Dose sections
  Total dose range is splitted into sections (equi-sized subintervals) and each section is calibrated separately.
  The calibrations for the dose section are smoothly attached to each other to have a smooth approximation over the entire dose range avoiding any ‘jump’ in dose values.
  
  • Smaller dose range means usually better approximation, i.e better accuracy is achieved.
    This approach works well when calibration data are sufficient smooth.
  • Low order for each section makes resulting method more stable and small enough dose sections (i.e. if number of sections high enough) produce better calibration accuracy.
    This approach works well when calibration data are sufficiently equi-distributed (e.g. calibration step wedge many dose levels).

Calibration quality context menu

Context menu shown below has 3 selection groups:

• Single dose section approximation,
• Multiple dose section.
• Legacy

	Minimum requirments	Linear data approximation over entire dose calibration range. Most stable approximation method Minimum required data: Single strip exposure and unexposed 0 Gy strip.
	Low approximation order	Quadratic polynomial approximation over full calibration range. Data requirements: Wedge with 3 or more dose levels.
	Medium order	Cubic polynomial approximation over full calibration range. Data requirements: Wedge with 4 or more dose levels.
	High order	Quartic polynomial approximation over full calibration range. Data requirements: Wedge with 5 or more dose levels.
	Piecewise min requirments	Linear polynomial approximation over multiple calibration dose sections. Very stable approximation with high accuracy at higher number of dose sections. Data requirements: Wedge with 3 or more dose levels.
	Piecewise low order	Linear polynomial approximation over multiple calibration dose sections. Highly accurate method, very stable approximation with lower number of dose sections. Default method for calibration quality. Data requirements: Wedge with 3 or more dose levels.
	Balanced composite	Legacy calibration protocol (balanced composite) using secondary scan with calibration data.

Lateral discretization stipulates the calibration resolution,
i.e. how many horizontal image pixels (scanner sensors) are used per calibration.

Lateral Discretization context menu

Each item in the context menu includes the number of image pixels per calibration that will be selected.

	Without lateral discretization	Single calibration correlation is used for all scanner sensors (all lateral locations). Highest number of image pixels per calibration. No compensation of lateral scanner variations, maximum averaging of film non-uniformity variations.
	Low calibration resolution	Recommended low number of calibrations. High number of image pixels per calibration. Adaptive to used calibration quality and width of the scanned image.
	Medium calibration resolution	Recommended medium number of calibrations. Medium number of image pixels per calibration. Adaptive to used calibration quality and wdith of the scanned image.
	Highest calibration resolution	Recommended highest number of calibrations. High number of image pixels per calibration. Adaptive to used calibration quality and width of the scanned image.
	Bitmap pixel per calibration	Select the number of pixels per calibration. 1 pixel per calibration - each sensor is calibrated separately, best compensation of lateral scanner variation, Note:The less pixels per calibration are used
	Suppress without fields	If selected and a calibration field is not available the lateral discretization is suppressed.
	Suppress with strips	If selected, the lateral discretization is suppress when calibration strip is present.

Dose prefiltering context menu

Filter selected in the context menu shown below is applied directly to the film scan before dose conversion.

	None	No filtering is applied.
	Arithmetic mean	Arithmetic average across the sampled points is applied.
	Geometric mean	Geometric average across the sampled points is applied.
	Median	Median value across the sampled points is applied.
	Least variation mean	Non-linear Kuwahara filter is applied to smooth the data.
	Least squares plane	Least squares plane is fitted to the sample data and plane value at filter position is used.
	Filter dimension	Filter dimension is the diameter of the circular region centered at filter location.