Overview

This tool maps a medical image or segmentation into a target geometry using an optional registration file and image interpolation method. It supports optional output geometry templates, padding, and super-sampling.

Usage

MitkMapImage -i <inputImage> -o <outputImage> [options]

Required Arguments

Argument	Short	Type	Description
`--input`	`-i`	File	Input image file to be mapped
`--output`	`-o`	File	Output path for the mapped image

Optional Arguments

Argument	Short	Type	Description
`--registration`	`-r`	File	Path to registration file (default: identity)
`--template`	`-t`	File	Reference image that defines output geometry (default: input geometry)
`--interpolator`	`-n`	Integer	Interpolation type (default: 2)
			1: Nearest Neighbour
			2: Linear
			3: BSpline3
			4: WSinc Hamming
			5: WSinc Welch
`--padding`	`-p`	Float	Padding value for unmapped voxels (default: 0.0)
`--super-sampling`	`-s`	List	Super-sampling factor(s): 1 or 3 values allowed
`--help`	`-h`	Flag	Print help information and exit

Examples

Map an image using a registration and reference image

MitkMapImage -i fixedImage.nii -r deformableReg.mapr -t refImage.nii -o outputImage.nii

Map using identity (no registration) and linear interpolation (default)

MitkMapImage -i input.nii -o output.nii

Use BSpline interpolation and pad unmapped voxels with -1000

MitkMapImage -i input.nii -o output.nii -r reg.mapr -n 3 -p -1000

Apply super-sampling (doubles resolution isotropically)

MitkMapImage -i input.nii -o output.nii -s 2

Anisotropic super-sampling

MitkMapImage -i input.nii -o output.nii -s 2 2 1

🔍 Notes

Tool does also support mapping of dynamic data (e.g. 3D+t images).
Supports both standard images and multi-label segmentations.
If the registration file is missing, an identity transform is used.
Super-sampling must either provide one value (isotropic) or three (for x, y, z).
If mapping a segmentation, the interpolaiton option will be ignored. Nearest neighbour will always be used.

Table of Contents