Overview

MitkMatchImage is a lightweight, command-line image registration tool built using the MITK (Medical Imaging Interaction Toolkit) and MatchPoint libraries. This tool enables users to register a moving image to a target image using a specified registration algorithm and supports advanced configurations like masking and parameter tuning. It is often used together with the application MitkMapImage which allows applying a registration to an image and map it accordingly into the target space.

Features

Register images using a plugin-based algorithm.
Apply optional image masks for targeted registration.
Accepts custom algorithm parameters via JSON input.
Supports multi-label image segmentation.
Outputs registration results as a file.

Usage

MitkMatchImage -m <movingImage> -t <targetImage> -a <algorithmDLL> -o <outputFile> [optional parameters]

Required Arguments

Argument	Short	Description
`--moving`	`-m`	Path to the moving image file.
`--target`	`-t`	Path to the target image file.
`--algorithm`	`-a`	Path to the deployed registration algorithm DLL.
`--output`	`-o`	Output file for the computed registration result.

Optional Arguments

Argument	Short	Description
`--moving_mask`	`-mm`	Mask for the moving image.
`--moving_mask_label`	`-mml`	Label name to extract from the moving mask if it is a MITK Segmentation with multiple labels.
`--target_mask`	`-tm`	Mask for the target image.
`--target_mask_label`	`-tml`	Label name to extract from the target mask if it is a MITK Segmentation with multiple labels.
`--parameters`	`-p`	JSON string of algorithm parameters.
`--help`	`-h`	Show help text.

Example

MitkMatchImage -m lungCT_moving.nii.gz -t lungCT_target.nii.gz -a rigidReg.dll -o outputReg.mapr -mm movingSeg.nrrd -mml "lung" -tm targetSeg.nrrd -tml "lung" -p '{"maxIterations": 200, "tolerance": 1e-6}'

What happens:
Registred the moving image lungCT_moving.nii.gz onto the target image lungCT_target.nii.gz by using the registration algorithm provided by rigidReg.dll. The algorithm will use a moving segmentation movingSeg.nrrd and a target segmentation targetSeg.nrrd to guide the registration. Both segmentations are supposed to be multi-label, so the label lung is selected for both. The resulting registration is stored in outputReg.mapr. The registration algorithm is also configured by passing some explicit parameter settings.

Notes

DLL Compatibility: Ensure the algorithm DLL is compiled with MatchPoint deployment support.
Masks: When using multi-label segmentations, specify a label name to extract the right region.
Time Series: Only the first time step of a 4D image is considered.