Motion Correction Workflow

Motion correction can only be applied to a dynamic input series. The aim is to correct for patient motion which is visible in the images and bring the anatomy into agreement across all the dynamic frames. The implementation uses the rigid matching approach, so it is only suitable when the motion doesn't result in deformation of the target tissue. Note that most appropriate way for PET and SPECT data is to correct motion during the image reconstruction, because otherwise the attenuation correction will not be fully accurate.

Please first load the dynamic images on the INPUT sub-page and make sure the Species setting is correct. Proceed with the motion correction button.

Image Inspection

On the REFERENCE sub-page inspect the motion in the data in order to see where the motion starts. The can be achieved by stepping through the frames using the slider

or by playing a cine across time

Reference Image for Motion Correction

There are various approaches for using rigid matching in the context of motion correction. One approach is to use REF Load for loading a suitable image to which the frames of the dynamic series are rigidly matched. Alternatively, a reference can be created out of the series itself in different ways with the button as illustrated below.

The choices work as follows:

Averaged	An average image is calculated from a range of frames and serves as the reference for the correction of the frames. The average is calculated from the sub-range defined by the upper selection range, which should only have negligible patient motion. Use the button for the actual reference creation.
Selected	The frame shown in the display will serve as the reference. Please note that if the pattern in the image changes significantly over time it will be difficult to motion correct successfully using a single frame. The use of markers is a way to potentially alleviate this problem. Use the button for the actual reference creation.
Previous	In this mode, motion correction matches each frame to its previous with the advantage that gradual pattern changes are less of a problem. On the other hand, successive matching errors might accumulate with this strategy. The final transformation per frame is obtained by combining the transformation matrices of all preceding frames. In this way multiple interpolations in the final image reslicing are avoided.
Following	This is the same principle as the Previous mode, but the method works from the last frame in the selection through the earliest one.

Correction Range

Optionally, configure a sub-range, wherein motion correction will be performed.

A reason to exclude a range of frames may be the lack of signal in the initial frames, and/or frames with a short acquisition duration during which patient motion is less likely. Excluded frames will be copied to the corrected series without changes, and the correction matrix of these frames will contain zero for all rotations/translations.

Matching Parameters

Motion correction uses the rigid matching technology and has the same two parameter presets and (default). The red bar above the buttons indicates which preset is active. As described for rigid matching, the parameters can be tailored if needed.

Starting the Motion Correction

Please use the Motion button to start the process, and inspect the results which are shown on the MATCHED sub-page. The resulting transformation is a sequence of rigid transformations as illustrated below.