gms | German Medical Science

In infancy mental, behavioral and motor dysfunctions result in abnormal movement patterns [Ref. 1]. Therefore movement analysis is the hallmark of neurological evaluation at that age group. In addition to the subjective analysis of video recordings quantified movement analysis by the means of 3D motion capturing can give objective decision support for physicians [Ref. 2]. Three-dimensional visualization is crucial for the analysis of movement data. This article presents the application of 3D motion capturing and visualization to infants and shows how 3D visualization can give new insight to distinguish between normal and abnormal movement patterns.

Infant motions are recorded with an electromagnetic tracking system (3D GuidanceTM, Ascension Technology Inc.) with a sample frequency of 50 Hz. Four sensors are attached with eudermic patches to the arm and the leg respectively. The infant is lying on its back on a mattress and can move arbitrarily for about 10 minutes while the movements are recorded. Each sensor delivers 6 degrees of freedom, i.e. position and orientation data.

Since these spatial data are relative to a global coordinate system (GCS), original sensor data cannot describe the body movements neither in a repeatable (if the infant is moved in the GCS) nor in a meaningful fashion. Therefore a biomechanical model of the infant anatomy [Ref. 3] is used to calculate the relative movements of the body segments (see Figure 1 [Fig. 1]). The relative rotations between them are meaningful (e.g. they describe the motion of the hand relative to the forearm) and repeatable since they are not dependent of the infant’s pose in the GCS. Figure 1 [Fig. 1] shows recorded poses of the arm sensors and the reconstructed arm model.

A 3D video player based on the visualization framework VTK was developed to give new insights to the quality of infant movements. Additionally to common features of 3D video players like zoom and change of camera perspective the application offers the following features:

• Movement trajectories (e.g. of the hand) can show the spatial distribution of movements (see Figure 2 [Fig. 2]).
• A movement trail (see Figure 1 [Fig. 1]) can give a better impression of fine movements.

These features can reveal movement characteristics that are not visible to the naked eye.

The movements of more than 30 infants were recorded. Physicians assessed the movements on the basis of video recordings according to the so called General Movements Analysis [Ref. 1] which allows the prediction of motor dysfunction. Normal movements show high complexity and variety, whereas abnormal movements are less complex. Figure 2 [Fig. 2] shows trajectories of arm movements diagnosed normal and arm movements diagnosed abnormal. The abnormal movements clearly show a more stereotype quality whereas the normal movements have a higher spatial variety. This visual impression indicates that it might be possible to use the quantitative recordings to distinguish between the two groups.

Data visualization is an important first step to reveal underlying characteristics. The presented three-dimensional visualization of spontaneous infant movements gives insight into their spatial-temporal structure. We plan to incorporate further information like velocity, acceleration or the spatial vicinity of movements. Based on these visualizations we might be able to generate new hypotheses about how to quantify differences between normal and abnormal movements.

This project has been funded by the Dietmar-Hopp-Stiftung, Walldorf, Germany.