(K-407) Comparison of On-Body vs. Around-Body Systems for Assessment of Movement Abnormalities in Health and Disease

Introduction:: Chronic disease is on the rise. Beyond direct organ involvement in chronic diseases e.g. heart failure and renal failure, generalized motion abnormalities have been described as well. Presently, motion disorders are largely characterized qualitatively. Our goal has been to develop methods to quantitate clinical motion abnormalities, thereby providing enhanced diagnostic information in the form of “digital biomarkers’ of disease. Our group developed quantitative measurement devices: 1. A stretchable electronic wearable patch (BioStamp, MC10) – presently FDA approved - measuring motion with 6 degrees of freedom – as an “on body” (contact) method, and 2. A video telemedicine system using external markers (MOCA 2.0) – an “around-body” (non-contact) method, measuring body element movement, e.g. arms, fingers using brightly colored markers. Here we compared the relative performance of each system assessing a series of defined movements.

Materials and Methods::

    Six subjects (3:3 male:female) had both wearable digital sensor patches (on-body method) and small (< =10mm2) color paper markers (around-body method) applied. Subjects performed 6 defined movements, categorized as either small (< 1 sq ft), medium (1-5 sq ft), or large ( >5 sq ft) (Figure 1). Each movement was performed for 30 seconds at both slow and fast pace, (n=3), for a total of 36 movement sessions, totaling to 216 motion trials across all subjects. Digital signals for the on-body method were streamed and recorded contemporaneous with around-body video recording of subject movement.
    The on-body system reports data in angular velocity and linear acceleration in three axes, which is converted into angular displacement, linear displacement, and linear velocity for comparison. The around-body system uses video recordings of the trial. Through color tracking and visual analysis, the XY-coordinates of each marker is determined within the frame in pixels. A known distance in the frame (typically the distance between two markers) is used to convert pixels into meters, giving linear displacement. From there, linear velocity, angular displacement, and angular velocity can be determined and used as direct comparators to the on-body system.

Results, Conclusions, and Discussions::

    Of the 216 original trials, 19 of the file pairs could not be correlated between BioStamp and MOCA 2.0. On average, this was caused by the around-body system being unable to track the motion due to lighting or color tracking limitations. Of the remaining 197 files, the average angular displacement correlation across all files was 82.75%, with a high of 95.66% correlation on the bicep curl, and a low of 60.04% for shoulder abduction/adduction. The average angular velocity correlation was 62.46%, with a high of 72.65% on the bicep curl, and a low of 45.57% on the shoulder abduction/adduction (Figure 2).
    Both on-body and around body systems offer readily deployable means for capturing nuances of human motion useful for movement disorder diagnosis and monitoring. The BioStamp can produce more accurate and detailed information about each movement than currently trackable by cameras. This becomes evident in fast, small movements such as the finger pinch, which is tracked at 60 reps/second. In addition, the BioStamp tracks motion in 3 dimensions, whereas the around-body system is currently limited to 2-dimensional data. The data collected by BioStamps relies on the force of gravity to orient the axes of movement, which can distort the data during stamp rotation.
    In contrast, the around-body video capture system utilizes video processing to directly yield positional data and therefore is free of introduced sensor error. However, MOCA 2.0 has the limitations of inefficacy with movements too small or fast for camera capture, or when lighting or interfering colors cause an inability to accurately track the markers. Whenever a marker can not be tracked in the frame, the position of the marker does not get updated from the previously known location. This causes sharp fluctuations in velocity that become apparent in the velocity correlation metrics as described (Figure 2B). The correlation between the two systems becomes much more apparent when looking at displacement data. Understanding system constraints will allow future patient studies to advance with greater accuracy and convenience based on system choice.

Acknowledgements (Optional): : This project was funded and supported by NIH R25 DK128859.

References (Optional): :