The reproducibility of signals from skin-mounted accelerometers following removal and replacement

Signals from skin-mounted accelerometers may contain measurement error when compared to those obtained from bone-mounted sensors. While this error may be minimized through various techniques, additional error may arise as a result of accelerometer removal from the skin and subsequent replacement. The purpose of this experiment was to determine if skin-mounted accelerometer signals remain similar before and after sensor replacement when sensors are stimulated within a consistent vibration environment. The spines of five porcine and five human cadavers were vibrated non-invasively and the resulting response measured using accelerometers glued to the skin overlying the vertebrae of interest (T-9 in porcine cadavers, L2-L4 in human cadavers). Accelerometers were removed, then replaced, to their perceived original positions. Accelerometer signals showed high repeatability within an original placement, however, once replaced, pre- and post-replacement signals were statistically dissimilar in all cadavers tested. Specifically, the similarity of pre- and post-replacement signals was poor for different skin types (porcine and human) and did not improve with accelerometer proximity to the vibration source. From these data, we conclude that accelerometers adhered to the skin overtop spinous processes are able to obtain highly repeatable signals for a given placement, however, signals change significantly after sensor replacement. When skin-mounted accelerometers are used in gait and other studies, investigators should consider performing test/re-test analyses to ensure that sensor data are not affected if sensor replacement is required by design or by accident.