Estimating foot strike in bare-breasted running from the position of the sternum

Background: When recording breast biomechanics with and without a bra, electromagnetic sensors can be attached directly to the skin of the breast to record breast motion, without the issue of marker occlusion in optical systems. However, as the sensors are wired and have a limited range, it is not practical to attach a sensor to the lower limb to establish gait events. Previous work e.g. [1] has used every other minimum position of the sternum (MINSTN) to divide running into gait cycles. With this latter method it is not possible to distinguish left and right foot strike. Secondly, due to the support moment [2], MINSTN is
likely to occur after foot strike. AIM: To determine if foot strike during bare-breasted running on a treadmill determined from a marker on the foot can be estimated from the position of the sternum. 
Methods: Twelve healthy female volunteers with a breast size of UK 34E (age=25±5 years, height=1.65±0.04 m, mass=63.8±5 kg, mean±SD) ran on a treadmill at 10 km/hr. Reflective markers were placed on the suprasternal notch (sternum), left nipple and left and right heel. Following a warm-up, data was recorded using a 15-camera system (Qualisys, Oqus, Sweden) for 10 seconds at 200 Hz. Data were processed in MATLAB (Mathworks Inc.) and Microsoft Excel. Marker trajectories were filtered using a Lowpass Butterworth Filter at 13 Hz. Foot strike was determined from the minimum vertical position of the heel marker (MINHEEL) [3]. Maximum and minimum medio-lateral (ML) position of the sternum were identified as left and right foot strike respectively. The MINSTN was calculated for each foot strike. A Bland-Altman analysis [4] was conducted using each step for the sternum ML position - MINHEEL and MINSTN- MINHEEL. RESULTS: Figure 1. shows on average a) sternum ML position was 20 ms (±60 ms) later than MINHEEL and b) MINSTN was 98 ms (±15 ms) later than MINHEEL. 
Conclusions: Although foot strike determined from sternum ML was closer to MINHEEL than MINSTN, it was more variable, occurring both before and after MINHEEL. Therefore, sternum ML can be used to distinguish left from right foot strike and MINSTN with an offset of ~100 ms could be used to estimate foot strike, when it is not possible to detect foot strike directly. Future work could refine this method using the gold-standard means of establishing foot strike with vertical ground reaction force. Discriminating between left and right foot strike can allow us to study potential differences in breast motion at foot strike between the ipsilateral and contralateral side.