Radar as a means to study the biomechanics of gait: A validity study.

摘要

The purpose of this study was to use the latest technology in video imaging techniques to help develop and validate the necessary software and mathematical theory to identify human walking parameters (stride length (SL)) using radar. Two different instruments were used simultaneously to measure the gait of 20 apparently healthy college-aged male subjects: an experimental gait-measuring radar device (single-frequency, narrow-band (10.5 GHz), continuous-wave) and a Motion Analysis Corporation High Resolution Motion Capture real-time system with a six infrared video camera (Falcon; 60 Hz) array. The first 10 subjects were used to develop algorithms, based on amplitude measurements (decibels (dB)), necessary to predict radar data from video-based data. The forefoot region of the foot exhibited a dB value within a range-22.9201 to −26.3641 during the swing phase. Amplitude signal overlap of larger body parts over smaller body parts limited the identification of the forefoot within all gait phases but the swing phase. Microsoft Excel (Visual Basic) macro programs were formulated to refine the radar forefoot data into frequency versus time curves and subsequently calculate position data. Interpolation and smoothing of curves were performed using curve estimations from the SPSS statistical software program. The next 10 subjects were used to test the device. A dependent samples t-test indicated a significant difference (p = .000) between SL measurements from the radar and video-based devices. Radar SL values were statistically larger than those measured through the video-based device. Frequency curves from radar exhibited an inconsistent flat slope near 150–170 Hz, which created large curve estimations. Further analysis through video-based analysis indicated a distinct possibility of the contra-lateral hand creating a signal overlap of the forefoot. In conclusion, the radar device cannot be recommended as a valid means to measure gross walking parameters at this stage in development. Suggestions for future research include removing signals created by the hand near 150–170 Hz to improve frequency curve estimations of the forefoot and systematically identify other body parts to attain segment and center of mass information.