The challenges of using A-mode ultrasound for kinematic analysis

Abstract

Amplitude-mode (A-mode) ultrasound (US) was proposed as non-radiant and non-invasive alternative technique for measuring bone kinematics and gait analysis. The system tracks the bone pose in 3D using optical markers and a motion tracking system, while A-mode US transducers measure the distance to the bony surface. However, detecting the actual bone depth using A-mode US is challenging, and the accuracy of the depth estimations remains unclear. Therefore, this study aimed to assess how accurate the A-mode US transducers can measure skin-to-bone distance. Multiple preliminary experiments were conducted and the outcomes were described in the appendices of this report. The main study included one right lower leg of a human cadaver, supplemented with three US holders and a total of nine US transducers. US estimated depths were calculated using the peak time in the A-mode signal and the speed of sound in soft tissues, and validated towards the ground truth depths from simultaneously conducted Computed Tomography (CT) scans at Ultra High Resolution (UHR). Two consecutive CT scans resulted in 18 CT ground truth depths, which were derived from the segmented 3D models of the tibia and the US transducers. Four measurements were excluded, and initially, for 3 of the 14 remaining outcomes (21%), the selected peak in the US signal was incorrect. After selecting the correct peaks, the US measurement overestimated the CT ground truth depth in 12/14 cases, and underestimated in the other two cases. Using the standard speed of sound in soft tissues (1540 m/s), the mean and standard deviation of the absolute difference between US estimated depth and CT ground truth depth was 0.9192 ± 0.4850 mm, whereas the best fitting speed of sound for the current setup (1424 m/s, according to one of the preliminary studies) decreased the mean absolute error to 0.6293 ± 0.6872 mm. In conclusion, if the correct amplitude peak in the US signal can be identified, the A-mode US transducers are capable of measuring skin-to-bone distance in a human cadaver leg with mean absolute errors below 1 mm.