Heel-Strike and Toe-Off Detection Algorithm Based on Deep Neural Networks Using Shank-Worn Inertial Sensors for Clinical Purpose

Inertial sensors are often used to detect heel-strike (HS) and toe-off (TO) events. However, in clinical practice, this placement can be challenging or even impossible due to foot deformities. This paper introduces a new algorithm for HS and TO event detection when sensors are placed on the lateral malleolus. This approach allows gait analysis in patients with foot deformities, as the sensor's secure fixation on the wide bone surface ensures stability during walking.

The algorithm is based on deep neural networks, making it adaptable for various pathological gait patterns through retraining. This is crucial in clinical practice, where numerous pathological gait patterns exist. The algorithm was integrated into a new wearable system for clinical gait analysis.

The paper also validates this wearable system. The proposed algorithm and system's performance were compared to a reference treadmill system using a capacitance-based pressure platform. A total of 117 healthy volunteers (62 males and 55 females, aged 24–55 years, 162–183 cm) participated in a 2-minute walking trial at different speeds.

The results showed mean accuracy ± precision of -0.021 ± 0.091 s for the gait cycle, 0.589 ± 1.144 steps/min for cadence, -0.051 ± 0.544% for stance phase, -0.37 ± 0.649% for single support, 0.296 ± 0.711% for double support, 0.132 ± 0.561% for load response, and 0.106 ± 0.661% for preswing. The limitations of the proposed algorithm and its comparison with state-of-the-art methods were also discussed.