Benchmarking Foot Trajectory Estimation Methods for Mobile Gait Analysis.

Area of Science:

• Biomechanics
• Wearable Technology
• Medical Engineering

Background:

• Mobile gait analysis systems using inertial sensors on shoes offer insights into motor impairments, aiding patient assessment in neurodegenerative diseases.
• Accurate reconstruction of foot trajectories from inertial data is crucial for these systems.
• A lack of standardized benchmark datasets hinders fair comparison of existing trajectory estimation methods.

Purpose of the Study:

• To implement and evaluate multiple orientation estimation and double integration schemes for foot trajectory reconstruction.
• To provide a fair and direct comparison of these methods using a consistent dataset.
• To identify the optimal processing pipeline for foot trajectory estimation in mobile gait analysis.

Main Methods:

• Three distinct orientation estimation algorithms were implemented.
• Three different double integration schemes were applied for trajectory reconstruction.
• All implemented methods were evaluated against a gold-standard marker-based motion capture system.
• A dataset comprising 735 strides from 16 healthy subjects was utilized for evaluation.

Main Results:

• The performance of the implemented methods was quantitatively ranked.
• Specific combinations of orientation estimation and double integration schemes demonstrated superior accuracy.
• The study identified a processing pipeline that is most suitable for mobile gait analysis.

Conclusions:

• The findings provide a clear ranking of different foot trajectory estimation methods.
• The identified optimal pipeline can enhance the accuracy and reliability of shoe-based mobile gait analysis.
• This work contributes to standardizing the evaluation of gait analysis techniques for medical applications.