Shoulder Glenohumeral Elevation Estimation based on Upper Arm Orientation.

Area of Science:

• Biomechanics
• Medical Imaging
• Robotics

Background:

• Accurate modeling of shoulder glenohumeral displacement is crucial for understanding upper arm movement and designing prosthetics.
• Existing methods for estimating humeral head elevation and translation have limitations in capturing complex joint kinematics.

Purpose of the Study:

• To compare four distinct modeling approaches for quantifying shoulder glenohumeral displacement.
• To evaluate the accuracy of biomechanics-inspired and Gaussian process regression models in estimating humeral head vertical and horizontal movements.

Main Methods:

• A biomechanics-inspired model with least squares parameter identification was developed.
• Three Gaussian process regression models were implemented using different input variable sets: joint angles and upper arm orientation (quaternions).
• Motion capture data was collected to quantitatively assess and compare the predictive performance of each model.

Main Results:

• Gaussian process regression utilizing overall upper arm orientation (quaternions) demonstrated superior accuracy in modeling glenohumeral displacement.
• Other tested models, including biomechanics-inspired and simpler Gaussian process approaches, also achieved acceptable accuracy levels for specific applications.
• The study highlights the effectiveness of data-driven approaches like Gaussian process regression for complex biomechanical modeling.

Conclusions:

• Upper arm orientation, represented by quaternions, is a key determinant for accurate glenohumeral displacement modeling.
• Gaussian process regression offers a robust and adaptable framework for estimating shoulder joint kinematics.
• Model selection should consider the trade-off between accuracy requirements and computational complexity for specific clinical or engineering tasks.