Machine learning algorithms for predicting scapular kinematics

Kristen F Nicholson; R Tyler Richardson; Elizabeth A Rapp van Roden; R Garry Quinton; Kert F Anzilotti; James G Richards

doi:10.1016/j.medengphy.2019.01.005

Machine learning algorithms for predicting scapular kinematics

Med Eng Phys. 2019 Mar:65:39-45. doi: 10.1016/j.medengphy.2019.01.005. Epub 2019 Feb 4.

Authors

Kristen F Nicholson¹, R Tyler Richardson², Elizabeth A Rapp van Roden³, R Garry Quinton⁴, Kert F Anzilotti⁵, James G Richards⁶

Affiliations

¹ Biomechanics and Movement Science Program, University of Delaware, Newark, DE, USA. Electronic address: kristent@udel.edu.
² Kinesiology Program, School of Behavioral Sciences and Education, Penn State Harrisburg, Middletown, PA, USA. Electronic address: tr12@psu.edu.
³ Biomechanics and Movement Science Program, University of Delaware, Newark, DE, USA. Electronic address: lizrapp@udel.edu.
⁴ Biomechanics and Movement Science Program, University of Delaware, Newark, DE, USA. Electronic address: drrgquinton@gmail.com.
⁵ Department of Radiology, Christiana Health Care System, Newark, DE, USA. Electronic address: KAnzilotti@christianacare.org.
⁶ Biomechanics and Movement Science Program, University of Delaware, Newark, DE, USA. Electronic address: jimr@udel.edu.

PMID: 30733173
DOI: 10.1016/j.medengphy.2019.01.005

Abstract

The goal of this study was to develop and validate a non-invasive approach to estimate scapular kinematics in individual patients. We hypothesized that machine learning algorithms could be developed using motion capture data to accurately estimate dynamic scapula orientation based on measured humeral orientations and acromion process positions. The accuracy of the algorithms was evaluated against a gold standard of biplane fluoroscopy using a 2D to 3D fluoroscopy/model matching process. Individualized neural networks were developed for nine healthy adult shoulders. These models were used to predict scapulothoracic kinematics, and the predicted kinematics were compared to kinematics obtained using biplane fluoroscopy to determine the accuracy of the machine learning algorithms. Results showed correlations between predicted kinematics and validation kinematics. Estimated kinematics were within 10 of validation kinematics. We concluded that individualized machine learning algorithms show promise for providing accurate, non-invasive measurements of scapulothoracic kinematics.

Keywords: Biomechanics; Machine learning; Neural networks; Shoulder mechanics.

MeSH terms

Adult
Biomechanical Phenomena
Fluoroscopy
Healthy Volunteers
Humans
Imaging, Three-Dimensional
Machine Learning*
Mechanical Phenomena*
Movement
Scapula / diagnostic imaging
Scapula / physiology*